UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Two- and three-dimensional velocity structure of the southwestern Canadian Cordillera from seismic refraction… Zelt, Barry Curtis 1994

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

Item Metadata

Download

Media
831-ubc_1994-954175.pdf [ 14.04MB ]
Metadata
JSON: 831-1.0052952.json
JSON-LD: 831-1.0052952-ld.json
RDF/XML (Pretty): 831-1.0052952-rdf.xml
RDF/JSON: 831-1.0052952-rdf.json
Turtle: 831-1.0052952-turtle.txt
N-Triples: 831-1.0052952-rdf-ntriples.txt
Original Record: 831-1.0052952-source.json
Full Text
831-1.0052952-fulltext.txt
Citation
831-1.0052952.ris

Full Text

TWO- AND THREE-DIMENSIONAL VELOCITYSTRUCTURE OF THE SOUTHWESTERN CANADIANCORDILLERA FROM SEISMIC REFRACTION DATAbyBARRY CURTIS ZELTB.Sc. Hons. (Physics and Applied Mathematics), University of Victoria, 1984M.Sc. (Geophysics), University of British Columbia, 1987A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIESDEPARTMENT OF GEOPHYSICS AND ASTRONOMYWe accept this thesis as conformingto the required standardTHE UNIVERSITY OF B COLUMBIASeptember 1994© Barry Curtis Zelt, 1994In presenting this thesis in partial fulfillment of therequirements for an advanced degree at the University of BritishColumbia, I agree that the Library shall make it freely availablefor reference and study. I further agree that permission forextensive copying of this thesis for scholarly purposes may begranted by the head of my department or by his or herrepresentatives. It is understood that copying or publication ofthis thesis for financial gain shall not be allowed without mywritten permission.(Signature)Department of Geophysics & AstronomyThe University of British ColumbiaVancouver, CanadaDate 26 September, 1994ABSTRACTSeismic refraction/wide-angle reflection data recorded on a triangular array in thesouthwestern Canadian Cordillera in 1989 as part of the Lithoprobe Southern Cordilleratransect are analyzed to determine the two- and three-dimensional velocity structure of thecrust and upper mantle. In-line data recorded along two sides of the triangle are interpreted for2—D structure using an iterative combination of traveltime inversion and forward modellingof amplitudes. An algorithm for the inversion of wide-angle seismic data to determine 3—Dvelocity structure and depth to reflecting interfaces is developed. The algorithm is based onan existing procedure for the inversion of first arrival traveltimes which includes: (i) forwardmodelling of traveltimes using a 3—D finite-difference algorithm; and (ii) a simple velocitymodel parameterization for the inversion which eliminates the need to solve a large systemof equations. The existing procedure is extended to allow: (i) fast and accurate forwardmodelling of reflection times; (ii) the inversion of reflection times to solve for depth to areflecting interface and/or velocity structure; (iii) the inversion of first arrival traveltimesto solve for depth to a refracting interface; and (iv) layer stripping. Application of thealgorithm for southern Cordillera data uses Pg to constrain upper crustal velocity structure,PmP to constrain lower crustal velocity structure and depth to Moho, and P to constrainupper mantle velocities and depth to Moho.Results for a line running along-strike in the southern Intermontane Belt (Quesnelliaterrane) reveal low average vertical velocity gradients, average depth to Moho of 32 km,and an upper mantle reflector —16 km below the Moho that may represent the base of thelithosphere. The upper and middle crust of the refraction model comprise the upper part ofthe Quesnellia terrane; the lower crust probably comprises parautochthonous and cratonicNorth America, but does not show the division into two components that is inferred fromreflection data, indicating that their physical properties are not significantly different withinthe resolution of the refraction data. The lower lithosphere of Quesnellia is absent and11presumably was recycled in the mantle.Results for a line running across the strike of the southernmost Coast Belt and easternInsular Belt reveal large lateral variations in velocity. The most significant of these variationsis a decrease in upper and middle crustal velocities to the east of the surface trace ofthe Harrison Fault, which likely represents the transition from Insular to Intermontanesuperterrane crust. Depth to Moho is 34—37 km beneath most of the Coast Belt, indicating asmall crustal root associated with the Cascade magmatic arc. The Moho decreases to 30 kmbeneath the eastern Insular Belt, which is much less than previous estimates. The inferredcrustal velocity structure beneath the Western Coast Belt (WCB) is consistent with the threelayer conductivity structure for this area, and suggests that the upper 8—12 km representsthe massive cover of plutonic rocks which characterizes the WCB. The middle and lowercrust beneath the WCB is interpreted as Wrangellia, which may extend at depth eastwardsas far as the Harrison Fault.The 3—D velocity model for the southwestern Canadian Cordillera is characterized by (i)significant lateral velocity variations at all depths that do not, in general, strongly correlatewith surface geological features or gravity data; (ii) higher average crustal velocities in theCoast Belt in comparison with the Intermontane Belt; (iii) relatively high velocity middleand lower crust in the southwest which correlates with a strong relative gravity high and mayoutline the eastern extent of lower Wrangellia; (iv) average upper mantle velocity of 7.85kmls; (v) depth to Moho of 33—36 km in the Intermontane Belt and 36—38 km throughoutmost of the Coast Belt, shallowing in the west to 33 km near the Insular-Coast contact.A correlation between thick crust and low heat flow suggests that a significant portion oftotal heat flow is sub-crustal in origin, possibly associated with (a recent) upflow of mantleconvection.111CONTENTSABSTRACT.LIST OF TABLESLIST OF FIGURESACKNOWLEDGEMENTSINTRODUCTION1.1 Background1.2 Thesis Outline1.3 Tectonic Setting: The Canadian Cordillera1.3.1 Geology and Regional Structural Pattern1.3.2 History of Formation1.4 Geophysics of the Southern Canadian Cordillera .1.5 The 1989 Southern Cordillera Refraction Experiment1.5.1 Location1.5.2 Field Experiment1.5.3 Surveying and Timing1.5.4 Recording Instruments and Data Collation2 2-D INTERPRETATION OF LINE 1:INTERMONTANE BELTIntroductionTectonic History and Geology of the Intermontane Belt .Previous Geophysical WorkSeismic Refraction DataInterpretation Method For In-line Data2—D Modelling of Line 1 Refraction DataUpper crustiv11• • • . viiiix• . . . xiv113336811111213132.12.22.32.42.52.62.6.115151516204650512.6.2 Middle Crust.562.6.3 Lower Crust and Upper Mantle 562.6.4 Sub-Moho Reflector 582.6.5 Model Resolution and Absolute Parameter Uncertainties . . . . 582.7 Principal Features of the Velocity Model 592.8 Comparison With Lithoprobe Reflection Data 612.9 Summary 663 2-D INTERPRETATION OF LINE 3: COAST BELT . 713.1 Introduction 713.2 Tectonic History and Geology of the Coast Belt 713.3 Previous Geophysical Work 743.4 Seismic Refraction Data 753.5 2—D Modelling of Line 3 Refraction Data 863.5.1 Model Resolution and Absolute Parameter Uncertainties . . . 1023.6 Principal Features of the Velocity Model 1023.7 Discussion of Results 1043.8 Summary 1104 3-D INTERPRETATION METHOD 1124.1 Introduction 1124.2 Tomographic Methods and Previous 3—D Seismic Surveys . . 1124.3 Inversion of First Arrival Traveltimes for Slowness 1154.4 Forward Modelling of Reflection Times 1204.4.1 Simple Method 1214.4.2 Accurate Method 1234.5 Inversion of Reflection Traveltimes for Depth 1274.5.1 Tests Using Synthetic Data 1294.5.1.1 Sine wave reflector model 131V5.3.15.3.25.3.35.3.45.4.15.4.25.4.31341341371421421421451451461461461471471491511561591811821821841851881921951951994.64.5.1.2 Vertical fault modelInversion of First Arrival Traveltimes for Depth4.7 3—D Modelling Algorithm56ANALYSIS OF 3—D DATAIntroduction3—D DatasetDescription of Fan-Shot and Other DataCorner Shots (SPs 1, 3 and 5)Shots at Line Midpoints (SPs 4, 12 and 17)Centre Shot (SP 7)Lines 4, 5 and 6Modelling the 3—D DataStarting ModelModelling Upper Crustal VelocitiesModelling Deeper StructureModel Resolution and UncertaintyResultsResults for a 1—D starting modelComparison With Other Geophysical DataMagnetotelluric DataHeat Flow DataGravity DataDiscussion of ResultsSummary5.15.25.35.45.55.65.75.85.96.16.25.6.15.7.15.7.25.7.3SUMMARY AND CONCLUSIONSAnalysis of the Spatial Seismic Refraction Recording MethodAnalysis of the 3—D Modelling Procedurevi6.3 Summary.2007 REFERENCES 209APPENDIX A Fan and Line 4 Record Sections 218viiLIST OF TABLESTable 2.1 Number of observations and average uncertainties for line 1 29Table 2.2 Inversion results for line 1 54Table 2.3 Lateral resolution and absolute uncertainty of parameters for line 1 59Table 3.1 Number of observations and average uncertainties for line 3 84Table 3.2 Inversion results for line 3 102Table 3.3 Lateral resolution and absolute uncertainty of parameters for line 3 . . . . 103Table 5.1 Number of data used in 3—D analysis 144Table 5.2 Rms traveltime residuals for inversion of Pg 151Table 5.3 Rms traveltime residuals for inversion of PP 156Table 5.4 Rms traveltime residuals for inversion of P 157Table 5.5 Lateral resolution and absolute uncertainty of 3—D model 159Table 5.6 Average velocity and standard deviation at depths of horizontal slices . . 163Table 5.7 Comparison of 3—D model and reflection interpretation TW1]’s 170Table 5.8 Average velocities and depth to Moho in the Coast and Intermontane Belts 189viiiLIST OF FIGURESFigure 1.1Figure 1.2Figure 1.3Figure 2.1Figure 2.2Figure 2.3Figure 2.4Figure 2.5Figure 2.6Figure 2.7Figure 2.8Figure 2.9Figure 2.10Figure 2.11Figure 2.12Figure 2.13Figure 2.14Figure 2.15Trace normalized record section for SP 3 into line 1Trace normalized record section for SP 18 into line 1 .Trace normalized record section for SP 2 into line 1Trace normalized record section for SP 17 into line 1 .Trace normalized record section for SP 16 into line 1 .Trace normalized record section for SP 19 into line 1 . .Trace normalized record section for SP 1 into line 1Representative ray paths for phases modelled on line 1 .Ray tracing diagram, traveltime comparison, synthetic andsection for SP 3 into line 1Ray tracing diagram, traveltime comparison, synthetic andsection for SP 18 into line 1Ray tracing diagram, traveltime comparison, synthetic andsection for SP 2 into line 1 .Ray tracing diagram, traveltimesection for SP 17 into line 1Ray tracing diagram, traveltimesection for SP 16 into line 1ix22232425262728observed record31observed record33observed record35observed record37observed record39Morphogeological belts and superterranes in the Canadian Cordillera . . . . 2Map of the study area showing recording geometry 4Terranes of the southwestern Canadian Cordillera 6Line 1 location map 17Previous refraction surveys in the southwestern Canadian Cordillera . . . . 1921comparison, synthetic andcomparison, synthetic andFigure 2.16 Ray tracing diagram, traveltime comparison, synthetic and observed recordsection for SP 19 into line 1 41Figure 2.17 Ray tracing diagram, traveltime comparison, synthetic and observed recordsection for SP 1 into line 1 43Figure 2.18 Location of line 1 boundary and velocity nodes 48Figure 2.19 Final 2—D velocity model for line 1 52Figure 2.20 Ray coverage provided by each phase for line 1 53Figure 2.21 Comparison of line 1 model with reflection profile 88—11 interpretation . 62Figure 3.1 Line 3 location map 72Figure 3.2 Trace normalized record section for SP 8 into line 3 77Figure 3.3 Trace normalized record sections for SPs 9 and 10 into line 3 78Figure 3.4 Trace normalized record section for SP 4 into line 3 79Figure 3.5 Trace normalized record section for SP 21 into line 3 80Figure 3.6 Trace normalized record section for SP 20 into line 3 81Figure 3.7 Trace normalized record section for SP 3 into line 3 82Figure 3.8 Representative ray paths for phases modelled on line 3 83Figure 3.9 Final 2—D velocity model for line 3 87Figure 3.10 Location of line 3 boundary and velocity nodes used in inversion 88Figure 3.11 Ray tracing diagram, traveltime comparison, synthetic and observed recordsection for SP 8 into line 3 89Figure 3.12 Ray tracing diagram, traveltime comparison, synthetic and observed recordsection for SP 9 into line 3 91Figure 3.13 Ray tracing diagram, traveltime comparison, synthetic and observed recordsection for SP 10 into line 3 92xFigure 3.14 Ray tracing diagram, traveltime comparison, synthetic and observed recordsection for SP 4 into line 3 93Figure 3.15 Ray tracing diagram, traveltime comparison, synthetic and observed recordsection for SP 21 into line 3 95Figure 3.16 Ray tracing diagram, traveltime comparison, synthetic and observed recordsection for SP 20 into line 3 97Figure 3.17 Ray tracing diagram, traveltime comparison, synthetic and observed recordsection for SP 3 into line 3 99Figure 3.18 Ray coverage provided by each phase for line 3 101Figure 4.1 Flowchart for modelling first arrival traveltimes for slowness 119Figure 4.2 General procedure for calculating finite-difference reflection traveltimes 121Figure 4.3 “Simple” method for generating finite-difference reflection times 122Figure 4.4 Test model for comparing methods of generating finite-difference reflectiontimes 123Figure 4.5 Errors for “simple” method of generating finite-difference reflection times 124Figure 4.6 “Accurate” method for generating finite-difference reflection times . . . . 125Figure 4.7 Errors for “accurate” method of generating finite-difference reflection times 128Figure 4.8 Flowchart for modelling reflection traveltimes for reflector structure . . . 130Figure 4.9 Schematic representation of 3—D finite-difference model used in tests and realdata analyses 131Figure 4.10 Results for sine wave reflector model 133Figure 4.11 Results for vertically faulted reflector model 135Figure 4.12 Illustration of inaccuracy in the inversion of first arrival traveltimes fordepth 137xiFigure 4.13 Flowchart for the complete 3—D modelling algorithm used in real dataanalysis 140Figure 5.1 Map of shot points and receivers use in 3—D analysis 143Figure 5.2 Elevation corrections applied to data used in 3—D analysis 145Figure 5.3 2—D modelling results for line 4 148Figure 5.4 Starting Moho depth model and upper-lower crust boundary 149Figure 5.5 Rms traveltime residual versus iteration number for modelling of upper crust 151Figure 5.6 Misfits from inversion of Pg traveltimes for upper crustal velocity . . .. 152Figure 5.7 Misfits from inversion of PmP traveltimes for lower crustal velocity . . . 154Figure 5.8 Misfits from inversion of P traveltimes for upper mantle velocity . . .. 155Figure 5.9 Checkerboard test for resolution of Pg data 158Figure 5.10 Horizontal slices through final model at z =2.8, 7.6, 12.4 and 18.4 km 161Figure 5.11 Horizontal slices through final model at z =23.2, 28.0, 32.8 and 37.6 km 162Figure 5.12 Location of 1—D velocity profiles and vertical slices through 3—D model 164Figure 5.13 1—D velocity profiles at various locations and average 1—D model . . . 165Figure 5.14 Velocity at base of upper crust, top and base of lower crust, and top of uppermantle 166Figure 5.15 Percentage of study area sampled versus depth 167Figure 5.16 Modelling results for 3—D Moho 169Figure 5.17 Vertical slice through 3—D model along north-south profile 172Figure 5.18 Vertical slices through 3—D model along profiles near lines 1—3 173Figure 5.19 Ray density plots for profiles near lines 1—3 174Figure 5.20 Vertical slices through 3—D model along four east-west profiles 177Figure 5.21 Ray density plots for four east-west profiles 178Figure 5.22 Vertical slices through 3—D model along profiles near lines 4, 5 and 10 179xuFigure 5.23 Ray density plots for profiles near lines 4, 5 and 10 180Figure 5.24 Representative results using a 1—D starting model 183Figure 5.25 Bouguer anomaly map and terrane map for 3—D study area 187Figure 5.26 Map of possible extent of lower Wrangellia in southern Coast Belt . . 190Figure 6.1 Hypothetical experimental design for improved 3—D recording 198Figure 6.2 Schematic interpretation of present crustal structure across the Coast andIntermontane Belts 205Figure A.1 Trace normalized record section for SP 1 into line 3 219Figure A.2 Trace normalized record section for SP 3 into line 2 220Figure A.3 Trace normalized record section for SP 5 into line 1 221Figure A.4 Trace normalized record section for SP 4 into line 1 222Figure A.5 Trace normalized record section for SP 4 into line 2 223Figure A.6 Trace normalized record section for SP 12 into line 1 224Figure A.7 Trace normalized record section for SP 12 into line 3 225Figure A.8 Trace normalized record section for SP 17 into line 2 226Figure A.9 Trace normalized record section for SP 17 into line 3 227Figure A.10 Trace normalized record section for SP 7 into line 1 228Figure A. 11 Trace normalized record section for SP 7 into line 2 229Figure A.12 Trace normalized record section for SP 7 into line 3 230Figure A.13 Trace normalized record sections for SPs 17 and 7 into line 4 231Figure A.14 Trace normalized record sections for SPs 3 and 1 into line 4 232Figure A.15 Trace normalized record sections for SPs 12 and 4 into line 4 233Figure A. 16 Trace normalized record sections for SPs 21 and 20 into line 4 234xli’ACKNOWLEDGEMENTSI thank my research supervisor, Bob Ellis, for his support, advice, patience, andconfidence in me over the years. I thank Ron Clowes for his frequent input and enthusiasticsupport of my work. Thanks to Cohn Zelt for too many Things to list, but including hisextensive advice on 2—D modelling, which CDs to buy, and what chord to play. Thanks toJohn Hole for numerous discussions and invaluable advice on 3—D modelling. Thanks toboth Cohn Zelt and John Hole for providing excellent 2—D and 3—D, respectively, analysiscodes. I thank Jim Monger for his efforts to educate me on Cordilleran geology, and for hisinterpretational input. Thanks to John Amor for his help, above and beyond the call of duty,with my many computer-related problems. I also wish to thank the 35 or so SCoRE ‘89 fieldparticipants, especially Cal Deschene, for making the recording program such a success.I sincerely thank my friends in the Department of Geophysics and Astronomy for makingmy time here so very enjoyable. Unfortunately, there are (and have been) too many to listindividually. Finally, for making life in and around Vancouver even more enjoyable, I thankmy parents, my brother Mike, and the cat behind Metallurgy.Principal financial support for the refraction program was from Lithoprobe, which isfunded by a Natural Sciences and Engineering Research Council of Canada (NSERC)Collaborative Project and Programs grant. Supplementary funding was derived from NSERCResearch Grants to R.M. Clowes, R.M. Ellis and E.R. Kanasewich. Analysis was fundedby the Energy Mines and Resources Canada Research Agreements Program and an NSERCResearch Grant to R.M. Ellis. In part, I have been financially supported by an NSERCPostgraduate Scholarship and a University of British Columbia Graduate Fellowship.gnixiv1 INTRODUCTION1.1 BackgroundUntil the 1960’s the Cordilleran orogeny was viewed as the end result of geosynclinaldeposition. With the advent of the plate tectonics hypothesis a new, mobilistic view ofCordilleran paleogeography, involving interactions between various oceanic plates and theNorth American plate and possible arc-continent collisions, has emerged (Monger 1993).Concomitant with this view has come the recognition of “terranes” (crustal blocks thatpreserve a distinct geological record) as fundamental building blocks of the continent (Irvingand Yole 1972; Coney et al. 1980; Monger et al. 1982; Monger 1993). The accretionof terranes to the continental margin is an important process by which large amounts ofnew continental crust are created. The Canadian Cordillera, with a recognizable history ofterrane accretion dating back to the Middle Jurassic (180—185 Ma), is a natural laboratoryfor the study of this and other processes associated with lithospheric evolution and growthof the continent.Geological investigations in the Canadian Cordillera began more than 120 years ago(Monger 1993). Geophysical studies of the crust and upper mantle began in earnest inthe 1950’s (Berry et al. 1971). These studies revealed much about the complex tectonicprocesses and current tectonic regime and architecture of the Cordillera and formed thegroundwork for more detailed investigations under the auspices of Lithoprobe, Canada’snational earth science research project initiated in 1984. The Lithoprobe Southern Cordilleratransect (Fig. 1.1 b) is one of ten study areas throughout Canada in which the processes andstructures associated with continental evolution are being addressed. The general objective ofthe Southern Cordillera transect is to understand the nature, timing, and dynamic evolutionof the Cordillera (Clowes et al. 1992), and in particular, the structures resulting from,and processes involved in, terrane accretion. To achieve this requires a multidisciplinaryapproach encompassing a broad range of geoscientific studies including the acquisition and1interpretation of seismic reflection, seismic refraction, gravity, heat flow and electromagneticdata, geological mapping, geochemical, geochronological, paleomagnetic and other studies(Clowes et al. 1992).490Figure 1.1 (a) Morphogeological belts of the Canadian Cordillera. Solid black lines formingtriangle denote profiles of the 1989 Southern Cordillera Refraction Experiment (SCoRE ‘89).Broken grey lines denote SCoRE ‘90 refraction profiles. Region in grey box is expanded in Fig.1.2. (b) Distribution of superterranes. The Insular superterrane comprises two smaller terranes, theIntermontane superterrane comprises four smaller terranes. Note that the eastern boundaries of thetwo superterranes approximately coincide with the Coast and Omineca Belts. Grey boxshows location of the Lithoprobe Southern Cordillera transect. (Modified from Mongeret al. 1994a.) (c) Location of three SCoRE ‘90 profiles referred to in text.Seismic refraction/wide-angle reflection data (hereinafter referred to as “refraction data”)were acquired by Lithoprobe in two stages: the 1989 Southern Cordillera RefractionExperiment (SCoRE ‘89), located in southwestern British Columbia, and the 1990 recordingprogram (SCoRE ‘90), located in southeastern British Columbia with one profile in theCoast Belt (Fig. 1.1 a). The general objectives of the surveys were to (i) provide quantitativevalues for velocity variation with depth; (ii) map laterally varying velocity structure; and (iii)map the topography of prominent velocity discontinuities such as the crust-mantle (Moho)boundary. This information places important constraints on the composition and state of the(a)200 km2crust and uppermost mantle. In addition, these surveys complement the Lithoprobe seismicreflection profiles in the southern Cordillera by providing important velocity information toaccurately translate the reflection data from time to depth sections and improve control onthe stacking and migration procedures of reflection data processing.An important aspect of the 1989 survey was the triangular recording geometry andabundance of broadside (fan) shots, providing, in addition to two-dimensional (2—D) coveragealong three long profiles, three-dimensional (3—D) coverage of a region centered on theCoast-Intermontane Belt (CB-IMB) boundary. This thesis presents results from SCoRE ‘89,in particular, the modelling and interpretation of data from two of the three main in-lineprofiles and an interpretation of the broadside data, including the development of a newprocedure with which to analyze these data.1.2 Thesis OutlineThe remainder of this chapter outlines the tectonic setting and general geophysicalcharacteristics of the Canadian Cordillera and presents details of the refraction survey.Chapter 2 describes the general method used to interpret the in-line data followed by acomplete 2—D analysis of the line 1 (Fig. 1.2) dataset. Chapter 3 presents a 2—D analysis ofline 3. Chapter 4 describes the methods used to interpret the broadside data and Chapter 5presents the results of the 3—D analysis. The results are summarized in Chapter 6.1.3 Tectonic Setting: The Canadian Cordillera1.3.1 Geology and Regional Structural PatternIn North America the name Cordillera refers to that part of the continent lying west ofthe eastern face of the Rocky Mountains. It comprises a series of nearly parallel mountainranges and includes intermontane valleys, basins and plateaus.In Canada, the Cordillera is divisible into five physiographically distinct belts (Mongerand Price 1979) (Fig. 1.la). The Foreland, Intermontane and Insular Belts comprise3480Figure 1.2 Study area for the 1989 Southern Cordillera Refraction Experiment. Numbered solidtriangles are shot points. Broken lines represent approximate location of receiver sites.Profile numbers are in circles. FRF, Fraser River Fault (broken line); PF, PasaytenFault; YF, Yalokom Fault; JFP, Juan de Fuca Plate; H, Hope; HM, 100 Mile House;K, Kamloops; P, Princeton; PA, Port Albemi; V, Vancouver; VI, Vancouver Island.Volcano symbols near SP 11 lie within the Garibaldi volcanic belt.unmetamorphosed and low-grade metamorphic rock and form the Cordilleran superstructure.The Omineca and Coast Belts are major regional, possibly tectonic, welts in which themetamorphic and plutonic infrastructure of the Cordillera is exposed. They comprisehigh-grade metamorphic and granitic rocks.The Foreland Belt consists mainly of mid-Proterozoic to Upper Jurassic miogeoclinal andplatformal carbonate and craton-derived clastic rock and overlying Jurassic-Paleogene clastic50°4rock. The Intermontane Belt consists primarily of upper Paleozoic to Tertiary volcanic andcomagmatic plutonic and sedimentary rock. The Insular Belt consists mainly of sedimentaryand volcanic rocks of mainly Paleozoic through Tertiary ages with comagmatic intrusiverocks. The Omineca Belt straddles the zone of overlap of autochthonous terranes to theeast and the allochthonous Intermontane superterrane to the west. It consists mainly ofArchean/Proterozoic crystalline basement, mid-Proterozoic to mid-Paleozoic sedimentary,volcanic and granitic rock that was either part of the Precambrian North American craton orformed along its margin, and upper Paleozoic and lower Mesozoic volcanic and sedimentaryrock in part, the eastern margin of the Intermontane superterrane. The Coast Belt lies alongthe boundary between the Intermontane and Insular superterranes. It consists primarily ofMiddle Jurassic through early Tertiary granitic rock and variably metamorphosed Paleozoicthrough Tertiary volcanic and sedimentary rocks.The Intermontane and Insular Belts, and flanking parts of the Omineca and CoastBelts, contain late Paleozoic and early Mesozoic strata and coeval granitic rocks that form,respectively, the Intermontane and Insular superterranes (Fig. 1.1). The Intermontanesuperterrane comprises four smaller terranes (Fig. 1.3) that were together by the end ofTriassic time (210 Ma) and was thrust eastward over North American rocks in the lateEarly Jurassic (180 Ma). The Insular superterrane comprises two smaller terranes thatwere amalgamated by Late Jurassic time (150 Ma) and collided with terranes to the east inCretaceous time (120 Ma). Rocks of the two superterranes probably formed in ocean basinsas oceanic crust and island arcs and later were added (accreted) to the ancient continentalmargin. In the south, between the two superterranes, are juxtaposed several small terranesof the Coast Belt comprising oceanic and island arc rocks (Fig. 1.3).The times at which the two superterranes were accreted correspond to the times at whichthe initiation of widespread Mesozoic metamorphism and crustal shortening occurred in thetwo tectonic welts. This observation, together with the rough correspondence between thelocations of the two welts and the eastern boundaries of the two superterranes, forms the basis552°500480120°Figure 1.3 Terranes of the southwestern Canadian Cordillera. Insular Belt terranes: WR,Wrangellia. Coast Belt terranes: BR, Bridge River; CD, Cadwallader; CK, Chilliwack; HA,Harrison; MT, Methow; SH, Shuksan. Intermontane Belt terranes: CC, Cache Creek; QN,Quesnellia; SM, Slide Mountain; ST, Stikinia. Also shown are PR, Pacific Rim; CR, Crescent(continental shelf terranes), and KO, Kootenay terrane (Omineca Belt). m, undifferentiatedmetamorphic rocks. Terrane boundaries from Gabrielse and Yorath (1989).of the popular collision model (Monger et al. 1982) for the origin of the Coast and OminecaBelts. This and other models for the formation of the Coast Belt, as well as more detaileddiscussions of the geology of the southwestern Cordillera are given in Chapters 2 and 3.1.3.2 History of FormationDetailed descriptions of the tectonic evolution of the Canadian Cordillera have been given128° 126° 124° 122°6by several authors (e.g., Monger and Price 1979; Coney et al. 1980; Gabrielse and Yorath1992; Oldow et al. 1989; Clowes et al. 1992; Monger et al. 1994a). The dominant tectonicprocess responsible for the present crustal structure is believed to be the accretion of two large,composite, allochthonous terranes to the ancient western margin of North America beginningabout 180 Ma (Monger et al. 1982). The present structure of the Cordillera, however, resultsfrom a complex series of events which began much earlier. The geological foundation forthese events was the western margin of the Late Archean to Early Proterozoic (2.8—1.8 Ga)North American crystalline basement, the southwestward extension of the Canadian Shield.During at least three periods of extension, in the Middle and Late Proterozoic (1600and 750 Ma) and early Paleozoic (500 Ma) the basement was rifted producing an ancientocean basin, or possibly back-arc basin, to the west of a west-facing passive margin. Thicksequences of sediments comprising shelf, slope and rise deposits accumulated in the basinforming a northeasterly tapering sedimentary (miogeoclinal) wedge which now forms the coreof the Rocky Mountains. Accumulation of sedimentary rock continued from about 1600 to380 Ma, or about three quarters of the total length of time during which the Cordilleracan be recognized as a distinct tectonic unit (Monger and Price 1979). The western marginremained passive until approximately the Late Devonian (370 Ma) when an island arc system,formed to the west between 400—500 Ma, may have collided with North America producingextensive volcanism and plutonism west of the miogeocline. In this scenario, the island arcseparated the miogeocline from the paleo-Pacific ocean to the west, presumably causing theformer passive margin to become the east flank of a mid-Paleozoic back-arc system.The basic configuration of the Canadian Cordillera, with its five distinctive geomorphological belts, was established between the latest Early Jurassic and early Tertiary (180—50 Ma).During this period two large previously amalgamated superterranes composed of intraoceanicarc and ocean floor rocks of Paleozoic and younger ages collided with and were accreted to7the continental margin. This process was accompanied and followed by crustal thickening byseveral mechanisms (primarily between 115 and 60 Ma): stacking of thrust sheets, dextraltranspression (i.e., compression with a component of right-lateral shear), arc magmatismrelated to subduction, regional metamorphism and subduction of Pacific ocean (Kula/Farallonplate) crust. The accretion of terranes expanded the continental area westward and wasaccompanied by a westward migration of the locus of convergence to the west of VancouverIsland where it continues today with the east-dipping subduction of the Juan de Fuca plate.Beginning in the early Late Cretaceous (85 Ma) and continuing through the Eocene (until40 Ma) large, right-lateral strike-slip faults within the interior of the Cordillera formed inresponse to dextral shear that probably resulted from oblique convergence between the NorthAmerica plate and various contiguous Pacific plates. These faults sliced through structuresand accommodated northward motion of the (fragmented) terranes relative to North America.In the early Eocene (60 Ma) the contractional regime changed to one of uplift andextension with continued strike-slip activity, likely caused by changes in the relative motion ofthe Farallon, Kula, Pacific and North America plates. The extensional activity was associatedwith subduction related magmatic arc activity.Since the mid-Tertiary (40 Ma) the Cordillera has remained relatively quiescent, whilein the west magmatic arc and back-arc activity has been associated with subduction of theJuan de Fuca plate, producing the Cascade magmatic arc, whose northern end is in thesouthern Coast Belt, and the Neogene Chilcotin basalts, which are in a back-arc setting inthe southern Intermontane Belt.1.4 Geophysics of the Southern Canadian CordilleraOur understanding of the geophysics and geology of the southern Canadian Cordillera hasimproved enormously as a result of recent multidisciplinary studies throughout Lithoprobe’s8Southern Cordillera Transect (Clowes et al. 1992). Prior to this concerted effort, begun in1984, earlier studies had succeeded in characterizing the primary geophysical features of thisregion (e.g., Berry et al. 1971; Gough 1986).Estimates of crustal thickness throughout the Canadian Cordillera were obtained fromearly seismic refraction work (e.g., White and Savage 1965; White et al. 1968; Bennett etal. 1975; Berry and Forsyth 1975; Cumming et al. 1979). These studies revealed a thin(30—33 km) crust beneath the Intermontane Belt, thickening to the east to 40—50 km beneaththe Rocky Mountains. Average crustal velocities were low (6.1—6.4 km/s) as were uppermantle velocities (<8.0 kmls). Very little reliable information about structure beneath theCoast Belt was obtained.More detailed velocity models for the Insular and westernmost Coast Belts were obtainedfrom refraction data recorded along Vancouver Island (McMechan and Spence 1983; Drewand Clowes 1990) and along a profile extending across Vancouver Island to the Garibaldivolcanic arc (Spence et al. 1985; Drew and Clowes 1990). In particular, the subductingJuan de Fuca plate was shown to have an average dip of 14° to 16° and a depth of 30-50km beneath Vancouver Island. Crustal velocities were much higher than to the east (6.4—7.0kmls) though deep structure (>20 km) was not well constrained. Depth to the Moho wasestimated to be 38 km beneath Vancouver Island.Rayleigh and Love wave phase velocities were used by Wickens (1977) to determinelithospheric thickness throughout southern British Columbia. No significant upper mantle( 240 km depth) low-velocity zone beneath Vancouver Island or the Coast Belt was found.Beneath the Intermontane Belt, a low-velocity zone 40—50 km thick was found, separatedfrom the Moho by a lid 15—20 km thick. Wickens concluded that the base of the lithospherecorresponds with the base of the crust in southwestern British Columbia, and extends slightlyinto the upper mantle to the northeast.9The lower crust and/or upper mantle beneath much of the southern Cordillera has highelectrical conductivity (in comparison with the craton) associated with three major conductivestructures (Gough 1986). The largest of these, the Canadian Cordilleran Regional (CCR)conductor, extends throughout the Intermontane and Omineca Belts, though its westernextent is poorly defined. The top of the CCR conductor lies within the lower crust (15—30km) and extends downward 30—70 km into the upper mantle. Gough (1986) argues thatthe CCR conductor consists of silicate melt below and highly saline water in the crust,both derived from mantle upfiow beneath the Intermontane and Omineca Belts. The othertwo conductive structures, the Northern Rockies conductor and the Southern Alberta-BritishColumbia conductor, occur in the eastern Cordillera and southeastern Alberta, and are muchsmaller in extent compared to the CCR.Heat flow throughout most of the southern Cordillera east of the Garibaldi volcanicbelt (Fig. 1.2) is relatively high (Davis and Lewis 1984; Lewis et al. 1988; Lewis et al.1992). From west to east the primary thermal pattern is: (i) low crustal temperatures fromwestern Vancouver Island to the heads of the Coast Mountain fjords (20 km west of theGaribaldi volcanic belt), (ii) very high and variable temperatures within the volcanic belt dueto hydrothermal cooling of intrusive bodies, and (iii) high and uniform crustal temperaturesin the central and eastern Cordillera to the east (Lewis et al. 1992). Low heat flow in theInsular Belt is associated with the heat sink of the subducting Juan de Fuca plate. The highand uniform heat flow in the east is attributed to a steady, deep mantle source of heat (Davisand Lewis 1984).Details of some of these previous studies and more recent studies relevant to this thesisare discussed in Chapters 2, 3 and 5.101.5 The 1989 Southern Cordillera Refraction Experiment1.5.1 LocationThe Lithoprobe 1989 Southern Cordillera Refraction Experiment (SCoRE ‘89) wasdesigned to image the crustal and upper mantle velocity structure of the southwesternCanadian Cordillera in both two and three dimensions. To meet this objective requiredboth in-line recording and an extensive fan-shot program. To this end, a roughly triangulararrangement of shots and receivers centered over the Fraser River Fault system was employed(Fig. 1.2). This geometry, called the “spatial seismic refraction recording” method byKanasewich and Chiu (1985), allowed for in-line recording along three strategically locatedand approximately linear profiles, and a fairly regular distribution of fan shots into theseprofiles.The three profiles forming the sides of the triangular array are numbered 1, 2 and 3(Fig. 1.2). Line 1 is a north-south line, 330-km-long, approximately along-strike throughthe Intermontane Belt. The north end (shot point (SP) 1)is 70 km northeast of 100 MileHouse; the south end (SP 3) is 3O km south of Princeton. Line 2 runs southwest from SP1 for 360 km to the southwest corner of the triangle on the Sechelt Peninsula (SP 5), 60km northwest of Vancouver, continuing -90 km to the west across the Strait of Georgia toabout 40 km northwest of Port Alberni on Vancouver Island (SP 8). Line 2 traverses thethree westernmost belts and is nearly perpendicular to the northwest trending strike of thelocal geology. Line 3 runs west from SP 3 obliquely across the Coast Belt for 240 km toSP 5. For modelling purposes the western extension of line 2 onto Vancouver Island (SP 8)from SP 5 is also considered to be a part of line 3.In addition to the three main profiles, three short profiles (4, 5 and 6) were recorded inthe interior of the triangle. Line 4 runs northeast for 90 km from SP 7 at the center of thearray (70 km north of Hope) to SP 17, the center shot along line 1 located 30 km west ofKamloops. Line 5 runs west from SP 7 for 30 km and line 6 runs 40 km to the southeast11from SP 7 approximately along the surficial boundary between the Coast and IntermontaneBelts (Pasayten Fault). Lines 5 and 6 are unreversed profiles.1.5.2 Field ExperimentData were recorded during three deployments of recording instruments. During the firstdeployment, line 2 (including the western extension to Vancouver Island) was recorded.Lines 1 and 5 were recorded on the second deployment and lines 3, 4 and 6 were recordedon the third deployment. A total of 296 portable seismograph units, recording the verticalcomponent of ground motion only, were used. These provided average receiver spacings of1.2 km along lines 1 and 3, and 1.6 km along line 2.Shots were detonated at seventeen locations around the triangle (note: SP 22 on theSechelt Peninsula did not detonate), one site interior to the triangle, two sites on VancouverIsland and at two locations offshore over the Juan de Fuca plate. Several of these sites wereused more than once. In general, three charge sizes, all in 20 cm diameter holes, were used,depending on the location of the shot point. Large (1800 kg of Nitropel® in two 40 m deepholes) charges were used at the corners of the triangle (SPs 1, 3 and 5) and at the west endof lines 2 and 3 (SP 8) on Vancouver Island. Medium size charges (800 kg in one 40 m deephole) were used at SP 7 in the center of the array and at sites located at the approximatemidpoint of each of the three profiles (SPs 4, 12 and 17). Small charges (200 kg in one 30 mdeep hole) were used at the remaining shot points (2, 6, 9, 10, 11, 13—16, 18—21). In general,all of the 800 and 1800 kg shots were recorded at all receiving sites around and interior tothe triangle, i.e., these were used for both in-line recording and as fan shots. The exceptionsare SP 5 which was recorded along line 1 only, and SP 8 which was recorded along lines 2and 3 only. The 200 kg shots were used for in-line recording only. The average shot spacingis about 55 km. The two offshore shots (SPs 98 and 99; Fig. 1.2), each comprising 1000kg of explosives, were recorded along line 2 only. They were included to allow an eastward12continuation, beyond the volcanic arc, of previous interpretations of the velocity structure ofthe Juan de Fuca subduction zone (Spence et al. 1985; Drew and Clowes 1990).1.5.3 Surveying and TimingAll shot points were positioned using global positioning system (GPS) receivers. Mostcoordinates are based on 3—D satellite fixes (i.e., four satellites used in the positioningsolution) and instrument specifications claim they should be accurate to within 10 m.Coordinates for SPs 3, 6, 9, 14, 15, 18 and 21 are based on 2—D fixes (i.e., three satellitesused in position solution) and should be accurate to within 50 m. Experience has shown thatthese error estimates are, in most cases, probably too optimistic. All shot sites were alsolocated on 1:50,000 scale topographic maps and, together with the GPS fixes, are probablyaccurate to within 50 m.Of the 866 receiver sites approximately 83% were located with 2—D and 3—D GPS fixes.The remaining sites were located using 1:50,000 scale topographic maps either because theGPS satellite fixes could not be obtained due to topographic or forest cover interference, orbecause it was possible to read coordinates to sufficient accuracy from maps. All receivercoordinates should be accurate to within 50 m.Timing was obtained by rating shot and receiver clocks against geostationary orbitingenvironmental satellite (GOES) time code before and after each deployment. Drift correctionswere made assuming a linear clock drift between deployment and retrieval. The estimatedclock accuracy in all cases is 10 ms.1.5.4 Recording Instruments and Data CollationThe 296 portable seismographs were of four different types: 154 Scintrex-EDA modelPRS1; 14 Scintrex-EDA model PRS4; 8 Geotech MCR 600 systems; and 120 USGS cassetteunits. The first three types recorded information digitally; the USGS units are analog13recorders. The relative velocity sensitivity of the four instrument types varies but the averagerange of uniform response is fairly consistent (2—12 Hz) (see Zelt et al. 1990).Data recorded on the analog instruments were obtained in digital form from the USGSand subsequently gained to produce units of nm/s common with the digital data. All datawere reduced to a common 120 Hz sample rate, converted to the Lithoprobe data storageformat (SEGY-LDS, version 2.0; Spencer et a!. 1989) and merged.142 2-D INTERPRETATION OF LINE 1: INTERMONTANE BELT2.1 IntroductionIn this chapter a 2—D velocity model for line 1, recorded along-strike within theIntermontane Belt (Fig. 1.2), is presented. A brief discussion of the geology, tectonic history,and geophysics of the Intermontane Belt is followed by a presentation and description of therefraction dataset for line 1. The general procedure for modelling the in-line data using aniterative combination of traveltime inversion and amplitude forward modelling is presentednext. This method also applies to the analysis of the line 3 (Chapter 3) data. Details ofthe specific modelling procedure for line 1 are discussed, followed by a presentation of thevelocity model. The chapter concludes with a comparison of the model with an interpretationof Lithoprobe reflection data (Cook et al. 1992) and a summary.Zelt et al. (1992) include the principal scientific information in this chapter. Mycontribution to this paper and the 1989 field program included: (1) lead role in receiver andshot point site location and surveying; (ii) minor survey design modifications; (iii) instrumentdeployment during recording program; (iv) subsequent collation of data and preparation offinal data tapes; (v) analysis and preliminary interpretation of data; and (vi) writing first draftof the paper.2.2 Tectonic History and Geology of the Intermontane BeltA detailed description of the tectonic history and geology of the Intermontane Belt isgiven by Monger et al. (1994a). The Intermontane superterrane is an amalgamation offour smaller terranes comprising ocean floor and oceanic volcanic rocks that were probablytogether by the end of Triassic time (210 Ma), and that were then accreted to the westernmargin of North America in the late Early Jurassic (ca. 180 Ma) by eastward thrusting overthe Proterozoic craton. Convergence with North America continued through to the Eocene(ca. 60 Ma) and resulted in crustal shortening and thickening, strike-slip faulting in thewestern Intermontane Belt, the initiation of widespread metamorphism and plutonism, and15formation of the Omineca Belt to the east. Compression and transpression were followedby east-west crustal extension, transtension, and continued strike-slip faulting in the westthroughout the Eocene (60—40 Ma). Since then this part of the Cordillera has remainedrelatively quiescent.Line 1 (Fig. 2.1) is located entirely within Quesnellia, one of the terranes ofthe Intermontane Superterrane. Here Quesnellia includes lower Mesozoic volcanic andsedimentary rocks and associated granites of arc affinity that are underlain by Late Devonianthrough Permian volcanic and sedimentary strata (Monger et al. 1 994a). It is greatlyelongated along a northwest-southeast axis, extending from northern British Columbia toWashington State. In the south, numerous normal faults involving Eocene and older rocksprovide evidence for crustal extension in Eocene time (Ewing 1980), resulting in the greatwidth (300 km) of the terrane at 50° N latitude.2.3 Previous Geophysical WorkPrevious geophysical data show the Intermontane Belt to be characterized by a thin crustand lithosphere, relatively high and uniform heat flow and high electrical conductivity.Limited regional seismic refraction data from the 1960’s and 1970’s exist (Fig. 2.2)but the large receiver spacing (.- 15 km) and frequently unreversed profiles did not allowfor well-constrained models. Further, the model resolution obtained from such surveys isnot adequate to enable correlation with even large-scale geological features. White et al.(1968) analyzed traveltime data recorded approximately along-strike within the IntermontaneBelt and inferred a thin (< 3 km) near surface layer of velocity ‘5. 1 km/s overlying a crustof average uniform velocity 6.1 km/s. Their average depth to Moho was 30 km with anupper mantle velocity of “probably” less than 8.0 km/s. Berry and Forsyth (1975) analyzedseveral profiles recorded in the Canadian Cordillera prior to 1971. For a profile runningnortheast from Vancouver Island to the western Omineca Belt (Fig. 2.2) their results in theIntermontane Belt include a 32 km thick crust of average velocity 6.4 km/s and an upper16121 120°Tertiary sediments and volcanicsEocene - Late Triassic plutonsCretaceous - Paleozoic sediments and volcanics(Early Jurassic and older rocks form Quesnellia)Cache Creek terraneCoast beltFigure 2.1 Line 1 location map showing shot point (A, with size proportional to charge) andreceiver locations (+) for line 1. Broken lines show location of Lithoprobe reflectionprofiles; 88—11 is emphasized. Generalized surface geology is from Wheeler and McFeely(1991). CNH, Central Nicola horst; K, Kamloops; P, Princeton.52°51°5001210 120°17mantle velocity 7.8 km/s. They observed evidence on one record section for an intermediatelayer approximately 11 km thick with average velocity 6.6 km/s, but this was not supportedby the reversed section. Cumming et al. (1979) examined data recorded along a partiallyreversed profile running east from the Highland Valley (Fig. 2.2). In the vicinity of line 1their crustal model consists of four layers: a thin (1 km) near-surface layer of velocity 5.7km/s. an upper crust 20 km thick with average velocity 6.05 km/s. a lower crust approximately10 km thick with velocity 6.9 km/s, and an upper mantle of velocity 7.8 km/s with depthto Moho approximately 30 km. The 6.9 km/s deep crustal layer, interpreted on the basisof traveltimes and amplitudes of secondary arrivals, was considered a significant feature oftheir model. Allowing for the large shot and receiver spacing and lack of true amplitudeinformation in the two older studies, the differences in interpretations are not surprising.Recently acquired high quality Lithoprobe deep seismic reflection data image featuresat all levels within the crust. Cook et al. (1992) present an interpretation of data recordedin the southern Intermontane Belt. Cook et al. (1991) place this interpretation within thecontext of the complete Southern Cordillera transect. The reflection data show the crust tobe highly reflective at all levels, structurally complex and not divisible into transparent upperand reflective lower layers. Cook et al. (1992) attribute much of the reflectivity to crustalstructures, e.g., fault zones and primary lithologic variations; however, Lewis et al. (1992)suggest layered fluid porosity as a source for some of the reflectivity. A detailed comparisonof the interpretation by Cook et al. (1992) in the vicinity of line 1 with the refraction modelis presented in section 2.8.Heat flow measurements in the Intermontane Belt show moderately high and uniformvalues averaging ‘73 mW/ma (e.g., Lewis et al. 1992; Lewis et al. 1985; Davis and Lewis1984), indicating temperatures of about 4500 C at 14 km and 730° C at 23 km (Lewis etal. 1992). The 4500 C isotherm is generally considered to mark the brittle-ductile transitionfor mafic deep crust and often correlates with the top of seismically reflective zones withinthe deep crust (e.g., Klemperer 1987). The 730° C isotherm corresponds to the transition18128°48°128°Figure 2.2 Previous refraction surveys in the southwestern Canadian Cordillera (numbered brokenlines) in relation to SCoRE ‘89 lines and shot points (A): (1) White et al. (1968); (2) Berry andForsyth (1975); (3) Cumming et al. (1979); (4) Spence et al. (1985); (5) McMechan and Spence(1983). FRF, Fraser River Fault; HV, Highland Valley; K, Kamloops; PA, Port Alberni; V, Vancouver.from “wet” amphibolite fades to “dry” granulite facies conditions and frequently correlateswith the base of reflective zones as, for example, in the southeastern Cordillera (Lewis et al.1992). The relatively high heat flow may be the result of a very thin lithosphere, perhapsnot much thicker than the crust (Lewis et al. 1985; Gough 1986). This is supported by astudy of phase velocities of Rayleigh and Love waves which indicate a shallow upper mantlelow-velocity layer 40—50 km thick approximately 15 to 20 km beneath the Moho throughoutmuch of the southern Intermontane Belt (Wickens 1977). This has been interpreted to indicatepartial melting of mantle material due to an upcurrent of mantle convection (Gough 1986).50°126° 124° 122° 120°19Electrical conductivity throughout the southern Cordillera is high (Gough 1986). Recentmagnetotelluric (MT) studies show that the crust in the Intermontane Belt is slightly moreresistive with respect to the Coast Belt to the west and Omineca Belt to the east and thatlateral changes in crustal conductivity are predominantly cross-strike (Jones et al. 1991).High electrical conductivity appears to correlate with temperatures above 4500 C, providingfurther evidence for mantle upflow, probably currently located beneath the Omineca Belt(Majorowicz et al. 1993). Studies of Nd and Sr isotope ratios in the southern Cordilleraindicate that the westward extent of the North American craton may be as far as 75 km westof the Okanagan Valley, near the location of refraction line 1 (Ghosh 1991).2.4 Seismic Refraction DataData acquisition parameters for the refraction survey are given in section 1.5. Line1 comprised seven shot points and 269 recording instruments (Fig. 2.1). All shotswere recorded by all instruments, resulting in a dataset comprising approximately 1850seismograms of which about 1730 have observable seismic arrivals. Approximately 2700traveltime picks of first and later arrivals were obtained from these traces. Phase identificationand picks were based on several different record section formats including large and smallscale plots, trace normalized and true relative amplitude scaling, and unfiltered and filtereddata with a range of band passes. The seven record sections are displayed in trace normalized(common maximum amplitude), unfiltered, reduced traveltime format in Figs. 2.3—2.9. Afew unusable traces have been removed from each record section. In the following discussion(and in Figs. 2.3—2.9) “distance” refers to model distance (i.e., distance from SP 3 at thesouth end of the line), and “offset” refers to specific shot-receiver offsets. Shot point namesfollowed by S or N refer specifically to traces recorded to the south or north of the shot,respectively.A total of eight distinct phases, illustrated in Fig. 2.10, are identified and used inthe modelling procedure (also see Figs. 2.3—2.9). Ps, Pg and P correspond to refracted20Is-0OQ0oo•o1 C’,0CD00CD‘-•0DCDoCDCD0—.0 CDsLine1Shot3I-NU)12 10 8 6 4 2 0F1060110160210260310Distance(km)4Line1Shot18NSqQ I12 10 8 6 2 0050100150200250300Distance(km)I-’•-t-toC)0-t FsLine1Shot2N1060110160210260310Distance(km)S10r10 -t 0 -tLine1Shot178— 6—N4— 2— 01060110160210260310Distance(km)(DID(.)CDo (DO—.-t C -t C,,CDsLine1Shot16Nt’.)((21060110160210260310Distance(km)DD,110 0 -.-t 0o-t C.,1. 10sLine1Shot19N(12 E050100150200250300Distance(km)CD 0 N CD C) 0 a CD C) C,, CD C,, CD CDT10 -I 0 CD -t 0sLine1Shot1Nt%)1060110160210260310Distance(km)Figure 2.10 Representative ray paths for phases modelled: P, Pg and P are rays refractedthrough the near surface layer, upper crust, and upper mantle, respectively. R1, R2, R3,Pr, and PmP are reflections from within the upper crust, base of the upper crust, baseof the middle crust, a “floating” reflector (see section 2.6.3) within the lower crust,crust-mantle boundary, and within the upper mantle, respectively.energy through the relatively low velocity near-surface layer, upper crust and upper mantle,respectively. R1, R2 and R3 are reflections from within the upper crust, base of the uppercrust, and base of the middle crust, respectively. PmP is a reflection from the crust-mantleboundary and pm is a relatively strong reflection from within the upper mantle. Not all phaseshave been identified on all record sections because of insufficient maximum shot-receiveroffset, low signal-to-noise ratio or absence of a particular phase. Table 2.1 summarizes thenumber of traveltime picks of each phase for each shot.Data quality, in general, is high and first arrivals can be easily identified across mostsections. Only on the far-offset traces of the SP 19-S record section (Fig. 2.8) is thesignal-to-noise ratio so low that phase identification is difficult. The location of line 1within a single terrane (Quesnellia) may explain in part the efficiency with which seismicenergy was propagated. The data show a wide degree of traveltime, amplitude and phasecoherency variations from shot-to-shot. The P phase shows great variability in apparentvelocity (3.8—5.5 kmls) and is not observed on the SP 1 record section (Fig. 2.9), probablyDISTANCE (km)0 50 100 150 200 250 300I I I IPS\ pmfo - /It ,*.. &28Table 2.1 Number of observations and average uncertainty in milliseconds (parentheses) for eachphase and each shot along line 1. Totala is the total number of observations and averageuncertainty in milliseconds (parentheses) for each shot. Totalb is the total number ofobservations and average uncertainty in milliseconds (parentheses) for each phase.Shot P and Pg Ri R2 R3 Pr PmP P Ptm Totalapoint3-N 122 (53) 79 (92) 44 (85) 116 (58) 13 (79) 374 (67)18-S 54 (32) 54 (32)18-N 128 (59) 23 (99) 93 (90) 15 (143) 87 (96) 22 (72) 368 (82)2-S 83 (47) 41 (112) 124 (68)2-N 104 (53) 30 116) 69 (103) 34 (102) 31 (95) 268 (84)17-S 93 (40) 53 (76) 42 (94) 49 (112) 237 (73)17-N 102 (42) 43 (101) 21 (142) 49 (69) 11 (66) 226 (69)16-S 106 (54) 44 (77) 37 (119) 64 (106) 4 (119) 255 (81)16-N 92 (56) 33 (108) 9 (119) 134 (73)19-S 99 (54) 47 (91) 7 (121) 17 (141) 42 (87) 212 (78)19-N 50 (32) 50 (32)1-S 117 (52) 31 (78) 56 (110) 47 (107) 120 (50) 27 (49) 398 (68)Totalb 1150 (50) 107 (82) 150 (98) 423 (98) 28 (137) 369 (103) 365 (68) 108 2700 (73)(75)indicating that the near-surface layer at the north end of the line is very thin (<500 m), withbasement velocities present near the surface.Beyond the Ps—Pg crossover point, Pg appears as the first arrival to offsets of about170—180 km. Pg is nearly linear on all sections, indicating low average gradients in theupper crust and shallow penetration of energy, with deviations principally due to near-surfacevelocity variations. The average apparent velocity is 6.2 km/s. There are no large traveltimeoffsets to indicate the presence of thick low velocity zones (LVZs) and no clear breakovers29to indicate large positive vertical velocity contrasts at layer boundaries within the uppermostpart of the crust. As seen on true relative amplitude displays of the data (Figs. 2.1 ld—2.17d),the relative amplitude of Pg varies strongly from shot-to-shot; indicative of lateral changesin velocity gradient near the top of the upper crust. For example, Pg is prominent on SP 2—N(Fig. 2.13) from 130 km to 220 km. on SP 17—N (Fig. 2.14) from 170 km to 220 km, andon SP 16—N (Fig. 2.15) from 190 km to 120 km. indicating a strong gradient region centeredapproximately between SPs 17 and 16. Pg is also relatively strong on SP 3 (Fig. 2.11) to60 km and on 18—S to about 30 km, indicating another high-gradient region beneath SP 18.The amplitude of Pg is anomalously high on both SPs 2—N and 17—N at about 190—200 km,possibly because of near-surface focusing of refracted energy at this location. A sudden dropin amplitude of Pg occurs at varying shot-receiver offsets on different record sections (e.g.,at 110 km on SP 3, 100 km offset on SP 2—N and SP 16—S), most likely indicative of a rapidtransition into a region of relatively low vertical velocity gradient.Two distinct reflection phases from within the upper crust are present in the data: R1on SP 18-N, 17-S and 1—S (see Figs. 2.4, 2.6 and 2.9); and R2 on SP 2—N, 17—N, 16—Sand 16-N (Figs. 2.5, 2.6, and 2.7). The R1 reflection on SP 18—N is not strong but on SP17—S, where it appears between 30—100 km nearly parallel to and about 0.4—0.5 s behindPg, it is clearly visible and coherent. On SP 1, R1 is most clearly visible between 250—290km, about 0.25—0.6 s behind Pg. The R2 reflection is, in general, very weak but can be mosteasily seen on SP-16—S (Fig. 2.7) between 50—100 km where it is 0.25—0.6 s behind Pg. R2provides constraints on velocities in the upper crust.A fairly strong wide-angle reflection (R3) with an asymptotic apparent velocity of about6.3 km/s is present on all sections where it can be traced from shot-receiver offsets of100—130 km (in front of PP) to offsets greater than 250 km, where it is t.i3 s behind P.This consistent and prominent phase is generated at the middle-lower crust boundary at adepth of 21—26 km and is the principal constraint on velocities at the base of the middle crust.30z.DISTANCE (km)0 50 100 150 200 250 300I I INS0-0-IH00CC\20‘Ir’ ii4-‘qi9,,I I I I I I I0 50 100 150 200 250 300CDISTANCE (km)Figure 2.11 Two-point ray tracing diagram showing total ray coverage (a) andcomparison between observed (short vertical bars representing uncertainty ofpicks) and calculated (dot) traveltimes (b) for SP 3 into line 1.31EFigure 2.11 (continued) Synthetic section (c) and observed record section(d) for SP 3 into line 1. Trace amplitudes are scaled to shot-receiver offset raised tothe 1.5 power. Data in (d) are band pass filtered from 2—10 Hz.0rI)HC--0S12100 200N10—8-6-42-0- I I I I I I I I I0 50 100 150 200 250 300Distance (km)32o I I0 50Figure 2.12 Two-point ray tracing diagram showing total ray coverage (a) andcomparison between observed (short vertical bars representing uncertainty ofpicks) and calculated (dot) traveltimes (b) for SP 18 into line 1.DISTANCE (km)0 50 100 150 200 250 300IN7-S0-00C\20CD000i2.coHC\2tTLl11óliftT,‘ F (b)I I I100 150 200 250 300DISTANCE (km)33Cl)HIIhLIIIIIIIII•IlIIIIIIIIII(IIIIIIiIIIII11111111 I’’ 11110 50 100 150 200 250 300S N1210-8-C,)H2-(d)0—250 300Figure 2.12 (continued) Synthetic section (c) and observed record section (d) for SP18 into line 1. Trace amplitudes are scaled to shot-receiver offset raised tothe 1.5 power. Data in (d) are band pass filtered from 2—10 Hz.C\1CD-IIIII 111111111 IIIIIIIIIIIIIIIIIIIIII0 50 100 150 200Distance (km)34DISTANCE (km)150 200Figure 2.13 Two-point ray tracing diagram showing total ray coverage (a) andcomparison between observed (short vertical bars representing uncertainty ofpicks) and calculated (dot) traveltimes (b) for SP 2 into line 1.S?C-50 100—250 N300-1cD--0C(C(a)N—‘(0-Cl]Hc\1-Cø1‘lily4j4’11/(b)I I I I0 50 100 150 200 250 300DISTANCE (km)35H150 250 300S N10——8—H2—I—)— I””” ‘ I I I0 50 100 150 200 250 300Distance (km)Figure 2.13 (continued) Synthetic section (c) and observed record section(d) for SP 2 into line 1. Trace amplitudes are scaled to shot-receiver offset raised tothe 1.5 power. Data in (d) are band pass filtered from 2—10 Hz.CD-1(c)11111 II 11111C0 50 100 200Ii lIIIII36NCDS DISTANCE (km)‘I0 50 100 150 200 250 300—NIc)1=CC0)-(a)—IIC\2-rN, /V.’, (b)I I I I0 50 100 150 200 250 300DISTANCE (km)Figure 2.14 Two-point ray tracing diagram showing total ray coverage (a) andcomparison between observed (short vertical bars representing uncertainty ofpicks) and calculated (dot) traveltimes (b) for SP 17 into line 1.37(1)H150 250 300(d)Figure 2.14 (continued) Synthetic section (c) and observed record section (d) for SP17 into line 1. Trace amplitudes are scaled to shot-receiver offset raised tothe 1.5 power. Data in (d) are band pass filtered from 2—10 Hz.C.1cc-(c)C-0S10010-8-2006—4-2-N0-—I0 50 100 150 200Distance (km)I I I250 30038.1I00Ic)‘ CDIS0 50 1000-DISTANCE (km)EE—150 200 250 300IN0C)(7/f//i0C)(a)frf4Li— I 4%S —t,.1 (b)0 I I I0 50 100 150 200 250 300DISTANCE (km)Figure 2.15 Two-point ray tracing diagram showing total ray coverage (a) andcomparison between observed (short vertical bars representing uncertainty ofpicks) and calculated (dot) traveltimes (b) for SP 16 into line 1.39‘I_CDI2IH— I I0 150 200 250 300S N10 — — a? CJC )%)OS MWflkc(KflWYPPUt1II1 FtI?WO SA41W1 NI 11111118—I?iII6: ;:;;;:;;:;:: WII1 iI— )tIIllIlIIIlII>IIllllIlIIlI IIIIIIIIIIIIIIIIIIIIII Ill III0 50 100 150 200Distance (km)Figure 2.15 (continued) Synthetic section (c) and observed record section (d) for SP16 into line 1. Trace amplitudes are scaled to shot-receiver offset raised tothe 1.5 power. Data in (ci) are band pass filtered from 2—10 Hz.I11111111 1111150 100BHuxIi II?(Ii1IIiIIIIIIIIIIIIIIIIIINIIIII III IIi 1IIII I I I I I I I I I I I I I250 30040cncoEs DISTANCE (km)0 50 100 150 200 250 300CC‘ICC\2IHNCCCLC)Cit*t 1Mf(b) I0 50 100 150 200 250 300DISTANCE (km)Figure 2.16 Two-point ray tracing diagram showing total ray coverage (a) andcomparison between observed (short vertical bars representing uncertainty ofpicks) and calculated (dot) traveltimes (b) for SP 19 into line .1.41Figure 2.16 (continued) Synthetic section (c) and observed record section (d) for SP19 into line 1. Trace amplitudes are scaled to shot-receiver offset raised tothe 1.5 power. Data in (d) are band pass filtered from 2—10 Hz.U)HC”?(c)0 50S100 150 200 250 3001210-8-U)H2-N(d)I I I I I0 50 100 150 200 250Distance (km)I I I I I I I I I I I I I I I 1 I I I30042C,)coEDISTANCE (km)150 200Figure 2.17 Two-point ray tracing diagram showing total ray coverage (a) andcomparison between observed (short vertical bars representing uncertainty ofpicks) and calculated (dot) traveltimes (b) for SP 1 into line 1.Sc? 50 100-250 N3000-0‘II00CQ0_,1’‘ ‘liP r%fi4+ja4 TV7%(b)9%CQ0- I I I I0 50 100 150 200 250 300DISTANCE (km)43Cl)co12108642250 300Figure 2.17 (continued) Synthetic section (c) and observed record section(d) for SP 1 into line 1. Trace amplitudes are scaled to shot-receiver offset raised tothe 1.5 power. Data in (d) are band pass filtered from 2—10 Hz.‘ICC(c)11111 11111- I0 50I I I150 200 250 300S N(d)0 I I I I I I I0 50 100 150 200Distance (km)I I I I I I I I I I I I I I I I I I I I44Reflections from the crust-mantle boundary (PmP) are seen on all sections. The characterof PmP varies from shot to shot as does the location of the critical point, which ranges from90—130 km shot-receiver offset. Apparent asymptotic velocities range from 6.6 to 7.0 km/sand the reduced arrival times near the critical point vary between 6.5 to 7.3 s. Theseobservations suggest the presence of small topographic variations along the crust-mantleboundary and significant lateral velocity variations at the base of the lower crust. On SP17—N (Fig. 2.6), there appears to be an offset in the PmP arrival times of about 0.7 s at265 km where arrivals to the south are earlier. This could be an indication of a step likediscontinuity or a localized region of complexity on the Moho. Similarly, PmP arrivals on SP19—S (Fig. 2.8) are offset by about 0.6 s with traces north of about 155 km arriving earlier.This phase samples the same part of the Moho as PP from SP 17—S and again points to alocalized region of disturbance or offset in or near the crust-mantle boundary.Upper mantle refractions (Ps) are observed as first arrivals on all sections with sufficientoffsets except for SP 19—S (Fig. 2.8) where the noise level is too high at large offsets. ThePgPn crossover point occurs at 170—180 km on all sections and the apparent velocity ofP is consistently 7.9—8.0 km/s. Like Pg. P is essentially linear with only minor deviationswhich are attributed to near surface velocity variations. There is no evidence in the Parrivals to support a significant offset of the Moho and thus it is likely that the offset PmPphases seen on SP 17—N and SP 16—S are caused by a localized region of complexity nearthe crust-mantle boundary.A relatively strong secondary arrival (Pm) 0.4 s or less after P can be observed onmost sections. It is strongest on SP 1 (Fig. 2.9) where it appears to be asymptotic with P.Coherent arrivals observed on SP 19 (Fig. 2.8) in the vicinity of P are most likely Ptm. Thestrength of this phase and its traveltime characteristics indicates that it is a reflection from asub-Moho boundary. Based on an interpretation of data within the Coast and Omineca Beltsrecorded as part of SCoRE ‘90, O’Leary et al. (1993) and Kanasewich et al. (1994) alsoinfer layering in the upper mantle; their results will be discussed in more detail in section 2.9.452.5 Interpretation Method For In-line DataTwo methods are used concurrently to interpret the in-line data. Traveltimes are modelledusing the 2—D ray-trace traveltime inversion algorithm of Zelt and Smith (1992). Theadditional constraints provided by amplitude modelling are incorporated into the velocitymodel using a modified version of the 2—D ray-trace forward modelling algorithm of Zeltand Ellis (1988) which uses the same velocity parameterization as the traveltime inversionalgorithm. The ray tracing component of both of these techniques is based on asymptoticray theory ((erven9 et al. 1977). The Zelt and Smith (1992) traveltime inversion routinewas favoured over conventional forward modelling of traveltimes because it represents amore objective and efficient procedure, and provides estimates of resolution, uncertainty andnon-uniqueness of model parameters (boundary depths and layer velocities). In addition, thehigh density shot and receiver coverage and presence of both refracted and reflected crustaland upper mantle arrivals on all record sections makes this dataset particularly well suitedfor inversion.Pick uncertainties are required to allow for the appropriate data weighting duringtraveltime inversion. Uncertainties between 25—175 ms were calculated automatically basedon a ratio of energy in a short (250 ms) time window before and after the pick. Table 2.1summarizes the average pick uncertainty for each phase and for each shot. The traveltime datawere checked to ensure that traveltime reciprocity was satisfied (within pick uncertainties)between all shot-receiver pair or “near” pairs (Zelt and Forsyth 1994). Pairs of data that didnot initially satisfy reciprocity were examined and either repicked or not used. Establishingtraveltime reciprocity helps to ensure that a consistent set of data, i.e., one that can befit within its uncertainties without introducing artificial lateral heterogeneities, is used formodelling (Zelt and Forsyth 1994).The 2—D isotropic velocity structure in both routines is parameterized into layerscomposed of variably sized trapezoids (Fig. 2.18a). Each layer boundary is specified byan arbitrary number of boundary nodes, with arbitrary spacing, connected by straight line46segments. A smooth layer boundary simulation (Zelt and Smith 1992) is used to reducethe effects of ray scattering and focusing which can arise with this type of parameterization,particularly for rays incident on a boundary near a node separating two boundary segmentswith different slopes. The smooth layer boundary simulation affects only the incident andemergent ray angles where the slope of the smoothed boundary is used. Boundaries weresampled at 250 equal-spaced (1.4 km) points and smoothed with 100 passes of a three-pointaveraging filter. The smoothed boundaries are shown in Fig. 2.1 8a.The P-wave velocity field within a layer is specified by placing velocity nodes along theupper and lower boundaries (Fig. 2.1 8a). The velocities along the boundaries are determinedby linear interpolation between velocity nodes. Velocities within a layer are determined bylinearly interpolating between the upper and lower boundary velocities (Zelt and Smith 1992).Figure 2.1 8a shows the locations of all boundary and velocity nodes for the final velocitymodel. The inversion routine treats the depth of boundary nodes and velocity at the velocitynodes as parameters to be adjusted to obtain the best possible match between the observedand calculated traveltimes. Depth and velocity at all boundary and velocity nodes are notnecessarily determined by inversion. For example, boundary nodes along the uppermostboundary define the surface topography and are kept fixed. Also, velocity and depth at somenodes may be determined by forward modelling (see section 2.6).Figure 2.1 8b shows the locations of boundary and velocity nodes where boundary depthsand velocities were determined by traveltime inversion for the final model. Node locationswere determined by several inversion tests using different node spacings. The number ofnodes, and their locations in the final model, were chosen so that an optimum trade-offbetween traveltime fit and model parameter resolution was achieved while still allowing raysto be traced to all observations. In general, it is desirable to look for the simplest (i.e., leaststructure) model that replicates the observations; thus the one with the fewest parameters waschosen. The number of independent model parameters can be reduced by fixing boundaryand velocity nodes for some iterations, specifying a zero velocity discontinuity across a4700-4lecgQ-io00It)00-Ilo0.40 FR_.__Upper mantle I4 +Boundary number Layer numberFigure 2.18 (a) Location of all boundary and velocity nodes used to define the line 1 model.Dotted vertical lines illustrate trapezoidal structure of the model used by the Zelt and Smith (1992)algorithm. Grey lines connecting boundary nodes are model boundaries. Broken lines are smoothedboundaries used to calculate the incident and emergent ray angles (see text). (b) Locations of nodesat which boundary depths and velocities were determined by traveltime inversion. Boundary 1represents the topography. Layer 1 = near-surface layer; Layer 9 = crust-mantle transition layer.The upper crust includes layer 6. FR, “floating” reflector (see section 2.6.3); M, Moho.48boundary, and fixing the vertical velocity gradient within a layer (see section 2.6). Thesemethods are generally used to increase the stability of the inversion when the traveltime datado not provide strong constraints on a particular feature of the velocity structure but whichmay be inferred from amplitude modelling.A “layer stripping” approach was used to model the observed data. Typically, traveltimedata were inverted to solve for boundary depths and velocities simultaneously within oneor more layers. Starting models for these inversions were usually simple 1—D models. Therelative amplitudes of phases used in the inversion and previously modelled phases werecompared qualitatively with observed amplitudes and, where necessary, velocity gradientsand/or velocity contrasts across reflecting boundaries were adjusted by forward modelling toimprove the match. In some cases, velocities were fixed as required by the amplitude data.In other cases small adjustments to the parameters determined by traveltime inversion wererequired to remove artificial shadow zones (e.g., caused by small negative vertical velocitygradients) or correct physically unrealistic situations (e.g., crossing model boundaries). In allcases these adjustments were within the estimated absolute uncertainties of the parameters andwere not considered significant. Velocity gradients and contrasts derived from this stage of themodelling were used as constraints in a subsequent inversion of the same traveltime data. Thisprocedure of alternating traveltime inversion and amplitude forward modelling was repeateduntil a satisfactory fit of observed and calculated traveltimes and amplitudes was achieved forsuccessively deeper layers. Due to the high degree of variability in the observed amplitudes ofeach phase (not uncommon in land-based surveys) and inaccuracies of amplitudes calculatedusing ray theory a qualitative comparison between observed and calculated relative amplitudeswas considered to be a sufficiently accurate criterion for matching the fits. Also, becausethe amplitude modelling required comparison with phases generated from deeper structure,several iterations through the entire modelling procedure were required to obtain values ofvelocity gradients and velocity contrasts throughout the model which adequately satisfied theamplitude data while still allowing a satisfactory fit to the traveltime data.49Important parameters in damped least squares inversion are overall damping and a prioriestimates of the uncertainty of model parameters. The damping parameter determines theoverall trade-off between the resolution and uncertainty of model parameters (Zelt and Smith1992). Various levels of damping and a range of parameter uncertainties were tried, but forthe final model a damping factor of 1.0 (i.e., equal weight placed on improving resolutionand uncertainty of parameters) was used in all but one case which is described below. Forall iterations the estimated uncertainties in velocity and boundary depth were 0.1 km/s and1.0 km, respectively.The final model that was chosen gave the best overall fit between observed and calculatedtraveltimes and amplitudes for all phases from all shots. Given the caveat stated above,more weight was placed on achieving a good fit to the observed traveltimes. In addition, theamplitudes of some intra-crustal reflections, which were stronger than first arrival amplitudes,could not be modelled accurately using a simple first-order velocity discontinuity. Suchreflections, for example, could be generated from zones of thin (less than one wavelengththickness), alternating high and low velocity layers (e.g., see Wenzel and Sandmeier 1988).The critical factors determining the resolution of seismic data are survey geometry(receiver spacing and number of shots) and frequency content of the data. The averagereceiver spacing along line 1 is 1.2 km while the dominant frequency range of the data is2—12 Hz. Thus, the maximum vertical resolution in depth is a few hundred metres and themaximum lateral resolution is approximately 1.2—2.4 km. Details of model resolution andabsolute parameter uncertainties are given in section 2.6.5.2.6 2—D Modelling of Line 1 Refraction DataThis section describes the modelling procedure in detail and points out the degree andsource of constraints supporting each model feature. The final velocity model for line 1 isshown in Fig. 2.19; layer boundaries representing gradient changes in the upper crust areomitted. Figure 2.1 8b shows the locations of nodes where velocities and boundary depths50were determined by traveltime inversion and identifies layer and boundary numbers referredto in the text. Ray tracing diagrams illustrating the coverage provided by each shot areshown in Figs. 2.lla—2.17a. Figure 2.20 illustrates the ray coverage provided by eachphase for all shots. A comparison of observed and calculated traveltimes for each shot isshown in Figs. 2.1 lb—2.17b. Synthetic sections generated from the model in Figure 2.19are presented in Figs. 2.1 lc—2.17c and the observed record sections plotted in true relativeamplitude format are shown in Figs. 2.lld—2.17d. To compensate for anelastic attenuation,amplitudes were calculated assuming a Q of 200 in the near-surface layer, 400 in layers2—6 (upper crust) and 1000 for deeper crustal and upper mantle layers. These values areestimates based on Q values for the northwestern Basin and Range Province (Benz et al.1990), a region of similar geophysical characteristics, e.g., high average elevation, relativelythin crust and lithosphere, high heat flow (Holbrook 1990). Note that possible sources oferror in the 2—D interpretation are out-of-plane effects due to 3—D variation in velocity andreflector surfaces, and the deviation (< 12 km) in recording geometry from a straight line.Table 2.2 summarizes the results of all traveltime inversions.The final velocity model is divided into four regions: upper, middle, and lower crust, andupper mantle (see Fig. 2.18b). The uppermost boundary in Fig. 2.19 accurately representsthe topography along the line, ranging from 370—1580 m. It is included in the model toeliminate the need for elevation corrections.2.6.1 Upper crustThe upper crust for this model is defined to be that portion of the crust fully sampled byPs, Pg, R1 and R2 ray paths, including the near surface layer, and extending to a maximumdepth of 15 km (boundary 7, Fig. 2.18b).Velocities in the near surface layer (layer 1) are constrained primarily by P arrivals.Because the P branch is typically very short, the velocities in this layer can be wellconstrained only near the shot point locations. Also, the P data do not constrain a vertical51(s/wN) AIIOO13AC’.! 0 CD C’iCD r CD CD CD CDI L ! I I Io o 0có ct CI I I0Lr,C”00c’ja)C-)CCe0010InFigure 2.19 Colour contour plot of velocities beneath line 1 (vertical exaggeration = 4:1).Numbers within model represent P-wave velocities in kilometres per second at various loãations.Heavy solid lines represent locations of reflectors constrained by the traveltime data. Broken whiteline indicates maximum depth of penetration of P rays. Depth refers to depth from sealevel. Yellow triangles represent shot point locations. M, Moho; FR, “floating” reflector(see section 2.6.3). White region around model is unconstrained. Location of surfacetrace of Coldwater Fault (CWF) and reflection profile 88—11 are shown at top.z00Co0 0Co(wi) qde052DEPTH(km)50403020100 IIIC12I-c,1-CC.qpo!pu!amøsqdqoppiooatW42siuodoqpuud02iowdwoijS!1ApoSo!2JAuq2ONsuqujoiqqPW!PU!ams.npunoq(rjuoqsams.nidJaAoaz-oqsUs(iJJ(I)1djMJujdwd()E’j(p)tj()Ij(q)2jSj(v):sqdqoXqppAo1d‘aorzaIn!I18DEPTH(km)DEPTH(km)141062—2141210864208DEPTH(km)76543210—1—2—IIII-cCi)DIQ.I-U‘0IU)C;’-o-o0-C;’0002DEPTH(km)2520151050 IIIIIoI1 III..III IIII‘III—2RMS(ms)No. of No. ofvelocity depth nodesnodesPsPg 1143 50 90 5.26 29 0R1 107 82 106 2.54 0 4R2 150 98 98 1.55 4 4R3 423 98 92 1.70 5 4PEP, P 734 86 103 2.70 14 14Ptm 108 75 72 1.96 0 3velocity gradient, and thus gradients were fixed at values of ‘—‘0.5 s1 determined by themaximum observed offset of P. In Fig 2.18b, velocity nodes are not shown at the base of alayer in which the vertical velocity gradient was fixed during the inversion. Depth to the baseof the near surface layer (boundary 2) was established by forward modelling the Pg arrivalsassuming a simple structure for the boundary. Additional velocity nodes located betweenshot point locations were included to account for some of the “kinks” in Pg. Velocities atthese nodes were determined by inverting Pg traveltimes, keeping the underlying upper crustalvelocities fixed at their final values (see below). The velocities at these intervening nodes andthickness of the near surface layer between shot points are poorly constrained. In effect, it isthe delay time, resulting from both the velocity and thickness, associated with this layer thatis actually modelled and considered well constrained. Since the resolution at the interveningnodes is low the solution for layer 1 is not unique but is a model that satisfies the data.The final model for the top 4 layers comprises 29 velocity nodes and fits 1143 Ps andPg arrivals with a normalized x2 of 5.26. The large x2 indicates the data have been underfit;it was not possible to achieve X21, probably because the data contain many small scaleTable 2.2 Final velocity model inversion results for line 1. Ps and Pg, and PmPand P have been combined, since they were inverted simultaneously. tRMSis the traveltime residual between observed and predicted data.Phase(s) No. of Average Normalizedinverted observations uncertainty x2 misfit(ms)54variations caused by heterogeneities that cannot be resolved. Also, the uncertainties assignedto the data may have been optimistic given the slightly crooked recording geometry andpossible 3—D effects. These same factors apply to phases discussed below.The R1 reflection phase constrains depth to an intra-crustal reflector (boundary 6).Velocities immediately above this boundary (layer 5) could not be constrained by R1 andthus the velocities in this layer were held fixed with a zero vertical velocity gradient and witha zero velocity discontinuity across boundary 5. The ray diagram for R1 indicates that twodistinct segments of boundary 6, separated by .i100 km, are sampled by Ri. Thus, whilethe source of R1 reflectivity may be different for these two segments, a single boundary wasused for modelling convenience. Four boundary nodes were used; three near the southernsegment which is constrained by reversed rays from two shot points, and one near the northernsegment, constrained by unreversed rays from one shot point. 107 data were fit withx2=2.54R2 arrivals constrain an intra-crustal reflector defined to be the base of the upper crust(boundary 7). The data were inverted to simultaneously solve for depth and velocity in layer6. However, because the data could not resolve vertical velocity gradients and since therewas no evidence for strong gradients, the gradient in this layer was fixed at zero. Layer 6was pinched out south of 70 km for modelling convenience; there are no data in the southto constrain velocities in this layer. R1 reflection amplitudes generated from boundary 6 aresmaller (relative to Pg) than observed amplitudes (e.g., see SP 17—S; Figs. 2.14c and 2.14d),however, a better match was not possible without requiring unreasonable velocities in layer6. This likely indicates that a first-order velocity discontinuity, required by the ray tracingalgorithm, is not an appropriate model for this boundary, thus no further improvement tothe amplitude fit was sought. The R1 phase could equally well be explained as reflectionsfrom a wide fault zone or fine scale layering within the crust. The final model for layer 6and boundary 7 contains four velocity nodes and four boundary nodes. Velocities at the twocentral nodes and all depths are well constrained and fit the 150 data with 2=1.55.The final upper crustal model has 33 velocity nodes and 8 boundary nodes (Fig. 2.18b).55The four upper crustal phases (Ps, Pg, R1 and R2) are fit with2=4.65 and an overall rmstraveltime residual of 92 ms.2.6.2 Middle CrustThe velocity structure of the middle crust (layer 7) and depth to its base (boundary 8)are constrained by R3 traveltimes. The vertical velocity gradient in this layer was fixedat zero because of the inability of R3 to constrain gradients and lack of evidence, in theform of strong amplitude far-offset Pg arrivals, for strong gradients. Five velocity nodes andfour boundary nodes were used. The final model fits 423 data with 2=1.70. Ray coverageprovided by R3 is very good (Fig. 2.20d). Velocities at the four northern nodes and depths atall boundary nodes are well constrained. Synthetic R2 amplitudes, generated by the velocitycontrast between layers 6 and 7, are reasonable given the caveat of section 2.5. R2 amplitudeson SP 16 (Fig. 2.15c) are somewhat low relative to Pg.2.6.3 Lower Crust and Upper MantleVelocities at the top of the lower crust (layer 8) were determined by forward modellingthe relative amplitudes of the R3 wide-angle reflection phase. These velocities are notconsidered strongly constrained. PmP and P arrivals were simultaneously inverted todetermine velocities at the base of the lower crust, depth to Moho and velocities in theupper mantle. Tests using a first-order discontinuity for the crust-mantle boundary resultedin calculated distances to PmP critical points that were significantly less than observed forsome shots. On this basis a transitional Moho (layer 9) was included with nodal velocitiesfixed between 7.4 and 7.75 km/s determined by forward modelling of PmP amplitudes. Testsshowed that a significant thickness for the transition was required only between 70—220 km.South and north of this region the transition layer was pinched out. PmP and P were thenmodelled as reflections from the top of, and refractions beneath, the transitional Moho. Asimilar approach was taken by Zelt and Smith (1992) to model Nevada PASSCAL data.56The upper mantle vertical velocity gradient was fixed at 0.013 s_I based on forwardmodelling of P amplitudes. Synthetic P amplitudes are slightly high relative to otherphases, indicating this gradient may be too high, but lower gradients resulted in shadowzones in P coverage. The damping was increased to 10.0 for the inversion because thelower boundary of the Moho transition layer is not well constrained by the traveltimes andits depth tended to fluctuate widely if less damping was used.The final model for the lower crust and upper mantle consists of seven velocity nodes atthe base of the crust (layer 8), seven boundary nodes along the top of the Moho transitionzone, three boundary nodes defining the base of the transition zone, and seven upper mantlevelocity nodes. Velocities at the base of the upper crust are not well constrained by m•Depth to the top of the Moho transition zone is well constrained by PmP and P. Velocitiesat the five central upper mantle velocity nodes are well constrained by P, while the base ofthe Moho transition zone is poorly constrained by P. The 734 data were fit withx2=2.70.A very short (13 km) “floating” reflector within the lower crust at .s215 km (Fig.2.19) was included to model a short segment of the PmP phase from both SPs 17 and19. As described in section 2.4 both of these phases display offsets (traveltime advances)of -0.6 seconds, possibly indicating some sort of disturbance on the crust-mantle boundary.However, it was not possible to accurately model the traveltimes by including a very shortstep in boundary 9. Instead, a “floating” reflector, with no associated velocity discontinuity(and hence no signature in the synthetic sections), was used to represent this feature. Itis considered significant because it is sampled by reverse coverage from two shot points.Location of the reflector was established by forward modelling 28 data (denoted Pr) whichwere fit with X2=0.92. Zelt and Forsyth (1994) also used floating reflectors to image asequence of steeply dipping reflectors representing tectonic layering and/or thin mylonitezones beneath the southeastern Grenville Province.572.6.4 Sub-Moho Reflectorpm reflections constrain depth to an upper mantle reflector (boundary 11). Velocities inthe upper mantle were fixed as determined above. Three boundary nodes were used, fittingthe 108 data withx2=1.96. Depths to all boundary nodes are well constrained. The negativevelocity contrast across this boundary shown in Fig. 2.19 (based on amplitude modelling)is one possible interpretation. pm reflection amplitudes could be equally well modelled witha small velocity increase across this boundary.2.6.5 Model Resolution and Absolute Parameter UncertaintiesMethods for estimating both the spatial resolution of a model about a specific velocityor boundary node and absolute uncertainty of specific parameters are described by Zeltand Smith (1992). The test for spatial resolution involves perturbing the final value of amodel parameter, generating synthetic traveltime data from the perturbed model, resettingthe perturbed parameter to its final value, and inverting the synthetic data to solve for allmodel parameters that were determined at the same time as the selected parameter. Theamount of smearing of the selected parameter’s perturbation into other parameters gives anindication of the spatial resolution about the selected parameter. The results of this test canbe difficult to interpret when both velocities and boundary depths are involved.A parameter’s absolute uncertainty is estimated by perturbing its final value and holding itfixed while inverting the observed data for all other model parameters that were determinedat the same time as the selected parameter. The absolute uncertainty is indicated by thesmallest perturbation for which the resulting model is unable to fit the data as well as thefinal model, either because i) it is impossible to trace rays to all observations, or ii) the x2values of the two models differ significantly (determined by an F test). The test for absoluteparameter uncertainty is not definitive because constraints imposed by forward modellingof amplitudes are not taken into account. However, both of these methods do account forthe local non-linearity of the traveltime inversion and reveal parameter trade-offs (C.A. Zelt,58personal communication, 1992), and thus provide insight into the reliability of a model. Theresults of these tests are shown in Table 2.3.Table 2.3 Estimated lateral resolution of the line 1 velocity model about specified nodes andabsolute uncertainty of velocities and depths. Resolution and uncertainty values are basedon tests applied to representative nodes. Tests were not applied to all nodes.Estimated lateral Estimated absoluteNodes resolution (km) uncertaintyVelocities in upper crust (layers 2-4) 50 - 70 0.15- 0.2 km/sBoundary 6 nodes 25 - 75 1.5 - 2.0 kmVelocities in upper crust (layer 6) 70 0.25 km/sBoundary 7 nodes 40 2.5 kmVelcocities in middle crust (layer 7) 50 0.2 km/sBoundary 8 nodes 40 3.5 kmVelocities at base of lower crust (layer 8) 30 0.15 km/sBoundary 9 nodes 60 3.0 kmBoundary 10 nodes 60 3.0 kmVelocities in upper mantle (layer 10) 25 0.2 km/sBoundary 11 nodes 45 1.5- 3.0 km2.7 Principal Features of the Velocity ModelThe principal features of the line 1 velocity model, as constrained by traveltime inversionand amplitude forward modelling, are shown in Fig. 2.19. There is a thin (<2.5 km) nearsurface layer of average velocity 4.5 km/s which thins to the north so that beneath SP 1 it is350 m thick. Velocities in this layer vary between 2.8 and 5.4 km/s with several localizedlow velocity regions that correlate well with delays seen in Pg. A comparison of surfacevelocities with a surface geology map (Wheeler and McFeely 1991) shows good correlationbetween these low velocity (<4.8 km/s) surface regions and several Tertiary sedimentary and59volcanic bodies and Eocene through Late Triassic plutons of unknown thickness traversedby the line (Fig. 2.1). The higher velocity regions of the near-surface layer are mainlyCretaceous through Paleozoic volcanics.The upper crust (layers 2—6) extends to a depth of 15 km, thinning north of SP 16 to10 km beneath SP 19. At the top velocities vary laterally between 5.95 and 6.4 km/s.Velocities are highest at the north and south ends of the model. Slower velocities occur justnorth of SP 2, south of SP 16, and beneath SP 19 (Fig. 2.19). Vertical velocity gradientsthroughout this region are low (-‘0.01 s_1). Two regions of relatively high gradient, inferredfrom amplitude modelling, occur at the top beneath SP 18 (0.07 s1) and between SPs 17and 16 (‘-‘0.05 s1).The R1 reflecting horizon within the upper crust outlines a transition into lower velocities(layer 6; 5.8—6.3 kmls) which continue to the base of the upper crust north of 70 km. The topof the low velocity zone is at 9 km north of SP 17, dipping to 15 km beneath SP 18 to thesouth where it is pinched out with the base of the upper crust. The average velocity contrastacross boundary 6 is —0.1 km/s, substantially less than the estimated absolute uncertaintiesfor velocities in the upper crust (Table 2.3).The middle-upper crust boundary is outlined by the R2 reflecting horizon. Velocities inthe middle crust are similar to those immediately above the upper crustal low velocity zone(6.05—6.55 km/s). Velocity is highest at the south end, resulting in a very small contrastacross the middle-upper crust boundary south of SP 18. This is consistent with a lack ofobserved R1 and R2 arrivals with reflection points at the south end (Fig. 2.20b and c).Beneath the low velocity zone the average velocity contrast is +0.1 km/s. The base of themiddle crust, outlined by the R3 reflecting horizon is at an average depth of ‘-.25 km southof SP 16, shallowing to 18 km beneath SP 19.The lower crust is ‘-4 km thick, increasing to 14 km at the north end. Velocities rangefrom 6.25—6.6 km/s at the top and 6.55—6.85 km/s at the bottom, except beneath SP 19 where60the high apparent velocity PmP reflection from SP 1 constrains a region of higher (7.15 km/s)velocity. A short “floating” reflector at 29 km depth (in the lower crust) is constrained byPmP arrivals from two shot points. It may represent a very short offset in the Moho, toosmall to be modelled by other means.The top of the Moho transition layer is at an average depth of 32 km depth but shallowsto 30 km beneath SP 2. This Moho depth is consistent with an earlier study of a partiallyreversed refraction profile (Cumming et al. 1979; see Fig. 2.2 for location). The transitionlayer is <3.5 km thick. Its base is nearly flat, varying between 32.5—34.5 km. The thickness ofthe transition layer is not well constrained, but it is included to provide an improved amplitudefit to PmP and because, if this zone is interpreted to represent a complex interlayered sequenceof high and low velocity material, it probably more realistically represents the true nature ofthe crust-mantle boundary (Mooney and Brocher 1987). Velocities at the top of the uppermantle average 7.85 km/s, ranging from a low of 7.6 km/s between SPs 16 and 19, to 7.95km/s beneath SPs 18 and 16. An upper mantle reflector is imaged at 47—51.5 km depth, orapproximately 16 km beneath the Moho and may represent the base of the lithosphere whichis known to be thin throughout this region (Wickens 1977). The velocity contrast across thisreflector, shown as negative in Fig. 2.19, is not well constrained.2.8 Comparison With Lithoprobe Reflection DataIn 1988, Lithoprobe recorded reflection profiles within the southern IntermontaneBelt; Cook et al. (1992) provide interpretations. Profile 88—11 crosses line 1 obliquelyapproximately midway between SPs 2 and 17 (Fig. 2.1). Figures 2.21a and c show thedata for Profile 88—11 and a simplified schematic interpretation of the data in the vicinityof line 1, with time-to-depth conversions based on the refraction model velocities at thelocation of Profile 88—11 (Fig. 2.21b). These lines are not coincident. Further, allowing forpossible crustal anisotropy and lateral averaging in the refraction velocity model, a precisecorrespondence between boundaries at the intersection of the two lines is not expected.61Figure 2.21 (a) Migrated and coherency filtered reflection data from Lithoprobe reflection profile88—il (Cook et al. 1992). CWF, surface trace of west-dipping Coidwater Fault; CNH, CentralNicola horst. Broken lines show region covered in (c). (b) 1—D velocity profile from the refractionmodel in Fig. 2.19 at the location of profile 88—il. (c) Schematic interpretation (no verticalexaggeration) of profile 88—11 in the vicinity of line 1 (modified from Fig. 7 of Cook et al.1992), with time to depth conversion based on refraction velocities in (b). Broken lineswithin Quesnellia represent reflections. The uppermost reflection is interpreted as theColdwater Fault by Cook et al. (1992). Grey lines between (b) and (c) show possiblecorrelations between refraction and reflection interpretations discussed in text.03CD-InIv Cl)15I 50km010-c0.a)304050-Parautochthonous N.A.Cratonic N.A.Upper Mantle62The crustal section of Cook et al. (1992) is interpreted in terms of three major components(Fig. 2.21c): (i) an upper layer ‘-‘ 23 km thick comprising primarily material of thePaleozoic—Mesozoic Quesnellia terrane, possibly thickened by imbrication and (or) ductiledeformation due to convergence of the terrane with North America; (ii) a middle layer 4km thick which projects updip eastward to the Vernon antiform (140 km east of line 1) andcan be probably correlated with the parautochthonous rocks which form a thick sequence ofthe upper crust in the eastern Cordillera; and (iii) an -6 km thick lower crust representing thewestwardly thinned extension of North American cratonic basement. The crust of Quesnelliashows a high degree of reflectivity, including several prominent west-dipping reflections(dashed lines in Fig. 2.21c). These features are related to Mesozoic or earliest Tertiarycompressional features within the Eocene Central Nicola horst (CNH on Fig. 2.21a). Thehorst is bounded on the west by the west-dipping normal Coidwater Fault (see Fig. 2.1 forsurface location).The refraction data image three crustal reflectors in the vicinity of Profile 88—11. Thefirst (R1) is modelled as the top of a weak LVZ at a depth of —11 km, and correlateswith the west-northwest dipping Coidwater Fault which projects to a depth of about 9—10km (r3.5 s two-way traveltime (TWTT) in Fig. 2.21a) beneath line 1. The origin of thereflectivity cannot be constrained but may be primary lithologic variation (e.g., see Hunchet al. 1985), as the Nicola volcanics are separated by the Coldwater Fault from underlyingNicola batholith and associated metamorphics, or anisotropic fabrics (e.g., see Jones and Nur1984) within the fault zone. The identification of the R1 reflector as the Coidwater Fault isattractive; however, this correlation is unlikely because of geometrical considerations. Thegeometry of the R1 reflector, imaged by the refraction data as dipping to the south from SP17, is inconsistent with the geometry of the fault surface, which crosses the line ‘—‘15 kmsouth of SP 2 (at 100 km). Thus, given the location of the R1 reflection points from SP17 (Fig. 2.20b), it is not likely that this phase is reflecting from the Coidwater Fault. R1reflections from SP 18 could be from the fault; however, this would require energy from this63phase to first cross the fault surface before reflecting from it. The different character of R1reflections from SPs 17 and 18 (R1 from SP 17 is stronger and more coherent) is consistentwith separate sources of reflectivity for these two R1 phases.The second refraction-imaged crustal reflector (R2), at 15 km depth correlates wellwith the top of a comparatively flat sequence of reflections beginning at —‘5.8 s TWT’l’ (16km) on the reflection section. This is one of several reflections in Quesnellia that form anantiformal pattern relating to the Central Nicola horst. Given the high reflectivity of thecrust as seen in Fig. 2.21a and the range of possible interpretations as to the positions ofthese reflectors, such a correlation may be coincidental. Alternatively, R2 could represent atransition from brittle to more ductile crust or a transition into a layered porous medium. Thebrittle-ductile transition occurs at a temperature of about 4500 C for mafic rocks which, fortypical heat flows in the Intermontane Belt, occurs at a depth of about 14 km (Lewis et al.1992). The brittle-ductile transition frequently correlates with the top of seismically reflectivezones in the deep crust. For example, Holbrook et al. (1991) compare seismic refractionand reflection interpretations from the northern Basin and Range of Nevada to determinethe origin of reflectivity from the deep crust. They interpret the upper limit of a stronglyreflective zone to be controlled by a transition from brittle to ductile behavior. Hyndman etal. (1991) argue that a correlation between the 450° C isotherm and the top of reflective zonescan be explained by a transition into layered fluid porosity, presumably accommodated by atransition to greenschist-amphibolite facies which is compatible with free aqueous porosity.The data of Cook et al. (1992), however, show strong reflectivity throughout the entire crustin the vicinity of line 1 (Fig. 2.21a). It is not possible to uniquely determine the sourceof the R2 reflection. However, the correlation between R2 and reflections associated witholder compressional structures present in the Eocene Central Nicola horst suggests that, atleast locally (i.e., where the refraction and reflection lines cross), ductile fabrics are the mostlikely source.The base of the middle crust of the refraction model (--‘23 km) correlates well with the64base of the upper crust of the reflection interpretation (22 km). This is interpreted by Cooket al. (1992) as the boundary between rocks of the Paleozoic—Mesozoic Quesnellia terrane(vp6. 15 km/s) and parautochthonous North American rocks (vpi’s6.46.6 km/s) which canbe traced down dip from surface exposures to the east. The strong R3 reflections generatedat this boundary may arise from a ductile shear zone which has accommodated movement ofQuesnellia during periods of compression and extension, or possibly from primary lithologiclayering due to igneous layering or mafic sill-like intrusions (e.g., see Holbrook et al. 1991).The depth to the R3 reflector also correlates well with the predicted depth to the 730° Cisotherm which, from heat flow data in the Intermontane Belt, occurs at about 23 km (Lewiset al. 1992). This isotherm sometimes corresponds to the lower limit of crustal reflectivity,although again the data of Cook et al. (1992) show strong reflectivity throughout the entirecrust (Fig. 2.21a). The 730° C isotherm corresponds to the top of granulite facies conditions,below which the crust would be expected to be largely dehydrated (Hyndman et al. 1991). Iffluids do play a major role in the crustal reflectivity of this region, then the middle crust of therefraction model may represent a region in which interconnected fluids exist and electricalconductivity is relatively high.A laterally coherent event in the reflection data at s TWTT (26 km; Fig. 2.21a),interpreted as a zone of detachment at the top of North American cratonic crust and alongwhich the overlying crust has been thrust eastward (Cook et al. 1992), has no identifiablesignature in the refraction data. This observation may have some bearing on the nature ofthis reflective zone since it may not be associated with a significant velocity contrast. Thelowest crustal unit of the reflection interpretation represents a thin and highly strained layerof North American crust (Cook et al. 1992). Velocities at these depths beneath profile 88—11range from 6.6—6.8 km/s (Fig. 2.21b), and are: (i) significantly less than typical lower crustalvelocities for the western craton i.e., -..7.2 km/s (Chandra and Cumming 1972); (ii) consistentwith velocities in the Omineca belt beneath line 8 (Kanasewich et al. 1994) and the westernend of line 9 (Zelt and White 1994) of SCoRE ‘90 (Fig. 1.la and c); and (iii) significantly65greater than velocities in the Foreland belt beneath the eastern end of line 9 (6.3—6.5 km/s;Zelt and White 1994). Beneath line 1, velocities near the base of the lower crust increaseto 7.15 km/s at the north end of the line. If the bottom half of the lower crust beneathline 1 does represent (modified) North American basement, then this unit shows an extremerange in velocity across the Cordillera (6.3—7. 15 km/s). This may be explained by havingdifferent levels (i.e., upper, middle and/or lower crust) of the original Precambrian cratonpresent in different locations with high velocities corresponding to deeper units. This modelwould require the removal of different levels of North American crust at different locationsand by different mechanisms, possibly by erosion or crustal delamination. Alternatively, thevariations in lower crustal velocity may be associated with different basement domains ofdiffering composition and tectonic history as suggested by Ross and Parrish (1991) for theOmineca Belt.Depth to the top of the Moho transition layer (31 km) agrees well with the depth to thereflection Moho (32 km) which has been interpreted as the base of the horizontal zone ofreflections at about 10.5 s TWTT in Fig. 2.21a. The duration of this reflective zone (—‘0.2s or -s 1 km) is thinner than the crust-mantle transition zone of the refraction model whichrepresents a zone of alternating high velocity (mafic) and low velocity rocks. The thicknessof this zone, however, is poorly constrained.2.9 SummaryAn iterative combination of traveltime inversion and amplitude forward modelling ofseismic refraction data has been used to determine along-strike variations in crust and uppermantle velocity structure within the Intermontane Belt. This procedure has proven efficient atproviding a well constrained velocity model with estimates of lateral resolution and absoluteuncertainty of model parameters. The near surface layer is thin and velocity variations inthis layer correlate well with surface geology. The average velocity gradient in the upperand middle crustal layers is low, although there are two regions of comparatively strong66gradient at the top of the upper crust. Velocities in the upper and middle crust are typically6.2 — 6.3 km/s with depth to the base of the middle crust at r25 km. There is a significant(0.2 km/s) velocity contrast at the middle-lower crust boundary. Velocities in the lowercrust average 6.4—6.75 km/s. Crustal thickness along this line averages 33 km, which isrelatively thin for a region of high average elevation (970 m). This likely relates to highcrustal temperatures which are confirmed by heat flow studies in the Cordillera (Lewis etal. 1992). The crust-mantle transition layer, with a maximum thickness of 3.5 km. is not awell constrained feature of the model but is inferred from consideration of PmP amplitudes.Velocities at the top of the upper mantle are low, varying between 7.6—7.95 km/s, consistentwith the suggestion of partial melting of the upper mantle below the Intermontane Belt dueto (recent) mantle upflow in this region (Gough 1986).An upper mantle reflector, approximately 16 km beneath the Moho, may represent thebase of the lithosphere which is considered to be thin throughout much of the IntermontaneBelt. This depth agrees well with the depth to the top of an upper mantle low velocityzone, approximately 15—20 km beneath the Moho, found by Wickens (1977) in his study ofRayleigh and Love wave phase velocities.There is good agreement between the refraction velocity model and an interpretation ofLithoprobe reflection profile 88—11 (Cook et al. 1992). Where the two lines intersect, depthsto wide-angle reflectors in the crust of the refraction model agree with depths to prominentreflection events. Depths to Moho agree to within 1 km.Strong reflections are generated from the middle-lower crust boundary and Moho. Weakerreflections are generated at the upper-middle crust boundary and from within the uppercrust. The origin of these reflections cannot be determined uniquely from the refractiondata. Likely sources include fault zones, ductile strain banding, trapped fluids or primarylithologic layering such as mafic sills (Klemperer and BIRPS Group 1987). Depths to intracrustal reflectors at 15 and 25 km correlate well with both reflection-imaged reflectors andpredicted depths to the 450° C and 730° C isotherms, a temperature range which is compatible67with free aqueous porosity (Hyndman et al. 1991). The middle-lower crust boundary at—‘24 km may represent a transition into a region of greater ductility which accommodatedmovement of Quesnellia along listric detachment faults during periods of Eocene extensionalactivity. Rocks immediately below this boundary have been correlated with parautochthonousNorth American rocks (Cook et al. 1992). There is no evidence in the refraction data forlower crustal type velocities above this zone and thus it appears that during accretion, thelower lithosphere of Quesnellia was removed, leaving this terrane “rootless”. In this senseQuesnellia is similar to the northern part of the Chugach terrane in southern Alaska which hasbeen interpreted as a wedge which overrode the deeper part of a neighboring terrane, losing itsroot in the process (Fuis and Plafker 1991). Fuis and Mooney (1990) interpret an analogoustectonic history for the Franciscan terrane in central California. The fate of Quesnellia’slower lithosphere is unknown, although presumably it was removed by subduction erosionand/or crustal delamination and recycled in the mantle.There are lateral variations in velocity and depth to reflecting boundaries but there isno evidence for rapid changes in velocity over short lateral distances (except in the nearsurface layer) or dips greater than .--‘5°. This may relate to the along-strike orientation ofthe refraction line which prevents the imaging of large-scale convergent and extensionalstructures associated with the development of the crust of this region of the Cordillera.Kanasewich et al. (1994) present an interpretation of refraction line 8 (Fig. 1.1 a and c)recorded along-strike in the Omineca Belt approximately parallel and 180 km east of line1 as part of SCoRE ‘90. Their interpretation also includes wide-angle reflections that canbe associated with a major decollement identified by surface geology and coincident seismicreflection data. An inversion of traveltimes indicates a substantive mid-crustal low velocityzone of varying thickness between depths of 12 and 25 km, in contrast to the weak LVZ (orperhaps very low vertical velocity gradient) inferred below the Intermontane Belt at depthsfrom 9 to 15 km. Both datasets are characterized by strong PmP reflections, indicative of aprominent velocity-density contrast at the crust-mantle transition. Depth to the top of a crust68mantle transition zone along line 8 averages about 36 km. about 4 km deeper than belowline 1, a result consistent with the earlier results of Cumming et al. (1979) and with recentreflection data (Cook et al. 1992), indicating crustal thickening to the east. With values of7.9 to 8.1 km/s. upper mantle velocities below the Omineca Belt appear to be somewhathigher than below the Intermontane Belt, somewhat surprising given that the location of anupcurrent of mantle convection may presently be located beneath the Omineca Belt (Gough1986). Kanasewich et al. (1994) identify and interpret sub-Moho wide-angle reflections froma boundary at an average depth of 48 km, very similar to the 47—52 km estimate for line1. If these sub-Moho boundaries do represent the top of the asthenosphere, then there maybe little change in total lithospheric thickness between the Omineca and Intermontane Belts.The velocity structures of the Intermontane and Omineca Belts show some similar generalcharacteristics but are clearly different in detail, presumably reflecting the different originsof the rocks forming the crust and the processes associated with their formation.O’Leary et al. (1993) present an interpretation of SCoRE ‘90 line 10 (Fig. 1.la and c)recorded along-strike in the Coast Belt, subparallel and roughly 100—250 km west of line1. Their model shows a similar three-layer division of the crust with a 10 km thick uppercrust, a middle crust extending to 23 km depth, and a lower crust extending to the Mohowith an average depth of 35 km. All boundaries are relatively flat in comparison with line1. Average upper and middle crustal velocities are very similar to those beneath line 1. Thelower crust beneath line 10 is divided into two layers on the basis of a wide-angle reflectionfrom what is interpreted as a thrust fault linked to the collision of the Insular superterrane andNorth America by J.M. Journeay (in Monger and Journey 1992). Otherwise, average lowercrustal velocities are similar to velocities beneath the southern half of line 1. O’Leary et al.(1993) also include a transitional Moho layer, approximately 2 km thick, but possibly up to3.5 km thick. Their average Moho depth of 35 km is only slightly greater than beneath line1 (32.5—34.5 km), but does suggest a slight thickening of the crust to the west beneath theCoast Belt. Upper mantle velocity beneath line 10 is modelled as 8.15 km/s, but is poorly69constrained. This suggests significantly higher velocity beneath the Coast Belt in comparisonto the Intermontane Belt. An upper mantle reflector at a depth of 70 km beneath line 10could represent the base of the lithosphere and, if so, may indicate a westward thickeningfrom values of —‘50 km beneath both the Intermontane and Omineca Belts.703 2-D INTERPRETATION OF LINE 3: COAST BELT3.1 IntroductionThis chapter presents a 2—D velocity model for line 3, recorded obliquely across-strikefrom the eastern Insular Belt through the southernmost Coast Belt to the western edge ofthe Intermontane Belt (Fig. 1.2). A brief discussion of the geology, tectonic history, andgeophysics of the study region is followed by a presentation and description of the refractiondataset for line 3. Details of the specific modelling procedure for line 3 are presentednext, followed by a presentation of the velocity model. This chapter concludes with aninterpretation of the model, a discussion and summary.Zelt et al. (1993) include the principal scientific information in this chapter. Mycontributions to this paper were the same as described in section 2.1 for Zelt et al. (1992).3.2 Tectonic History and Geology of the Coast BeltThe Coast Belt straddles the boundary between the Insular and Intermontane superterranes(Fig. 1.1) and can be subdivided into two distinct tectonic units based on its premid-Cretaceous (pre-collision) configuration (Crickmay 1930; J.W.H. Monger, personalcommunication, 1994). The Western Coast Belt (WCB; Fig. 3.1) is bounded on the west bythe Insular-Coast Belt boundary in the Strait of Georgia, and on the east by the CentralCoast Belt detachment (CCBD; Fig. 3.1), a steeply dipping reverse fault. The WCBcomprises Wrangellia (Insular superterrane), Early Cretaceous granitic rocks and, in thesouth, the Harrison terrane, a former island arc that was linked to Wrangellia by the MiddleJurassic (Monger and Journeay 1992). The Eastern Coast Belt comprises the Bridge River,Cadwallader and Methow terranes, which represent marine clastic basins built on oceanicbasement (Monger et al. 1994b). This region was the focus of Late Cretaceous—earlyTertiary deformation and granitic intrusion. Monger et al. (1994b) propose that the basinswere inter-arc, existing before the accretion of the Insular superterrane in the mid-Cretaceous.Rocks in the Eastern Cascades (EC; Fig. 3.1) are correlative across the Fraser River-Straight71COAST BELTFigure 3.1 Line 3 location map showing shot point (A, with numbers) and receiver locations (+)for line 3. Other SCoRE ‘89 lines are shown in grey. Nearby reflection and refraction surveysmentioned in text are indicated by heavy broken lines. Major geologic features and tectonicelements are: BKFS, Bralorne-Kwoiek Creek Fault System; BR, Bridge River terrane; CBTS, CoastBelt Thrust System; CCBD, Central Coast Belt detachment; CMC, Cascade metamorphic core; EC,Eastern Cascades; ECB, Eastern Coast Belt; FRF, Fraser River Fault; FR-SCF, Fraser River-StraightCreek Fault; GB, Georgia Basin (stippled region); HF, Harrison Fault; EL, Harrison Lake; HOF,Hozameen Fault; lB. Insular Belt; IIB, approximate location of Insular-Intermontane boundary atdepth as inferred from reflection data (Varsek et al. 1993); 1MB, Intermontane Belt; LI, LasquetiIsland; MT, Methow terrane; NWCS, Northwest Cascades System; OLF, Owl Lake Fault; PF,Pasayten Fault; RLF, Ross Lake Fault; TLF, Thomas Lake Fault; VF, Vedder Fault; WCB, WesternCoast Belt; WR, Wrangellia. Geology from Monger et al. (1994a) and Varsek et al. (1993).Creek Fault with those in the Eastern Coast Belt. Recent (40—0 Ma) activity in the CoastBelt has been dominated by Cascade arc magmatism related to the subduction of the Juan deFuca plate. Other products of recent tectonism include uplift within the last 10 Ma (Parrish1983), and the development of northeast-trending Neogene (possibly extensional) faults (e.g.,Vedder Fault; Fig. 3.1) (Monger and Journeay, 1994).Several models for the tectonic evolution of the Coast Belt exist (Monger et al. 1982;Gehrels and Saleeby 1985; van der Heyden 1992; Monger et al. 1994b). In what has perhaps125’W 124’W 123’W 122GW 121’W 120’W72been the most widely accepted model (Monger et al. 1982), the Coast Belt is viewed in partas a metamorphic and structural welt which resulted from the mid-Cretaceous collision ofthe Insular superterrane with the western margin of North America, the previously attachedIntermontane superterrane. The collision is associated with the closure of an interveningocean basin and led to the development of a collisional suture which was later overprintedby magmatic activity in the present Coast Belt, presumably arising from a new east-dippingsubduction zone beneath the Insular superterrane (van der Heyden 1992). The Cretaceoussuture between the two superterranes in this model is the Central Coast Belt detachmentwhich occurs in the eastern part of the Coast Belt Thrust System (CBTS; Fig. 3.1). TheCBTS is a west vergent contractional belt that formed along the eastern margin of theInsular superterrane in early Late Cretaceous time (97—9 1 Ma; Journeay and Friedman 1993),containing elements of the southwestern Coast Belt in the lower part of the thrust stack andelements of the southeastern Coast Belt in its upper part.van der Heyden (1992) has proposed an Andean arc model for the Coast Belt involvingthe accretion of a single superterrane, comprising previously amalgamated terranes of theInsular superterrane and Stikinia, to ancestral North America in the Middle Jurassic. TheCoast Belt is viewed as a complex succession of magmatic arcs related to the prolongedsubduction of oceanic lithosphere beneath the new western margin. In this model the oceanicBridge River and Cache Creek terranes represent the suture between Stikinia and EarlyJurassic North America.More recently, Monger et al (1994b) have proposed a “compromise” model in whichthe Jura-Cretaceous arc and accretionary complex was transected acutely and offset along asinistral strike-slip fault system in the Early Cretaceous (120—100 Ma), trapping the inter-arcbasins of the southeastern Coast Belt to the east.Major thrust sheets and bounding faults of the Coast Belt Thrust System extend southwardinto Washington and provide a structural link between thrust systems of the northwestCascades (Fig. 3.1). The southeastern Coast Mountains between Harrison Lake and the73Fraser River Fault share several partly coeval, though lithologically distinct terranes withthe northwest Cascades (Monger 1991). In the line 3 study area, these are the Harrisonterrane, which occurs both west and east of Harrison Lake (Monger 1991), Shuksan, BridgeRiver and Methow terranes. Cascade structures wrap around the east side of Harrison Lakeand extend to the northwest (Journeay and Csontos 1989). East of the Fraser River-StraightCreek Fault system, and bounded on the east by the Ross Lake Fault, lies a narrow, northwardextension of the Cascade metamorphic core (Fig. 3.1) comprised of high-grade metamorphicand Late Cretaceous-Early Tertiary granitic rocks. The Cascade metamorphic core probablydiffers from the Northwest Cascades mainly in its higher metamorphic grade and abundanceof granitic rock. High grade metamorphic rocks in the central Coast Belt correlate withrocks in the Cascade metamorphic core but have been offset by 80—190 km of late Eocenestrike-slip motion along the Fraser River-Straight Creek Fault (Misch 1977; Monger 1985;Coleman and Parrish 1991; Friedman and van der Heyden 1992). This displacement hasalso offset terranes in the Eastern Coast Belt.3.3 Previous Geophysical WorkRefraction data recorded in the southernmost Cordillera prior to 1971 (Berry and Forsyth1975) (Fig. 2.2) employed large receiver spacings (13 km) and shots at line ends only (i.e.,separated by 1500 km). Their model for the Coast Belt consisted of a crust increasing inthickness from 22 km in the west to 28 km in the east with velocity 6.4 km/s. Upper mantlevelocity was 7.8 km/s. They inferred a significantly different structure for the Insular Beltwhich included a high velocity 7.1 km/s lower crustal layer between 30 and 43 km depth.More recently, seismic refraction data recorded along Vancouver Island (VISP 80 line IV:McMechan and Spence 1983; Drew and Clowes 1990) and along a profile extending acrossVancouver Island from the volcanic arc (VISP 80 line I: Spence et al. 1985; Drew andClowes 1990) (see Fig. 2.2 for locations) have proven more effective in imaging structurewithin the Insular and westernmost Coast Belt, in particular that of the subducting Juan de74Fuca plate. Models for line I, however, are well-constrained to a depth of only 18 km forthe WCB although they do include a Moho at 38 km.Results from the Pacific Northwest (PACNW) 1991 refraction survey for a profileimmediately to the south of (and perpendicular to) line 3 (profile 1; Fig. 3.1) indicate acrustal thickness of 38 km for the Northwest Cascades System (Gridley 1993). Details ofthis model will be discussed later.Seismic reflection data recorded across the southern Coast mountains about 130 kmnorth of line 3 (Varsek et al. 1993) are interpreted to show the wedging of the Intermontanesuperterrane into the middle and lower crust of the Insular superterrane in the vicinity of theCoast Belt Thrust System, just east of the Owl Lake Fault (see Fig. 3.1 for location).As an additional component of the Lithoprobe Southern Cordillera transect, extensivemagnetotelluric (MT) measurements have been made along lines approximately coincidentwith the reflection profiles. Initial interpretations concentrated on prominent resistivitycontrasts across the Fraser River Fault (Jones et al. 1992a) and an overview 2-Dcharacterization of crustal resistivities across the southern Cordillera at a latitude of about50° (Jones et a!. 1 992b). Interpretation of the MT data indicates that the Fraser River Faultpenetrates the entire crust; the upper crust to the west of the fault is more resistive than tothe east (Jones et a!. 1992a); and the lower crust (below is23 km) is less resistive to the westthan to the east (Jones et al. 1992b). Using part of the Lithoprobe data and additional dataacquired up to 250 km further north and northwest, Gough and Majorowicz (1992) interprethigher crustal resistivities beneath the eastern edge of the Coast Belt and Intermontane Beltto arise from less conductive plutonic granodiorites.3.4 Seismic Refraction DataRecording parameters for line 3 are given in section 1.5. Shot points 9 and 10 wererecorded by receivers between SPs 8 and 10 only and thus constrain only the uppermostcrustal structure at the west end of the line. SPs 21 and 20 were recorded by receivers75between SPs 10 and 3 only. The resulting dataset for line 3 comprises approximately 1250traces, although ‘150 of these, particularly at offsets greater than 250 km for SP 8 andgreater than 100 km for SP 20-W, were too noisy to allow identification of seismic arrivals.Approximately 1250 picks of first and later arrival traveltimes were obtained and used in theanalysis. Each pick was assigned an uncertainty using the method described in section 2.5.The seven record sections are displayed in trace normalized, unfiltered, reduced traveltimeformat in Figs. 3.2—3.7. A total of seven distinct phases, illustrated in Fig. 3.8 are identifiedand used in the modelling procedure (also see Figs. 3.2—3.7). P, Pg, P and P refer torefracted energy which has turned through the near surface layer, upper and lower crust,and upper mantle, respectively. R1, R2 and PmP are reflections from the base of the upper,middle and lower crust (Moho), respectively. Table 3.1 summarizes the number of traveltimepicks and average pick uncertainty of each phase for each shot.First arrivals on the first few near-offset traces from most shots are refractions througha thin near surface layer (Ps). The apparent velocity of this phase varies between 4.9 and5.2 kmls, except beneath SPs 10 and 3 where the apparent velocity is closer to 5.8 km/s.Although P is missing on SP 9 because there are no traces within 10 km of the shot, thehighest that Ps could be here is 4.6 km/s as constrained by the arrival times of Pg. Inaddition, Pg arrivals on the first few traces in both directions from this shot are delayedrelative to farther-offset Pg arrivals, indicating that the near surface velocity within about15 km of the shot point is considerably less than surrounding values. The P branch on SP3 (Fig. 3.7) is quite long (40 km) indicating a thick near surface layer to the west of theshot, but 50 km to the west, beneath SP 20 (Fig. 3.6), P is very short suggesting that thenear surface layer is thin there.Pg appears as the first arrival from the Ps-Pg crossover point to offsets of about 165 km.A few short “kinks” representing traveltime delays and advances over distance scales of afew to ten traces are obvious along this phase. These usually occur consistently at the samelocation from section to section indicating they are most likely caused by near surface high76C..tD00—.C..-to—...o()0 -t C,,0o-t00s—GOEc,-.- -t()o—Z00<-0 — 0NCl) 00.Cl)Cl)C..wLine3Shot8EU2050100150200250300Distance(km)o 20 40 60 80 100 120Distance (km)Figure 3.3 Trace normalized record sections for (a) SP 9 and (b) SP 10into line 3. See Fig. 3.2 caption for other information.w Line 3 Shot 9 E54-3-2-1—0PgT11j1UIPgCl)ccEU]ccEL543210I I I I — — I TT j1II0 20 40 60 80 100 120w Line 3 Shot 10 E:uI iN=111 HILHW 7 Ri I— HHUL PgT- -I I I I I I I I“A78wLine3Shot4E0 *.* h-’.(1 00C) 0 C,,2CD C,,CD CD8 7 6C125 4 3 1 0050100150200250300Distance(km)80100120140160180200220240260280300320340Distance(km)wLine3Shot21E(Do*CD0CI)00 C8 7 6 5 4 3 2 1 0wLine3Shot200 CD()—.oI-’ C,,C 0E00Cl] E8 7 6 5 4 3 2 1 0100120140160180200220240260280300320340(km)DistanceO)oN—.(D 0 0 -S C’,0 — C’,wLine3Shot3E00Ci210 8 6 4 2 0050100150200250300Distance(km)dO-4 ‘-114rxQOcy)00 50 100DISTANCE (km)150 200Figure 3.8 Representative ray paths for phases modelled: P, Pg. P and P are rays refractedthrough the near surface layer, upper crust, lower crust, and upper mantle, respectively. R1, R2, andPmP are reflections from the base of the upper, middle, and lower crust, respectively.0250 3000c\2and low velocity anomalies. For example, a delay of Pg between 25—40 km from the westend of the line on SP 8 (Fig. 3.2) is consistent with delays seen on SPs 9-W, 10-W and 4-W(Figs. 3.3 and 3.4) at the same location and correlates with the near surface low velocityzone surrounding SP 9 mentioned above. This example can also be correlated with a majorgeological feature; the delayed traces correspond to receiver sites situated in the GeorgiaBasin, which contains Upper Cretaceous through Miocene clastic rocks (Fig. 3.1). Otherkinks in Pg do not correlate as well with surface geology.An increase in the apparent velocity of the Pg branch from values in the 6.0—6.3 km/srange to values in the 6.35—6.45 km/s range on most sections is indicative of layering withinthe upper crust. These breakovers occur at offsets varying between 65—125 km indicatingsignificant lateral variations in the structure.P, with an apparent velocity of 7.1—7.2 km/s, appears as the first arrival at offsets beyond165 km on SPs 8 and 4-W and between 115—150 km on SP 3. This phase provides evidence83Table 3.1 Number of observations and average uncertainty in milliseconds (parentheses) for eachphase and each shot along line 3. Totala is the total number of observations and averageuncertainty in milliseconds (parentheses) for each shot. Total1’ is the total number ofobservations and average uncertainty in milliseconds (parentheses) for each phase.Shot point P and P8 R1 R2 Pc PmP P Totala8-E 84 (43) 12 (53) 15 (49) 44 (70) 74 (124) 229 (75)9-W 18 (25) 18 (25)19-E 40 (34) 12(100) 52 (49)10-W 44 (45) 21(75) 65 (54)l0-E 14 (26) 14 (26)4-W 103 (37) 17 (59) 14 (63) 15 (108) 149 (49)4-E 109 (49) 13 (66) 20 (110) 21 (104) 163 (65)21-W 108 (48) 20 (154) 128 (64)21.E 58 (54) 58 (54)20-W 54 (38) 54 (38)20-E 40 (38) 40 938)3-W 115 (46) 19 (52) 19 (62) 104 (111) 28 (108) 285 (77)Total” 787 (43) 75 (70) 54 (73) 77 (118) 234 (118) 28 (108) 1255 (63)for a high velocity lower crustal layer. Two intracrustal reflected phases are observed andmodelled. R1 reflections were recorded from SPs 8, 9-E, 10-W and 4 and typically appearclose (<300 ms) behind Pg at offsets in the 45—100 km range. R2 is a wide-angle reflectionobserved on SPs 8, 4-E and 3 at offsets in the 115—175 km range and is typically about600 ms or less behind Pg.Upper mantle refracted arrivals, Pn, are observed on SP 3 between 20—110 km at 7.1 swith an apparent velocity of 7.9 km/s. The amplitude of this phase is very low, likelyindicative of a very low vertical velocity gradient or perhaps high attenuation in the upper84mantle. P is not observed on the two other record sections with sufficiently large shot-receiver offsets (SPs 8 and 20-W). The low signal-to-noise ratio at large offsets on both ofthese sections, together with the expected low amplitude of P (based on the low amplitudeP seen on SP 3), likely explain this absence. A paucity of P appears to be a consistentfeature of data recorded along other wide-angle reflection lines in both the Coast Belt (i.e.,line 10, O’Leary et al. 1993; line 2, McLean and Spence 1992) and Insular Belt (McMechanand Spence 1983) and to the south in northern Washington state (Gridley 1993). Thiscontrasts sharply with data recorded in the Intermontane Belt (line 1) where P is recordedvery clearly, even from small (200 kg) shots.Reflections from the crust-mantle boundary (PmP) are observed on SPs 8, 4, 21-W and3. The character and times of arrival of this phase differ from section to section suggestingsubstantial lateral variation in the nature of the Moho. Location of the PmP critical pointis difficult to determine from true relative amplitude plots due to the large fluctuations inamplitudes recorded from trace to trace. On SP 8, PmP can be most easily recognized whenthe data are scaled to a common maximum amplitude (Fig. 3.2). PmP then appears as afairly wide band (-.‘0.5 s duration) of relatively high amplitude energy between 80—170 kmover which the reduced arrival times vary between 5.1—5.9 s. Wide angle PmP arrivals canbe followed to —‘240 km where the reduced arrival time is 6.9 s.PmP on SP 4-W (Fig. 3.4) appears as an increase in energy 0.4—0.5 s behind P andPg (at 6.2 s) between 0—30 km. Unlike SP 8, where PmP is clearly present beginning atoffsets of 90 km, here there is no indication of PmP until 150 km offset. PmP arrivals thatare present are incoherent. PmP on SP 4-E can be seen most clearly between 280-3 15 kmat 6.4 s with the critical point possibly at 290-3 10 km. The presence of PmP at offsets of95—130 km, and relative coherency of this phase, is again in contrast to Pm recorded tothe west from the same shot.PmP on SP 21-W (Fig. 3.5) is observed between 90—125 km beginning at -.6.1 s. Again,the phase is not sharp but consists of a band of energy almost 1 second in duration. Location85of the critical point cannot be precisely determined but appears to be near or beyond themaximum offset of 150 km.By far the most prominent and laterally coherent Moho reflection is present on SP 3(Fig. 3.7) where it can be seen between 60—250 km. It is very prominent between 170—230km (6.4—6.7 s) where the frequency content is much higher than at larger offsets (8.6 s at60 km). The wide-angle PmP arrivals are seen most easily when the data are scaled to acommon maximum amplitude (Fig. 3.7). The apparent asymptotic velocity of this phase is‘7. 1 km/s. providing another indication of high velocities in the lower crust. Location ofthe critical point is, as on other sections, difficult to determine.3.5 2—D Modelling of Line 3 Refraction DataThe general procedure for interpreting in-line data was described in section 2.5. Thissection describes the modelling procedure in detail and points out the nature of the constraintssupporting the final model (Fig. 3.9a). Figure 3.10 shows the locations of nodes wherevelocity and boundary depths were determined by traveltime inversion, and identifies layerand boundary numbers referred to in the text. Ray tracing diagrams illustrating the coverageprovided by each shot are shown in Figs. 3.1 la—3.17a. Figure 3.18 illustrates the raycoverage provided by each phase for all shots. Comparisons between observed and calculatedtraveltimes are shown in Figs. 3.1 lb-3. 17b. Synthetic sections generated from the model inFigure 3.9a are presented in Figs. 3.1 lc—3.17c and the observed record sections plotted intrue relative amplitude format are shown in Figs. 3.lld—3.17d. To compensate for anela.sticattenuation, amplitudes were calculated assuming a simple Q structure based on valuesinferred for the Pacific Northwest by Singh and Herrmann (1983). Table 3.2 summarizesthe results of all traveltime inversions.The final velocity model (Fig. 3.9a) is divided into five regions: near surface layer, upper,middle and lower crust and upper mantle. The upper boundary in Fig. 3.9a is a smoothedapproximation to the surface topography along the line. The true elevation profile ranges86ijpu(v)tuS13wsarusjoqwiCsnoiso3SA\4flOSu&l1uwpsptrU13JOA11AOOJZSSpuS)OJuoinjdSfloUIWfljoAsusaidatdouoipjddiU1UUS3fl4‘HS!IIUflO‘NOULT2UOS!.TJH“fl2iO‘spmsiqiouopsquozjouopisodwpcoiddsusaidaij(v)woijusinojoouuttpONj:sIAs)uo!wuuoju!iu!sICqdoiqopuoiouop.rudpsiq(v)upowi(ipopijouo1thu!3wpsjdwt(q)uowuuojuiqioioj61t(j:uoixI!IA)£UjqnuqSOAjooidinoiuooinojo(v)6tanii_IIi I111111 p...ala’0101010101cob,bb,bkb’c 000000001000VELOCITY(km/s).0)0)0)0)bbraa)VELOCITY(kmfs)-J0)00DEPTH(km) .Ci)Mo0000 IIDEPTH(km) C,)-00000(‘10G) w00-LC;’01) 00 I0--‘—I--“lizIAC,) 0- 00DISTANCE (km) E0 50 100 150 200 250 300Figure 3.10 Locations of nodes where boundary depths and velocities weredetermined by inversion of the line 3 traveltime data. Boundary 1 representsthe topography. Layer 1 = near-surface layer. M, Moho.between 0 and 1340 m and contains many large fluctuations (e.g., site to site differences ofseveral hundreds of metres). Elevation corrections (typically less than 25 ms) were applied tothe observed traveltimes to compensate for the difference between the true elevation profileand profile of the model.Velocities in the near-surface layer at the location of each shot point, except SP 9, weredetermined by inverting P traveltimes. Because the P branch is typically very short, thevelocities in this layer can be well constrained only near the shot point locations. Also, thedata do not constrain a velocity gradient and thus gradients were fixed for this inversion usingvalues determined by the maximum observed offset of P. Significant lateral variations in thevelocity and thickness of the near-surface layer were added later through forward modellingof rays transiting the layer to account for some of the “kinks” in Pg.wI0D Boundary node• Velocity node Layer BoundaryV Shotpoint number number88N—.‘ )-CI)H—DISTANCE (km) E50 100 150 200 250 300I I0‘I0CQC’)IH(a)7111rl—t0 I I I0 50 100 150 200 250 300DISTANCE (km)(b)Figure 3.11 Two-point ray tracing diagram showing total ray coverage (a) andcomparison between observed (short vertical bars representing uncertainty ofpicks) and calculated (dot) traveltimes (b) for SP 8 into line 3.89NCl)H(I)HFigure 3.11 (continued) Synthetic section (c) and observed record section(d) for SP 8 into line 3. Trace amplitudes are scaled to shot-receiver offset raised tothe 1.5 power. Data in (d) are band pass filtered from 3—15 Hz.CD10‘IW E0 50 100 150 200 250 300Distance (km)90Figure 3.12 Two-point ray tracing diagram showing total ray coverage (a), comparisonbetween observed (short vertical bars representing uncertainty of picks) and calculated(dot) traveltimes (b), synthetic section (c) and observed record section (d) for SP 9into line 3. Trace amplitudes in (c) and (d) are scaled to shot-receiver offset raised tothe 1.5 power. Data in (d) are band pass filtered from 3—15 Hz.Depths to the base of the near-surface layer in the vicinity of the shot points areconstrained by Pg arrivals. Boundary nodes were placed on boundary 2 at shot pointlocations. Pg arrivals were also inverted to determine velocities in the upper crust. Twolayers (layers 2 and 3) represent the region within which refracted Pg energy turns. Arrivalsout to and beyond offsets of about 100 km were modelled as turning rays through layers2 and 3, respectively. Velocity gradients in these two layers were fixed in the inversion asdetermined by forward modelling of the Pg amplitudes. Depths to the base of layers 2 andw DISTANCE (km) EDt1F(a)ID0IHID0-0 50 100w E‘a a(b)CI)E5432100 50 100DISTANCE (km)0 20 40 60 80 100 120Distance (km)91Cf)C\10- I I50 100 o 20 40 60 80 100 120DISTANCE (km) Distance (km)Figure 3.13 Two-point ray tracing diagram showing total ray coverage (a), comparisonbetween observed (short vertical bars representing uncertainty of picks) and calculated(dot) traveltimes (b), synthetic section (c) and observed record section (d) for SP 10into line 3. Trace amplitudes in (c) and (d) are scaled to shot-receiver offset raised tothe 1.5 power. Data in (d) are band pass filtered from 3-45 Hz.3 (boundaries 3 and 4) were fixed at the shallowest depths that would allow turning rays toreach all receivers that recorded Pg.R1 traveltimes were inverted to solve for depth to the base of the upper crust (boundary5). Because there was no evidence to suggest that significantly higher or lower velocitieswere present beneath the maximum depth of penetration of Pg (boundary 4), velocities at thebase of the upper crust (layer 4) were fixed at values present at the base of layer 3, resultingin a zero vertical gradient. The R1 traveltime data suggested a thinning of layer 4 to thew DISTANCE (km)0 50 1000E0c)LC)(l200w E:“ +1 u(b)10-.--.‘ 4U)3292IC)NCD-IC)ICH0 50 100DISTANCE (km)200Figure 3.14 Two-point ray tracing diagram showing total ray coverage (a) andcomparison between observed (short vertical bars representing uncertainty ofpicks) and calculated (dot) traveltimes (b) for SP 4 into line 3.w0-\,,150 250E300N(a)i&ij44(b)‘ttit Id—I‘4-0 1,I I I0 50 100 150 200 250 300DISTANCE (km)93NCDLOHC\l0 50 250w8— —_______7.-6-C/) 5-3-H2-1—U-— I Io 50Figure 3.14 (continued) Synthetic section (c) and observed record section(d) for SP 4 into line 3. Trace amplitudes are scaled to shot-receiver offset raised tothe 1.5 power. Data in (d) are band pass filtered from 3—15 Hz.200E100 150 200 250 300Distance (km)94DISTANCE (km)100 150 200 250I I I I100 150 200 250 300DISTANCE (km)Figure 3.15 Two-point ray tracing diagram showing total ray coverage (a) andcomparison between observed (short vertical bars representing uncertainty ofpicks) and calculated (dot) traveltimes (b) for SP 21 into line 3.east. This layer was pinched out at 208 km as a modelling convenience. It is possible thatthe more shallow R1 reflections (from SP 4-E; see Fig. 3.14a) are unrelated to the originof the deeper R1 reflections to the west.Velocities at the top of the mid-crustal layer were determined by forward modelling R1amplitudes. Based on a lack of evidence to support a significantly thick low velocity zonein this region of the model (i.e., layer 5) a positive velocity contrast across boundary 5 wasw0‘-4300E0C\10CDIH(0-IC)ITIt,t *14UI—1 (b)95N-ccir)Ci2C I I100 150 200 250 300W E87-6-5-4-3-2-1 (d)— I I I80 100 120 140 160 180 200 220 240Distance (km)Figure 3.15 (continued) Synthetic section (c) and observed record section (d) for SP21 into line 3. Trace amplitudes are scaled to shot-receiver offset raised tothe 1.5 power. Data in (d) are band pass filtered from 3—15 Hz.assumed.A simultaneous inversion of R2, P, Pm and P was performed to determine velocitiesat the base of the middle crust, in the lower crust and upper mantle, and depths to the baseof the middle and lower crust (boundaries 6 and 7). Velocities at the base of the middle crustare primarily constrained by wide-angle R2 reflections, although the lateral coverage affordedby this phase is limited (see Fig. 3.18c). In addition, velocities based on this phase representC/)260 280 300 320 34096w DISTANCE (km) E100 150 200 250 300I I I I ICH(a)10-CI)I- /H/(b),a’_0- I I100 150 200 250 300DISTANCE (km)Figure 3.16 Two-point ray tracing diagram showing total ray coverage (a) andcomparison between observed (short vertical bars representing uncertainty ofpicks) and calculated (dot) traveltimes (b) for SP 20 into line 3.upper bounds on the actual velocities. Depth to the base of the middle crust (boundary 6)is primarily constrained by R2 and P traveltimes. A single boundary node (i.e., horizontalboundary) was sufficient to represent this boundary.Velocities in the lower crust are constrained by both P and PmP arrivals. Verticalgradients in the central part of the profile were fixed based on results from amplitudemodelling of P. Because of the weak gradient inferred here, only the uppermost partof this layer is constrained by P (see Fig. 3.18c). Velocities at the top of this layer near97CDH(c)__0---100 150 200 250 300W E8-7-6-C/] 5-H2-1- (d)____________O________ __________— III liii III I I III100 120 140 160 180 200 220 240 260 280 300 320 340Distance (km)Figure 3.16 (continued) Synthetic section (c) and observed record section (d) for SP20 into line 3. Trace amplitudes are scaled to shot-receiver offset raised tothe 1.5 power. Data in (d) are band pass filtered from 3—15 Hz.the western and eastern end (nodes at 34 and 300 km) were fixed based on R2 amplitudes.Velocities at the base beneath these two points were allowed to vary.The Moho is constrained by PmP and P arrivals. Five was the minimum number ofboundary nodes required to represent the topography of this boundary. The upper mantlevelocity is constrained by unreversed P arrivals from SP 3. One velocity node, i.e., constantvelocity at the top of the upper mantle, was sufficient to provide a good fit. Because of thevery low observed P amplitude, the gradient was fixed in the inversion to the lowest possible98CtE:ciHFigure 3.17 Two-point ray tracing diagram showing total ray coverage (a) andcomparison between observed (short vertical bars representing uncertainty ofpicks) and calculated (dot) traveltimes (b) for SP 3 into line 3.w DISTANCE (km)0 50 100 150 200 250 300I I IE(a)C‘I0)-NC,]IH-4-0-t’i VtI t 4(b)I I I I0 50 100 150 200 250 300DISTANCE (km)99CNr))If)H— I I0 50 100 150 200 250 300w10-—8-6-2-(d)0 — IIII III0 503.17 (continued) Synthetic section (c) and observed record section (d) forSP 3 into line 3. Trace amplitudes are scaled to shot-receiver offset raised tothe 1.5 power. Data in (d) are band pass filtered from 3—15 Hz.150 200Distance (km)1000 50 100 150 200 250 300C10ZC’2F0-4ía)S0ía)0 50 100DISTANCE (km)150 200C-250 300NF0-4CDPgandPsrays (a)C.-4C0 50 100 150 200 250 300I I I I I I aR2 and Pc rays50 100 15004::(c)2000C -C --4C -C\1C-Cr)C -250 300Figure 3.18 Ray coverage provided by each phase: (a) P and P5; (b) R1; (c) R2 and P; (ci)PmP and P. Rays connecting all shot-receiver pairs are shown. Layer boundariesare indicated by broken lines. Note that vertical scale varies from panel topanel. Shot points that recorded each phase are indicated by•.101tRMS(ms)Table 3.2 Final velocity model inversion results for line 3. R2, Pc, PmP, and Phave been combined, since they were inverted simultaneously. tRMs is thetraveltime residual between observed and predicted data.Phase(s) No. of Average Normalizedinverted observa- uncertainty x2 misfittions (ms)value that would allow rays to be traced to all receivers. Additional forward modelling ofthe amplitudes was performed after the inversion resulting in a zero gradient at the west endof the model to explain the absence of P on SP 8.3.5.1 Model Resolution and Absolute Parameter UncertaintiesA description of the methods for estimating spatial resolution of a model about a specificdepth or boundary node and absolute uncertainties of parameters is given in section 2.6.5.The results of these tests for the line 3 model are shown in Table 3.3.3.6 Principal Features of the Velocity ModelThe essential features of the velocity model are shown in Fig. 3.9a. The near surfacelayer is about 1.2 km thick on average but is significantly thinner beneath SPs 8 and 20(60O m) and thicker beneath SP 3 (-3.2 km). Average velocities are 5.2 and 5.6 km/sat the top and bottom, respectively. Superimposed on this average velocity structure areseveral low velocity anomalies responsible for spatially short delays, most noticeable in Pg,mentioned above. Velocities in these anomalies average 3.2 km/s although they are notNo. of No. ofvelocity boundarynodes nodesPs 70 31 65 4.43 6 0Pg, layer 2 548 39 71 4.52 14 0Pg, layer 3 169 61 71 3.09 11 0R1 75 70 56 1.18 0 4R2, Pc. PmP, Pn 393 101 88 1.30 11 6102Table 3.3 Estimated lateral resolution of the line 3 velocity model about specified nodes andabsolute uncertainty of velocities and depths. Resolution and uncertainty values are basedon tests applied to representative nodes. Tests were not applied to all nodes.Estimated lateral Estimated absoluteNodes resolution (km) uncertaintyVelocities in upper crust (layers 2 and 3) 10 - 25 0.1 - 0.3 km/sTwo eastemmost boundary nodes on boundary 5 25 - 30 3 - 4 kmTwo westernmost boundary nodes on boundary 5 60 - 90 1.5 - 2 kmVelocities at base of middle crust (layer 5) 50 - 70 0.2 km/sBoundary node on boundary 6 100 2- 3 kmVelocities at top of lower crust 35 0.2 km/sVelocities at base of lower crust 100 0.2- 0.3 km/sBoundary nodes on boundary 7 (Moho) 50 - 70 1.5 - 2.0 kmUpper mantle velocity 100 - 150 0.1 km/sstrongly constrained since in most cases rays do not turn through the anomalies. The mostsignificant, in terms of lateral extent, is beneath SP 9.Velocities at the top of the upper crust average 6.1—6.25 km/s and are generally lowereast of the Harrison Fault. Velocities at the base average 6.4—6.45 km/s west of the HarrisonFault (HF, Fig. 3.9a) but are considerably lower (6.2 km/s) to the east. The base of theupper crust, which is defined by R1 reflections from a .-s6.4 over 6.6—6.7 km/s velocitycontrast in the western two thirds of the model, decreases from .‘ 12.5 km at the west end to-.‘8 km at 200 km. East of this, the boundary is not well constrained because of a lack ofR1 reflections, which is consistent with the smaller velocity contrast east of SP 21. In thissense, the distinction between upper and middle crust at the east end of the line becomesless marked.Velocities in the middle crust show large lateral variations with a significant transition103into lower velocities occurring near SP 21. Velocities are in the 6.6—6.9 km/s range to thewest and 6.35—6.45 km/s to the east. Although these values are based primarily on R2reflections the coverage afforded by this phase on the western and eastern sides of the modelis roughly equivalent (Fig. 3.1 8c). The lateral variations in velocity within this layer arethus likely significant as they are based on similar constraints. The base of the middle crustis at 21 km. The lower crustal velocity structure is more homogeneous with velocities in the6.9—7.05 km/s range, though again there is a slight trend to lower velocities at the easternend of the line.Depth to the Moho increases from 29.5 km beneath SP 9 to 37 km beneath the centralpart of the line. Depth decreases to 34 km just east of SP 20. Velocity at the top of the uppermantle is a relatively low 7.65 km/s with a gradient of 0.01 east of 150 km and a zerogradient to the west. The low medium velocity arises from a combination of the observedP apparent velocity of 7.9 km/s and the topography on the Moho defined by PmP.3.7 Discussion of ResultsFigure 3.9b shows an interpretation of the velocity model of Fig. 3.9a based partially onother geophysical and geological information. Low velocities in the near surface layer anduppermost crust centered at about 30 km represent the Georgia Basin. In the vicinity of line 3the basin extends from the eastern edge of Vancouver Island into the Strait of Georgia to justwest of Lasqueti Island (Fig. 3.1) and is underlain by Wrangellian basement (Monger 1990);it is probably no more than 2 km thick (White and Clowes 1984). The average velocityvalues for the basin (4 km/s) and underlying uppermost crust (6.1—6.2 km/s) agree withthose determined by White and Clowes (1984) from a small-scale refraction survey. Theiridentification of the 6.1-6.2 km/s crustal layer as Coast Belt granitic rocks is, however, atvariance with the boundary between Wrangellia and the Coast Mountains located by surfaceexposures (east of their refraction profiles). It is more probable that this layer, beneath theStrait of Georgia, comprises Wrangellia.104Beginning at about the Insular-Coast Belt boundary (85 km) and continuing east to about175 km, velocities at the top of the upper crust are low with respect to velocities at the samedepths to the west (Fig. 3.9b). The position and extent of this zone of low velocitiescorrelate well with the Jura-Cretaceous plutonic rocks of the WCB which link Wrangelliato the Harrison terrane west of the Harrison Fault. Reflection data recorded further north(profile 88-16, Fig. 3.1) have failed to outline the geometry of the intrusives. The gravitymodel of Dehier (1991) for the westernmost Coast Belt requires a thin (<5 km) layer of lowdensity plutonic rocks over higher density (Wrangellian) rocks for the western 20—30 kmfollowed by a rapid thickening to the east. Given the lateral resolution of the velocity modelat upper crustal depths (see Table 3.3), this location is consistent with the beginning of therelatively low upper crustal velocities at -.80 km. Magnetotelluric investigations (Jones et al.1992b) show the upper s14 km of the WCB’s crust to be significantly more resistive thandeeper structure by a factor of 7—30. This thickness is similar to the upper crustal thicknesseast of 100 km. The consistency of these two depth estimates and association of resistiveupper crust with plutonic granitoids (Gough and Majorowicz 1992) leads to the interpretationof the upper crust of the refraction model east of 100 km (to a depth of 13 km in the west,shallowing to 8 km in the east) as the intrusive mass of WCB plutonic rocks.East of -.‘220 km, velocities in the uppermost crust again decrease and remain relativelylow to the eastern edge of the model. The western edge of this zone (at 220 km) coincideswith the westernmost surface exposure of the Harrison terrane and marks the eastern boundaryof abundant plutonic rocks in the WCB. There is no evidence for the existence of differentterranes between eastern Vancouver Island (Wrangellia) and the Harrison terrane; for certainthey were together by 165 Ma (Monger 1991). On this basis the high velocity middle crustwest of the Harrison Fault, and upper crust west of 100 km is interpreted as Wrangellia.This is consistent with an interpretation of seismic reflection data recorded 130 km to thenorth (Varsek et al. 1993). Velocities in these layers are also consistent with those interpretedfor Wrangellia beneath Vancouver Island and the westernmost Coast Belt from refraction data105(McMechan and Spence 1983; Spence et al. 1985; Drew and Clowes 1990).The ‘C’ zone (Fig. 3.9b) is an east-dipping reflective zone interpreted by Clowes et al.(1987) to be the base of Wrangellia, beneath which lies high velocity underplated materialof the subducting Juan de Fuca plate. It is clearly imaged in reflection data recorded acrossVancouver Island (Green et al. 1986; Clowes et al. 1987) and is tentatively correlated withreflectors on reflection profile 88-16 (Fig. 3.1) on the mainland (Clowes 1990; Varsek et al.1993). Beneath SP 8 the C zone is at 16 km depth; however, its position to the east isuncertain. Reflection data suggest it is listric into the Moho. Teleseismic receiver functionanalysis (Cassidy and Ellis 1993), which provides S-wave structure, indicates the C zonemay flatten into the lower crust at ‘26 km to at least as far east as SP 10. Regardless,the refraction data do not image the reflective C zone; there is no correlatable zone of highvelocities in the lower crust at the west end of the model. The absence of any signaturefrom this zone may be (i) because velocity anomalies associated with it are not resolvableby the data —for example, the reflectivity has been associated with a decollement zone atthe base of Wrangellia (Clowes 1990); (ii) because of its presumed location at the edge ofthe velocity model (Fig. 3.9b) where turning ray coverage is poor and constraints tend tobe weakest; or (iii) related to the sub-perpendicular orientation and position of line 3 withrespect to the other studies across the Insular-Coast Belt boundary. Estimates of the positionof the C zone from these other studies indicate that, at most, underplated material wouldoccupy only a small wedge of lower crustal material west of “-i 100 km (Fig. 3.9b).Based on a lack of other evidence to the contrary, the deep (>21km ) crust east ofthe C zone, and extending perhaps as far as 250 km, is interpreted as the lower crust ofWrangellia. The division of the crust beneath the WCB into three layers (plutons, middle andlower Wrangellia) and depths to intervening boundaries are consistent with the conductivitystructure for this area (Jones et al. 1 992b). Thus the resistive (>1000 lm) upper crust isassociated with plutonic rocks; less resistive (.‘ 150 flm) middle crust with the middle crustof Wrangellia; and conductive (.‘30 Zm) lower crust with the lower crust of Wrangellia.106The high velocity middle crust extends eastward to about 250 km, near the Harrison Fault.Lower crustal velocities are also, on average, lower to the east of this location (6.75—6.95versus 7.0 km/s). This location likely coincides with the eastern limit of deep Insularsuperterrane (Wrangellia?) crust. The exact location and nature of the transition from Insularto Intermontane crust is not resolved by the refraction data due to the low lateral resolution ofvelocities in the middle and lower crust (Table 3.3). Nevertheless, this estimate is consistent,given the northwest strike of the geology in this region, with the reflection interpretationof Varsek et al. (1993) which shows the wedging of the Intermontane superterrane into theInsular superterrane ‘—20 km to the east of the Owl Lake Fault, approximately beneath theCentral Coast Belt detachment (see Fig. 3.1 for location).Upper crustal velocities are significantly lower east of ‘-.220 km. These coincide withseveral north-south elongated terranes and overlying granitic rock traversed by the line:Harrison and Shuksan terranes, which are correlated with Cascades structures (Monger 1991),and Bridge River and Methow terranes of the Eastern Coast Belt.Distinct terranes, and depth extent of these terranes, are not resolved by the refractiondata. Based on the interpretation of reflection data they are presumably thin (<10 km) thrustsheets (Varsek et al. 1993), the lower crusts of the terranes having been delaminated orflaked off during Late Cretaceous contraction as the Insular and Intermontane superterranescollided. East of #s250 km, and beneath these terranes, sits Intermontane superterrane crust,possibly the middle and lower crust of Quesnellia present in the Intermontane Belt east ofthe Pasayten Fault (Fig. 3.1). Mid-crustal velocities are consistent with those beneath line1 (situated in Quesnellia; see Chapter 2). Lower crustal velocities are higher than beneathline 1 (—‘6.65—6.9 km/s versus 6.55 km/s). The crustal velocity structure east of -.250 km ismostly consistent with that at the northern end of PACNW profile 1 (see Fig. 3.1 for location)(Gridley 1993). The primary difference is that the upper crust beneath PACNW profile 1, to adepth of about 11 km, comprises three layers (not including a thin low velocity near-surfacelayer): The first extends to a depth of 4 km with velocities of 5.2—5.7 kmls, the second107extends to a depth of 7 km with velocities of 6.8—7.0 km/s, and the third, with velocitiesof 4.5—5.0 km/s, marks a strong low velocity zone. The extremely high velocity secondlayer is interpreted to be an ophiolitic body, correlatable with outcrops of mafic and ultramafic rocks along the western margin of the northern Cascades. The underlying low-velocitylayer is interpreted as a sedimentary layer. There is no counterpart for this three layer uppercrustal structure in the line 3 model. Ophiolites do occur in the southeastern Coast Beltin the Bridge River, Methow, Shuksan and Cadwallader terranes (J.W.H. Monger, personalcommunication, 1994) but are not resolvable by the refraction data. Thickness of the uppercrust is similar as is the average velocity in the middle crust (6.4 km/s), depth to the base ofthe middle crust (21 km), velocity at the top of the lower crust (6.7 kmls) and depth to Moho(34—37 km). Velocities at the base of the lower crust are somewhat higher (7.3 km/s versus6.95 km/s along line 3), though not strongly constrained. The similarity in crustal structuresupports a link between structures of the Northwest Cascades and the Coast Belt east ofHarrison Lake and is consistent with geological mapping which indicates that NorthwestCascades structures wrap around Harrison Lake to the east and extend northwest (Crickmay1930; Journeay and Csontos 1989). The surface location of the Insular-Intermontane sutureis the Central Coast Belt detachment, which at the latitude of line 3 occurs near the easternboundary of the Harrison terrane. This is consistent with the velocity model given theresolving power of the refraction data.The velocity structure beneath line 10 (Fig. 3.1; O’Leary et al. 1993) shows a greateraffinity with velocities to the east of the Harrison Fault. In particular, velocities within,and depths to the base of the upper and middle crust are similar. Lower crustal velocitiesbeneath line 3 are higher (6.75—6.9 km/s versus 6.45—6.7 kmls). There is also close agreementwith depth to Moho at the eastern end of line 3 (—‘35 km; however O’Leary et al. (1993)interpret a 2 km thick crust-mantle transition zone that was not required in the line 3 model.Upper mantle velocities, which are not strongly constrained on either model, are significantlydifferent (7.65 km/s versus 8.15 km/s beneath line 10). An upper mantle reflector at a depth108of 70 km beneath line 10 is not observed beneath line 3. The lack of high velocities in themiddle and lower crust beneath line 10, interpreted as middle and lower Wrangellia beneathline 3, suggests that north of line 3 Wrangellia does not extend eastwards as far as line 10.O’Leary et al. (1993) compared the line 10 model with two models for the collision zonebetween the Insular superterrane and North America (J.M. Journeay in Monger and Journeay1992; Varsek et al. 1993). The similarity in gross velocity structure between line 10 and line3 east of the Harrison Fault in part led them to favour the crustal delamination model of J.M.Journeay (in Monger and Journeay 1992) in which the Insular superterrane was displacedalong east vergent faults over rocks of the southwestern Coast Belt.Moho depth throughout most of the Coast Belt beneath line 3 is 34—37 km. To the north,beneath line 10, the Moho is at 35 km with little variation in depth (O’Leary et al. 1993). Tothe south, Moho depth is 34—37 km beneath the northern end of PACNW profile 1 (Gridley1993). These results indicate that depth to Moho does not vary significantly throughout muchof the southern Coast Belt and northern Cascades. A significant shallowing does occur inthe west where the depth decreases to 29.5 km beneath the eastern Insular Belt. This issignificantly less than previous estimates of crustal thickness in the Insular Belt (Berry andForsyth 1975; McMechan and Spence 1983; Spence et al. 1985; Drew and Clowes 1990)in which Moho constraints were weak. East of 150 km the TW’T’T to the refraction Moho(.‘ 11 s) is consistent with reflection data recorded to the north (‘-.‘ 11.4 s; Varsek et al. 1993).To the west of 150 km, reflection data for profile 88-16 (Fig. 3.1) do not delineate clearly theposition of the Moho as there is a prominent wedge-like zone of reflectivity from S (30km) to 12 s (40 km); Varsek et al. (1993) place the reflection Moho at 12 s. The latteris significantly deeper than the refraction Moho. In contrast, results derived in this studysuggest that the top of the prominent band (at l-.d30 km depth; 9 s TWTT) is the Moho. Theinterpreted low velocity (7.65 kmls) below the shallow Moho, combined with the reflectionsignature, might be indicating a thick crust-mantle transition in the westernmost Coast Belt.Clowes (1990) has also alluded to this possibility. Alternatively, Varsek et al. (1993) show109that the top of this reflective package may represent the eastward continuation from beneathVancouver Island of either the reflective C or B zones (Green et al. 1986; Clowes et al.1987). Interpretation of the reflective wedge on profile 88-16 has important implications onthe structural significance of the C and E reflectors and possible fate of material displacedby the subducting Juan de Fuca plate (Varsek et al. 1993).3.8 SummarySeismic refraction data recorded across the eastern Insular and Coast Belts reveal largelateral variations in velocities, particularly within the upper and middle crust, that correlatewell with geological and other geophysical information. The most striking feature of themodel is the decrease in velocities to the east in the vicinity of the Harrison Fault in theregion where the surficial boundary between terranes of the Western and Eastern Coast Beltsoccurs. This location may represent the transition (suture?) from Insular to Intermontanesuperterrane crust at middle and lower crustal levels. This location is consistent with aninterpretation of reflection data recorded further north (Varsek et al. 1993) which showsthe wedging of Intermontane crust into Insular crust about 20 km east of the Owl LakeFault. Crustal velocities east of the Harrison Fault show a greater affinity with IntermontaneBelt velocities beneath line 1 (Chapter 2) and velocities in the northwest Cascades beneathPACNW line 1 (Gridley 1993). The crust of the WCB comprises three layers: low velocity,relatively high resistivity plutonic rocks which extend to about 12 km depth; and highervelocity, more conductive middle and lower crust, probably of Wrangellia. To the west,beneath the Insular Belt the upper crust comprises relatively high velocity Wrangellian rocks.An eastward extension of the seismically reflective C zone from beneath Vancouver Island isnot imaged by the refraction data suggesting that it may not be associated with a significantP-wave velocity contrast at its presumed location beneath the Strait of Georgia. Crustalthickness is greatest beneath the WCB (s37 km) and thins slightly to the east to about 34km beneath the Eastern Coast Belt. Beneath the eastern Insular Belt, the Moho is at about11030 km, significantly thinner than previous estimates.1114 3-D INTERPRETATION METHOD4.1 IntroductionA major objective of the refraction survey was to image the three-dimensional velocitystructure of the crust and upper mantle within the triangular study region. To meet thisobjective a large number of fan shots, situated at strategic locations throughout the array,were recorded, with minor exceptions, by all receiver sites providing 3—D ray coverage witha broad range of offsets and azimuths as required for traveltime tomography.This chapter begins with a discussion of tomographic methods and several previous 3—Dseismic surveys. This is followed by a description of the methods used to analyze the SCoREdata, including a review of the first arrival traveltime inversion procedure of Hole (1992).Details of the forward modelling and inversion of reflection traveltimes, including synthetictests, and inversion of first arrival traveltimes for depth are also presented. The chapter endswith an outline of the complete 3—D modelling algorithm.Some of the techniques described in this chapter were developed in collaboration withJ.A. Hole. Hole was primarily responsible for the forward modelling, reflection modelling,and tomographic and refracting interface inversions, while I was primarily responsible forthe inversion of reflection times and incorporation of these times into the tomographyand interface inversions; however, collaboration was involved at several stages of thedevelopment.4.2 Tomographic Methods and Previous 3—D Seismic SurveysA number of methods for imaging 3—D velocity variations exist; see Thurber and Aki(1987) for a brief review. Most of these have been applied to the study of earthquake arrivaltimes from either teleseismic or local events. Popular methods of solution include directdamped least squares (Aki et al. 1977) and algebraic reconstruction (Humphreys and Clayton1988). Phillips and Fehler (1991) compare the popular methods. When local earthquake data112are used a simultaneous inversion to determine both seismic velocity structure and locationparameters is frequently performed (e.g., Thurber 1983). An important distinction betweendifferent techniques is the method used to represent the earth’s velocity structure. The typeof parameterization used, along with the amount, type and quality of data, governs the typeof structure, and its resolution, that a tomographic inversion procedure will produce (Zhouet al. 1992). Also important is the type of three-dimensional ray tracing scheme employed,as this too can affect the accuracy of the results.For example, Aki et al. (1977) divide the earth into constant velocity blocks and performa one step inversion of teleseismic P-arrival times using a homogeneous halfspace as theirinitial model. This greatly simplifies the ray tracing but results show that the final solutionis highly dependent on the starting model (Hawley et al. 1981). Thurber (1983), performinga simultaneous inversion of local earthquake and explosion traveltimes, defines the velocityfield at a large number of regularly spaced grid nodes and uses interpolation to determinethe velocity at intervening locations. He approximates the true 3—D raypath by examining alarge number of rays comprising arcs of varying radii, and lying in planes of varying dips,and choosing the arc with the smallest traveltime.Most tomographic studies involving earthquake data use only first arrival times andseldom employ accurate 3—D ray tracing or allow the incorporation of velocity discontinuities.An exception to this is the study of Zhou et al. (1992) who invert refracted arrivals fromlocal earthquakes to determine 3—D structure beneath northeastern Japan. They parameterizethe earth as a series of layers separated by seismic velocity discontinuities, and use a 3—Dgrid net for each layer to define the velocity. Accurate raypaths between the discontinuitiesare calculated using the pseudobending technique of Urn and Thurber (1987). Snell’s law isused to find the intersection points between the rays and discontinuities.In comparison with earthquake analysis techniques, methods designed specifically toimage 3—D structure from crustal scale refraction experiments are far fewer in number,probably because of the relative scarcity of large 3—D refraction surveys, i.e., surveys which113include a large number of fan shots and are designed specifically to image 3—D structure.Recent studies include the work of Pavlis (1986) who used a circular recording geometry (witha radius of 190 metres) and 24 fan shots to investigate small scale variations with a methodthat reduces the 3—D imaging problem to a 2—D one. Toomey et al. (1990) investigate 3—Dstructure beneath the East Pacific Rise over an area of --‘18x 18 km2. They used a rectangulargrid of 15 ocean bottom seismographs to record 480 explosive shots and several airgunprofiles. They used the tomographic imaging technique of Thurber (1983) together with thepseudo-ray bending method of Urn and Thurber (1987) to trace rays. Amato et al. (1991)derive 3—D upper crustal structure beneath a volcanic complex in Italy. They employed a42 km-per-side triangular recording geometry supplemented by additional intersecting in-lineprofiles. They used data from both reversed and unreversed in-line profiles, three fan shots,and earthquake data. To obtain 3—D structure they used the method of Thurber (1983).The first use of the “spatial seismic refraction recording” method, i.e., a triangular arrayof shots and receivers as described in section 1.5, in Canada was a refraction survey of theWilliston Basin, southern Saskatchewan in 1981 (Kanasewich and Chiu 1985). This surveycomprised three 288 km-long lines in the form of an equilateral triangle. Shot points locatedat each vertex were recorded by all receiver sites around the triangle. An average stationspacing of 6.5 km (45 receivers per side) resulted in a total dataset of -.‘400 traces, of which“.‘ 135 were fan shot recordings. Thus, while the scale of this survey is similar to SCoRE‘89, the total number of traces available for analysis was significantly less as the SCoRE ‘89dataset comprises 10,300 traces, of which .4000 were fan shot recordings.To solve for 3—D velocity structure beneath the triangular array Kanasewich and Chiu(1985) chose to use a simple model parameterization and a damped least squares inversionprocedure. By assuming constant velocity layers with (piecewise) plane interfaces theywere able to use a fast and accurate ray tracing scheme (Chander 1977) to forward modelreflected and critically refracted arrivals. Using a layer stripping procedure they invertedthree reflected phases (two intracrustal and PmP) and upper mantle refracted arrivals to solve114for the velocity in four layers and the structure of the three intervening boundaries, eachcomprising three normally faulted planes. While most parameters in their study were welldetermined due to the overdetermined nature of the procedure, the ability of their method toresolve detailed variations in both velocity and depth was severely limited by their simplemodel parameterization and small dataset.The method of Kanasewich and Chiu (1985) was extended by Phadke and Kanasewich(1989) to allow the incorporation of vertical velocity gradients within each layer andpolynomial surfaces. They also used a layer stripping approach and employed damped leastsquares inversion to solve for velocities in two layers and the structure of two interfacesbeneath a reflection line recorded across Vancouver Island. The very poor 3—D ray coverageafforded by their data restricted their ability to solve for both gradients and structure onthe interfaces.Thus, most tomographic methods offer advantages and disadvantages which can makethe choice of a scheme difficult. As mentioned, drawbacks can include the use of:1) only first arrival traveltimes; rarely are both refracted and reflected arrivals used.2) inaccurate forward modelling (ray tracing) techniques, or accurate methods which arecomputationally very slow.3) model parameterizations that limit the resolving power of the data; usually this isgoverned by the choice of inversion scheme.4) inversion schemes which can be computationally very slow for large problems.4.3 Inversion of First Arrival Traveltimes for SlownessThe tomographic inversion procedure of Hole (1992), with modifications to allow thecalculation and inversion of reflection traveltimes, is used to determine 3—D velocity structurefrom SCoRE ‘89 data. This method is computationally efficient and allows for a denselysampled velocity model. Key elements of this procedure, described in more detail below,include: (i) forward modelling of traveltimes using a 3—D finite-difference algorithm; and115(ii) a choice of inversion parameterization that greatly simplifies and speeds up the linearizedinversion (by taking advantage of the linear independence of ideal rays).Details of the theory for the inversion of first arrival traveltimes for slowness are givenby Hole (1992) and therefore only a summary is given here.The traveltime t of a seismic arrival following a raypath L thorough a three dimensionalvolume of reciprocal velocity (slowness) u ist= fu(r)dt , (4.1)L[u(r))where dl is an incremental length along the path L. Since the raypath depends upon theslowness field the relationship between t and u is nonlinear. Applying Fermat’s principle,i.e., traveltime along the raypath is stationary with respect to perturbations of the path, leadsto the approximationSt = JSu(r)dl , (4.2)L[u0(r)]where uo is some reference slowness field, Su = u(r) — u0(r) is the slowness perturbationand St = t(u(r)) — t(u0(r)) is the traveltime perturbation or residual. This is the standardlinear relationship between traveltime residual (data) and slowness perturbation (model) usedfor traveltime inversion. Usually, an iterative solution to this linearized representation of thenon-linear problem is sought.Taking St to be the ith datum, equation (4.2) can be written= fSu(r)gi(r)dV, i = ito M , (4.3)where gj(r) is a function which describes the ith raypath through the volume V. and M isthe number of data. The functions g(r) are called data kernels. Parameterizing the slownessmodel Su(r) asNSu(r) = cr,h3( ) , (4.4).1=1116where h,(r), j = 1 to N are a set of basis functions, N is the number of model parameters,and the aj are coefficients to be determined, equation (4.3) becomes= EaiJhj(r)Yi(r)dV . (4.5)3=1In most tomographic inversion procedures the velocity model is parameterized in terms ofa set of boxes (pixels) with(1, if r inside jth box;h,(r)= j 0, otherwise.For large problems (i.e., many pixels) this parameterization requires the inversion of a verylarge and sparse matrix. To eliminate this requirement Hole (1992) follows Harris et al.(1990) and defines both the data kernels and basis functions to be deltalike functions locatedalong the ray path withg,(r) h,(r)= { urn , for r along the jth raypath; (4.6)0, otherwise,where A can be thought of as the cross sectional area of the infinitesimally narrow ray.Setting g, = h, produces the smallest model (i.e., minimizes llSu(r)2 ). Also, since themodel is parameterized along raypaths, N = M. Substituting equation (4.6) into (4.5) givesA8t1crj=3where ij is the length of the jth ray, so that (4.4) gives(St,/13, for r along the jth raypath;Su(r) = (4.7)1 0, otherwiseas the slowness perturbation due to the jth datum and solution to equation (4.2). Theperturbations are added to the slowness model to create a new reference model zto (r). Thefinal slowness model is determined by iterating until some stopping criterion is satisfied, e.g.,the rms traveltime residual reaches an acceptably small level.Parameterizing the model in terms of functions defined along raypaths for the inversion isnot a convenient choice for the forward modelling of traveltimes. Hole (1992) uses the threedimensional finite-difference algorithm of Vidale (1990) to compute first arrival traveltimes.117The velocity model is defined at a set of uniformly spaced grid nodes in 3—D. First arrivaltraveltimes are calculated at each node of the grid using finite-difference operators basedon the eikonal equation of ray tracing. The operators use the average slowness across agrid cell and are equivalent to a plane wave approximation across a grid cell (Hole 1992).The traveltime at an arbitrarily located receiver is determined using tn-linear interpolationof the traveltimes at the eight surrounding grid nodes. First arrival raypaths are found byfollowing the steepest gradient in traveltime from the receiver to the shot point (Vidale1988). The slowness perturbations in equation (4.7) are defined along the raypaths andmust be transformed into the same parameterization (i.e., a 3—D grid) used in the forwardmodelling. Hole (1992) calculates the slowness perturbation at each grid node as the averageof all the Lu terms in some volume surrounding the grid node. This gridding procedure alsoserves as a smoothing operator which stabilizes the inversion. Additional smoothing of thegnidded Lu terms is performed by applying a 3—D moving average (MA) filter. A flowchartoutlining the complete tomographic inversion procedure for determining velocity structurefrom first arrival traveltime data is shown in Fig. 4.1. Synthetic examples demonstrating thecapabilities of the procedure, and methods for estimating the resolution of resultant modelsare given by Hole (1992). An application of the procedure to real data from a seismicrefraction survey in the Queen Charlotte Basin is given by Hole et al. (1993a).The inversion procedure can be simply modified so that a layer stripping procedure can beused. This is necessary to allow the velocity perturbations of deep phases to be applied onlyalong the part of the ray path that is most strongly constrained by the phase. For example,perturbations due to P traveltime residuals are applied only along the sub-Moho part of theraypath. Thus, rather than using the entire path length to calculate the slowness perturbationsin equation (4.7), the length of the raypath beneath a specified 3—D surface is used.The inversion procedure for first arrivals (including layer stripping) can also be appliedto reflected phases (section 4.4). Thus, for example, PmP traveltimes can be inverted forlower crustal velocities and/or depth to Moho (section 4.5).118Figure 4.1 flowchart for the modelling of first arrival traveltimes for3—D velocity structure based on the method of Hole (1992).-!!---nishedarray to keep track of sum of6u terms in each cellarray to keep track of number ofrays penetrating each cellDo forward modelling for shot number nshot:calculate traveltimes to each noint of 3-DBackpropagate ray number nray, keeping track of cellsCalculate length of ray 1, traveltime residual 6t, and slowness6u=&’I. Undate Z&(iik) andSloppingC1i1C1i4)fl satislie([?(e.g., total mis (Itsmall enough?)Update the velocity model:v(jjk) = 1/(6u(jjk) + 1/v(ijk))1194.4 Forward Modelling of Reflection TimesThe 3—D finite-difference algorithm of Vidale (1990) calculates first arrival traveltimesto each node of a 3—D grid. A method for calculating reflection traveltimes that uses thisalgorithm has been presented by Hole et al. (1993a). The method relies on an extension tothe finite-difference algorithm that allows the computations to begin from an arbitrary wallof a previously computed 3—D traveltime field.The reflector depth model is parameterized by a two dimensional grid of nodes withthe same spacing as the velocity model. Prior to the calculations, velocities beneath thereflector are fixed, e.g., replaced with velocities immediately above the reflector and extendeddownwards with a zero vertical gradient, to ensure that the first arrival traveltimes at thereflecting surface are from the downgoing wave and not from waves that have turned beneaththe reflector. The general procedure for generating reflection times consists of the followingsteps:1) Calculate downgoing times from shot point to each grid node. In particular, this givesthe times at the reflector.2) Traveltimes at grid nodes above the reflector are replaced with large “dummy” values.3) The downgoing traveltimes on the bottom face of the model are used as a source to beginthe calculation of upgoing times to each grid node. Previously computed downgoingtimes, in particular the large “dummy” values, are replaced with upgoing times whenthe upgoing times are earlier.Effectively, this results in the downgoing (incident) time field at the reflector being usedas a source for the propagation of the upgoing (reflected) time field (see Fig. 4.2). Holeet al. (1993a) used this procedure to forward model reflection times and determine depthto Moho assuming a horizontal reflector. The accuracy of this procedure depends on howthe downgoing times at the reflector are handled. Two methods will be presented and theiraccuracy compared.120E1o20- 30A ftNE1o20- 30AfNFigure 4.2 illustration of the general procedure for calculating reflection times above a reflector(heavy solid line). (a) Calculate downgoing times throughout model (represented by wavefronts)from shot point (triangle). (b) Replace traveltimes at grid nodes above reflector with large“dummy” times. (c) Use traveltimes along bottom face of the model as a source to begincalculation of upgoing times, replacing previously computed times (i.e., “dummy” values) that arelater with upgoing times. Raypath (broken line) from receiver (dot) to reflector is found byfollowing traveltime gradient through upgoing time field. Raypath allows intersection point of raywith reflector to be determined. Upgoing raypath from reflector to shot is found by following thetraveltime gradient through the downgoing time field back towards the source (panel a).4.4.1 Simple MethodThe first method is illustrated in Fig. 4.3. The true reflector is discretized by movingeach depth point to the nearest grid node (Fig. 4.3a). If the depth to the true reflector variesin excess of the node spacing the discretized reflector will contain steps. Completing steps 2and 3 of the general procedure generates an upgoing time field contaminated by diffractionsfrom the corners of the discretized reflector (Fig. 4.3b). The diffractions lead to artifacts, inthe form of discontinuities, in the time field. The magnitude of these effects are governed0 50 100 150 200 250 3000 50 100 150 200 250 3000 50 100 150 200 250 300Distance (km)121by the grid cell size and topography of the reflector. For a horizontal reflector, as used inthe study of Hole et al. (1993a), these effects disappear._____________Figure 4.3 “Simple” method for generating reflected times throughout 3—D grid (illustrated in2—D). (a) True reflector (heavy broken line) is discretized by moving each depth node to the nearestgrid node (open circles). Incoming plane wavefield (dotted lines) and raypaths, and desired (true)reflected raypaths from true reflector are shown. (b) Shows typical results (i.e., after completingsteps 2 and 3 of general procedure; see text): diffractions (circular dotted lines) from the corners ofthe discretized boundary lead to artifacts in the form of discontinuities in the reflected time field.For the purposes of traveltime inversion the reflected traveltime, location of the reflectionpoint, and angle of reflection (calculated as the angle between the incident ray, whosedirection is given by the average traveltime gradient at the reflector, and normal tothe reflector, calculated from the average gradient in reflector depth) must be calculatedaccurately. The accuracy of these parameters were tested against ray-traced results usingthe model in Fig. 4.4. The model comprises a 6 km/s constant velocity medium and areflector dipping at 45°. The discretized reflector would be represented by a “staircase”type structure. A line of receivers is oriented parallel with the X-axis, the strike direction,so that the perpendicular distance between the reflector and recording line is a constant26.9 km. The 3—D finite-difference model has a cell size of 1 km3 and dimensionsflx X fly x n3 = 201 x 51 x 54. Errors in the calculated traveltimes along the recording122line, reflection point location, and angle of reflection as a function of shot-receiver offsetare shown in Fig. 4.5.xykmFigure 4.4 Model used to test the two methods for generating reflected times with first-arrivalfinite-difference algorithm. Dimensions are n x n, x n = 201 x 51 x 54; cell size is 1 km3.Velocity is constant (6 kmls) and there is no dip in the direction of the recording line.Traveltime errors are large, ranging from —100 to +300 ms. Errors in the Y and Zcoordinate of the reflection point are less than 2 km; errors in the X coordinate are up to±6 km. Errors in the angle of reflection are relatively small (—3.3° to +2°) at offsets lessthan .s4O km, but are much larger for greater offsets, such as would be encountered for realMoho reflections (shot-receiver offsets for observable arrivals are typically >50 km).Since accurate estimates of the reflection angle are necessary for inversion these errors areproblematic. The poor performance of this method is motivation for the method describedbelow.4.4.2 Accurate MethodThis method does not involve discretization of the reflector in the manner of the previousmethod. Instead, the downgoing wavefield is used to analytically calculate the reflected timeat one or more grid nodes immediately above, and below, the reflector (see Fig. 4.6). Theanalytically calculated times are used to propagate the reflected time field upwards. This1230.30.200—0.1() 50 100 150:. Error in angle ofreflection•—.—N>. ...—.•••1..(c)—— I I I I I I50 1 50 100 150shot-receiver offset (km) shot-receiver offset (kin)Figure 4.5 Error in calculated values for the (a) traveltime, (b) X, Y and Z coordinates ofthe reflection point, and (c) angle of reflection as a function of shot-receiver offsetusing the “simple” method. Errors are measured against ray-traced results.method is based on the assumption that the incident and reflected waveflelds are locallyplanar and the reflector surface is locally planar.The details of the development in this section are presented by Hole and Zelt (1994).In this paper, Hole was responsible for the development of modifications to an existing3—D finite-difference algorithm (Vidale 1990) for large velocity contrasts, and primarilyresponsible for the forward modelling of reflection times; however, there was substantialcollaboration on the latter technique.Let z:,, be the depth at reflector node (i, i) and dI,J,k be the depth within the velocitymodel at grid node level k. Zj,j is not constrained to lie at the depth of a grid node. Timesare analytically calculated for each node (i, k), k = k0 to k2 of the velocity model, where(i,j, ko) is the shallowest node such that d,,,k0 is greater than the reflector depth at the ninenodes surrounding and including node (i,j), and (i,j, k2) is the deepest node such that d,,,k2is less than the reflector depth at the nine nodes surrounding and including node (i, j) (Fig.I I I I I I I I I I I I IError In traveltime -AtRMS = 0.092 s, . ‘.‘.• •• —. ••.. ...._•••%_.•.••*..••I I I•1 I.ii._’t I i I I I I I0—5—15 -124k=k2=k’k=I,Figure 4.6 “Accurate” method for generating reflected times to nodes (i, i k), k = k0 to k2 (largesolid dots) above and below reflector node (i, j) (large square dot). Illustrated in 2—D. (a) Slope oftrue boundary (grey line) is approximated by line (plane in 3—D) fit through surrounding two (fourin 3—D) depth points (small square dots). Incoming plane wavefield (broken lines), i.e., traveltimegradient direction g is determined from times at the four black dots (six in 3—D). The verticalcomponent of the traveltime gradient and distance between node (i, i k2) and reflector are used tocalculate downgoing time at reflector. (b) Direction of reflected ray and reflected traveltime gradienth is determined from incident ray direction and unit normal to the reflector fi using Snell’s law.Reflected time at nodes (i, k), k = k0 to k2 are calculated from time at reflector, the verticalreflected traveltime gradient and the distance between the nodes and reflector. (c) Location of nodes(black dots) used in calculations for a reflector that does not locally cross a vertical grid nodeboundary. In this case, a time is analytically calculated only for the nodes at (i, k0) and (i, k1).4.6a). The incident ray direction g used to calculate the reflected times at nodes (i, , k),k = k0 to k2, is calculated from the average traveltime gradient defined by the downgoingtimes at the four nodes surrounding (i,j, k2), and nodes (i,j, k’) and (i,j, ku). Defining(i, , k1) to be the deepest node such that d:,3,kz < z:,,, then k’ is the minimum of k1 andk=k0y1251c2 + 1, and k” is the minimum of k1 — 1 and k2. Note that the magnitude of g is equal tothe local slowness. The components of g are— ti+1,j,k2 — tj_1,j,k226— ti,j+1,k2 — ti,j_1,k2 4gy— 26— t,j,kl — ti,j,kIIg—where S is the grid node spacing. For the situation illustrated in Fig. 4.6a, k’ = k1, k” = k2,and k0 = k1 + 1. For a reflector that does not vertically cross a grid cell boundary at any ofthe nine surrounding nodes as illustrated in Fig. 4.6c, k’ = k1, k” = k1 —1, and k0 = k1 +1.Note that (i,j, k0) is below the reflector. In fact, it is not necessary to analytically calculate“back-propagated” reflected times at nodes deeper than (i, , k1) (i.e., below the reflector) forthe forward modelling (i.e., calculation of reflected times). Times are analytically calculatedat one (or more) nodes below the reflector, however, to allow accurate ray tracing from areceiver to the reflector (required for traveltime inversion), since the ray direction in the cellcontaining the reflector is determined by nodes both above and below the reflector.The upward facing normal vector to the reflector surface, n, is calculated from the averagegradient in reflector depth determined from the depths at the four grid nodes surroundingnode (i, i) in the 2—D reflector model as— Zi+1,j — zi_1,j26— Z:,i — Zj,j_1 (4.8)26n = —1The time at reflector node (i, j) (large open square in Fig. 4.6a) is determined from theknown time at node (i,j, k2), ti,3,k2 the vertical gradient of the traveltime, g, and thevertical distance between node (i,j, k2) and the reflector, ast(z:j) ti,j,k2 + gz(zi,, — dI,,,k2) . (4.9)The direction of the reflected ray h (i.e., the traveltime gradient of the reflected wavefield, with magnitude equal to the local slowness) is determined from the incident ray g and126the unit normal to the surface, ñ= n/I!n!I, using Snell’s lawh=g—2(gñ)ñ , (4.10)and the reflected time at nodes (i,j, k), k = k0 to k2, t,,,k, is given byti,j,k t(z,.,) — h(zi., — d,,,k) , k = ko to k2 . (4.11)Times are analytically calculated at one or more nodes immediately above and below eachnode of the reflector model using equation 4.11. Steps 2 and 3 of the general procedure arefollowed to produce the reflected time field. This method avoids the problem of the “simple”method because, in effect, analytically calculated reflection times immediately above thereflector are used as a source to propagate the reflected time field throughout the model.For very steeply dipping reflectors, k2 will be several grid nodes above the reflector andk1. As the dip of the reflector increases, the distance between k2 and the reflector increasesand the assumption of local plane waves becomes invalid. In practice, the calculations areaccurate for dips less than e—35°.Results of the test for accuracy using the model in Fig. 4.4 are shown in Fig. 4.7. Themaximum traveltime error is 18 ms; maximum error in the reflection point coordinates is0.35 km; and, for offsets >50 km. the error in the angle of reflection is <<10. Tests usingother more complicated models show that this method is always significantly more accuratethen the “simple” method and errors are small, as required to allow accurate inversion ofreflection traveltimes for depth.4.5 Inversion of Reflection Traveltimes for DepthA simple inversion scheme which, like the inversion procedure for velocity, eliminatesthe need for matrix inversion is used to determine reflector depth from reflection data. Theprocedure is analogous to that described by Hole et al. (1992) for the inversion of refractedfirst arrivals to determine interface structure. In both procedures a set of depth perturbations1270.030.0200.0101 I I—.__E 0)N—1 - : Error in angle ofreflection(c)—2 I I I I0 50 100 150shot-receiver offset (km) shot-receiver offset (km)Figure 4.7 Error in calculated values for the (a) traveltime, (b) X, Y and Z coordinates ofthe reflection point, and (c) angle of reflection as a function of shot-receiver offsetusing the “accurate” method. Errors are measured against ray-traced results.are gridded to obtain a reflector (interface) perturbation model, and both assume the velocitymodel is accurate.For reflections, the partial derivative relating a small change in reflector depth Sz to thechange in traveltime St isSt 2cosO—= cosa, (4.12)Sz Vwhere 0 is the angle between the ray and interface normal, a is the dip of the reflector(measured as the angle between the normal to reflector and vertical), and v is the velocityabove the reflector (Zelt and Smith 1992). The right hand side of this equation can becalculated for each ray from the velocity and reflector models and ray path. The change indepth Lzk for the kth traveltime residual tk is thenzZk= St/Sz (4.13)I I I I I ‘ ‘ I I 1 I I ‘ IError in traveltime- AtRMS=0.012S—•_._•.....__ (a) —— I I I I I I I I I50 100 150128where the depth perturbation applies at the reflection point.A set of M data will give M depth perturbations LZk(x, y), k = 1 to M, which aregridded to obtain a reflector perturbation model lZjj. The choice of gridding scheme isimportant, especially when there are large gaps in reflection point coverage, or no coveragearound the edge of the reflector model. The LaPlace interpolation method, which fits thedepth perturbations to a function z(x, y) satisfying the equation V2[z(x, y)] = 0 , is agood choice because it introduces minimal structure between the depth perturbations. For realdata with errors, the gridded surface will generally contain spikes arising from closely spacedreflection sampling points that are associated with significantly different depth perturbations.Spikes are removed and the gridded surface is smoothed by applying a 2—D MA filter. Aflowchart outlining the iteration procedure for determining reflector depth is shown in Fig.4.8.The resolving power of this method is governed primarily by the density of reflectionpoints, errors in the data, and errors in the assumed velocities (Hole et al. 1992). Thesefactors control the size of the operator used to smooth the depth perturbation surface. Testsusing synthetic data as presented in the following section are the most effective means ofevaluating the abilities of this procedure.4.5.1 Tests Using Synthetic DataTwo examples are presented to illustrate the accuracy with which known structure canbe recovered using the inversion procedure outlined above. Both examples use the SCoRE‘89 shot and receiver recording geometry and the same model dimensions (n X fly X z =221 x 301 x 56 3.7 million nodes, or 264 x 360 x 66 km) and grid node spacing (1.2km) employed in the analysis of real data presented in Chapter 5 (see Fig. 4.9). Noise-free synthetic data were generated for the same shot receiver pairs that recorded real PmParrivals. However, only data from the seven large shot points at the line ends, midpoints, andcenter of the triangle are used in the examples. Velocities were fixed during the inversions.129Figure 4.8 Flowchart for the modelling of reflection traveltimes for 3—D reflector structure.Calculate ray-reflector intersection point (y), 0, a,Calculate &‘3z = (2 cosO cosa)/vCalculate depth perturbation for ray nray == biI(8t13z)Sluppiiagriiterioii atiSlie(I!(e.g.. toliI ems (ItsimII eiioiigh?)—--nish130YFigure 4.9 Schematic representation of the 3—D finite-difference velocitymodel used in synthetic tests and real data analyses.The velocity model used for both tests is based on the 2—D velocity models for lines 1, 3(Chapters 2 and 3) and 2 (McLean 1994). Velocities interior to the triangle were determinedby linearly interpolating in the strike direction (roughly NW-SE); velocities exterior to thetriangle were extrapolated outward from the 2—D models in directions perpendicular to thevertical plane comprising each 2—D model. The 3—D velocity model constructed in this waywas smoothed with a 7 x 7 x 3—point MA filter to give the model used in the tests. Thismodel is identical to the starting 3—D velocity model use in the real data analysis (Chapter 5).4.5.1.1 Sine wave reflector modelThe reflector model (Fig. 4.lOa) used to generate the synthetic data was produced byadding two sinusoidal surfaces. The primary features of the model within the study area arethe shallow dome-like feature centered in the southern part of the triangle, and deep bowl-like feature centered on the west side of the northern part of the triangle. The minimum andmaximum depths are 29 and 37 km. A total of 1358 data were generated; shot receiver offsetsnz =z 2jf131were limited to a maximum of 250 km. The starting reflector model was a horizontal planeat 35 km depth. Figure 4.lOd shows the traveltime residuals for the starting model plottedagainst the Y coordinate of the reflection point. The rms traveltime residual for the startingmodel is 352 ms. The depth perturbations used in the first iteration are shown in Fig. 4.lOe.The roughly sinusoidal pattern in both figures is an expression of the true reflector geometry.At each iteration the gridded reflector perturbation model was smoothed with one pass ofa 21 x 21—point MA filter. Figure 4.lOb shows the final reflector depth model determined afterthree iterations (rms traveltime residual = 61 ms) and the reflection point locations. Furtheriterations did not decrease the residuals. Structure between and exterior to the reflectionpoints arises from the gridding and subsequent smoothing of the depth perturbations and isunconstrained. In this example, the smallest resolvable feature would be 13 x 13 km (halfthe size of the smoothing operator).The final model is in good agreement with the true model. The shape and depth of thebowl-like feature in the north is accurately recovered. The depth to the dome-like structure inthe south is in good agreement with the true model, but the feature itself is slightly distorted.Other features around the edges of the model outside the region sampled by the data are notrecovered. The final model does correctly include an increase in depth at the lower rightcorner of the model based on reflection points which sample the edge of this feature of thetrue model. The difference between the final and true model (Fig. 4.lOc) shows that errors inthe region of coverage are generally within ±1 km, i.e., less than the grid node spacing (1.2km). Errors in the data and/or velocities will, of course, increase the errors in the final model.Figures 4.lOf and g show the traveltime residuals and depth perturbations used in thefinal (third) iteration. The traveltime residuals are small and tightly clustered around tt = 0s. The depth perturbations are also relatively small and clustered around 1z = 0 km. Thelarger scatter in both plots for smaller Y reflection point coordinates is likely related to thepoorer recovery of the southern dome-like feature.13229.0 30.6 322 33.8 35.4___ ___reflector depth (km)Figure 4.10 (a) Contour plot of sine wave reflector model. White lines represent approximatelocation of recording lines; A, shot points used in synthetic tests. (b) Final model after threeiterations. Dots are PmP reflection points and represent point constraints on depth to reflector. (c)Difference between final and true models. (d) rms traveltime residuals for first iteration plotted as afunction of the Y coordinate of the reflection point. (e) Depth perturbations used in first iteration.(f rms traveltime residuals used in third iteration. (g) Depth perturbations used in third iteration.350300250350300200>-25015010050.200>-150100150X (km)035030025020015010050Starting model residualsATRUS.O.352e N.1358.5 .3 -1 1 3 5error in reflector depth (km)Residuals used In 3rd It.AThUS.O.061s N.1353300002o0.L00.(f)o._IM (a)300’0I.L0 -••(g)1O—5 u 5 iO& (tan)0150x (kni)37.01334.5.1.2 Vertical fault modelThe reflector model (shown in Fig. 4.1 la) represents a vertically faulted Moho. Depthsto the reflector are 31 and 36 km to the northeast and southwest of the fault respectively.The fault is located approximately along the surface trace of the CB-IMB boundary. A totalof 1965 data were used. Data to the northeast and southwest of the fault were generatedseparately from horizontal reflectors at 31 and 36 km. respectively. That is, the fault modelshown in Fig. 4.1 la was not used to generate all data simultaneously because the methodfor calculating reflections is not valid for very steeply dipping faults. The starting reflectormodel was a horizontal plane at 34 km. Smoothing operator dimensions were identical tothe previous example. Figures 4.1 id and e show the starting traveltime residuals and depthperturbations used in the first iteration. The offset nature of the residuals and perturbationsis an expression of the faulted true model. The rms traveltime residual for the starting modelwas 304 ms.The final model, after three iterations, is shown in Fig. 4.1 lb. and the difference isshown in Fig. 4.1 lc. The final model images the vertical fault as a dipping fault withapproximately the correct throw. The base of the fault is close to the true fault location. Thewidth of the dipping fault zone is 25 km. approximately the width of the operator usedto smooth the reflector depth perturbations. Away from the fault the depths agree to within±1 km (this is less than the grid node spacing) of the true model in the region sampled bythe data. The final traveltime residuals and depth perturbations are shown in Fig. 4.llf andg. The relatively large scatter on these plots and large final rms traveltime residual (133 ms)arises because it is not possible to exactly reconstruct the true model: the method used togenerate reflections is invalid for vertical faults. In addition, smoothing the updated reflectormodel limits the maximum dip with which a vertical fault can be represented.4.6 Inversion of First Arrival Traveltimes for DepthHole et al. (1992) describe a procedure for inverting first arrival traveltime data to find1343500,05 :.error in reflector depth (km)Starting model reslduele Residuals used In 3rd It.AThMS 0.304 $ N1965Figure 4.11 (a) Contour plot of vertically faulted reflector model. (b) Final model after threeiterations. Dots are PmP reflection points and represent point constraints on depth to reflector. (c)Difference between final and true models. (d) rms traveltime residuals for first iteration plotted as afunction of the Y coordinate of the reflection point. (e) Depth perturbations used in first iteration.(j9 rms traveltime residuals used in third iteration. (g) Depth perturbations used in third iteration.>-300250200>-150100500350300250__200>-15010050150X (km)5300I.200.100I___d)t (a)50 100 150 200 250X (km)S.0200I0029.0 30.8 32.6 34.4 36.2reflector depth (km)300• ..4,:: A•• (g)•10z (km)3.0135the structure on an interface separating two media of different, and known, velocities. Holeet al. (1993b) present an application of the procedure to data from a seismic refraction surveyin the Queen Charlotte Basin. The inversion scheme is similar to that outlined in section 4.5for the inversion of reflection times for depth, with equation (4.12) replaced bySt cos Ui cos 62—=— cosa, (4.14)OZ V V2where 01 and 62 are the angles between the ray and interface normal measured above andbelow the interface, and vi and vi are the velocities immediately above and below theinterface.The procedure described by Hole et al. (1992) was designed specifically for geometriesin which each ray path crosses the interface only once. For example, they describe anapplication of the procedure to determine depth to the basement interface beneath the QueenCharolette Basin in which the recording line was situated above the basin and source pointswere outside the basin. For this type of geometry the depth perturbation for each traveltimeresidual, given by equation (4.13), is applied at the ray-interface intersection point. For themore usual case where the ray crosses the interface twice, for example, P at the Moho, thisprocedure can be generalized in a straightforward manner.Rather than applying the entire traveltime residual to one crossing, the residual is splitbetween the two crossings. The manner in which the residual is split is arbitrary, but theleast biased choice is to apply half of the residual at each crossing, so thatt/2IZ= St/Sz ‘for both crossings. For each ray the correct factor for splitting the residual is unknown.Thus it is important to have uniform ray coverage (i.e., with a broad range of azimuths)to reduce modelling errors that arise from applying the incorrect fraction of the residual ateach crossing.A problem which can affect the performance of the inversion arises because ofinaccuracies in the calculated angle of refraction (62 in equation 4.14) at the first, i.e., closest1360Cj.CDISTANCE (km)150 200Figure 4.12 P rays traced through a typical velocity model. In the vicinity of the critical point(box) the refracted rays fan out through a relatively large range of angles (02) over a small region.For a typical grid cell size the error in the calculated value of 02 will be large because ofinaccuracies in the finite-difference calculations at this region of strong wavefront curvature.to the shot point, ray-interface crossing. This is illustrated in Fig. 4.12 which shows a suiteof P rays traced through a typical velocity model. In the vicinity of the critical point (boxin Fig. 4.12) a narrow cone of rays incident on the interface fans out upon refraction througha wide range of angles (typically 750 9 900). As represented in the finite-differencemodel with a grid cell size of “-‘ 1.2 1cm, this occurs over a few grid cells. Since only asingle ray direction can be represented within each cell (given by the direction of the averagetraveltime gradient), the resolution of the 02 values is very poor. For a typical model this willlead to errors in the calculated value of St/Sz of 10—40%. The accuracy can be increasedby regridding the traveltime field in the vicinity of the critical point i.e., interpolating thetraveltimes onto a finer grid to more accurately calculate 02. Note however, that the errorsquoted above are potentially much smaller than that due to the equal split of the residualbetween the two crossings. Also, because the inversion procedure is iterative, these errorsshould not seriously affect the accuracy of the inversion assuming isotropic ray coverage.Instead, it is the rate of convergence that is affected.4.7 3—D Modelling AlgorithmThis section presents a modelling algorithm employing the basic techniques describedabove: forward modelling and inversion of first arrivals and reflection traveltimes for velocity137and depth. The steps of the algorithm are based on the constraints provided by the data. Forthe real data analysis the dataset comprises the arrival times of three phases: Pg, P (firstarrivals) and PmP (Moho reflections). Pg constrains only the upper-most crustal velocitystructure. There are no first arrival data to constrain deeper crustal velocities. Reasonableoptions for dealing with the lower crust are: (i) use a (fixed) simple lower crustal velocitymodel, e.g., a 1—D average model based on the 2—D in-line models, or a 3—D modelconstructed from the 2—D models by interpolation between the models; or (ii) use PmP(with layer stripping) to solve for lower crustal velocities (as well as for Moho topography).In the first option the PmP data misfits will only be mapped into Moho depth perturbationsand it is accepted that the lower crust may include gross errors in velocity. In the secondoption the PmP data misfits are mapped into both depth and velocity perturbations, however,it is not known how the misfits should be partitioned into the two types of perturbations.Regardless of the option chosen it is clear that the classic trade-off between velocity anddepth introduces a degree of non-uniqueness that is difficult to assess. The results of manysynthetic tests and preliminary tests with real data show that option (2) is preferable becauseit is more effective in reducing the rms traveltime error and because it allows for someinference on possible lower crustal velocity variations. The same reasoning can be appliedto the use of P to constrain both upper mantle velocities and depth to Moho since againthere is an ambiguity regarding the source of P misfits. Thus, both PmP and P are usedto constrain velocities and depth to Moho.Steps in the 3—D modelling algorithm are illustrated in the flowchart of Fig. 4.13.Initially, the iterative procedure outlined in section 4.3 is used to invert Pg traveltimes forupper crustal velocities. Upper crustal velocities are then held fixed for all subsequentinversions of deeper phases (layer stripping). To do this it is necessary to construct a3—D surface (upper-lower crust boundary) separating the region of the velocity modelconstrained by Pg from deeper regions. The surface is constructed by gridding a set ofpoints corresponding to the maximum depth of penetration in each vertical column of grid138cells comprising the velocity model. After gridding the surface is smoothed with a 2—D MAfilter to produce the final upper-lower crust boundary. A constant depth value is added toeach constraint point prior to gridding to account for the smoothing.The deeper phases (PmP and P) are then inverted quasi-simultaneously for lower crustalvelocities, Moho depth, and upper mantle velocities. A true, simultaneous, inversionis not performed. Instead, each step of the quasi-simultaneous inversion comprises: 1)inverting PmP traveltimes for lower crustal velocities; 2) inverting PmP and P traveltimessimultaneously for Moho depth using the procedures outlined in sections 4.5 and 4.6; and 3)inverting P traveltimes for upper mantle velocities (layer stripping is used to keep crustalvelocities fixed during the inversion). These three steps are repeated until a stopping criterionis satisfied or until the rms traveltime residual stops decreasing.In synthetic tests this procedure was sufficient to adequately fit the data. Tests withreal data, however, showed that after a few iterations for deeper structure the rms traveltimeresidual stopped decreasing without the data (particularly PmP) being adequately fit. Thisis apparently caused by the convergence to a “final” Moho depth model after only a fewiterations. In further iterations, small (insignificant) adjustments to the Moho in the secondstep of the quasi-simultaneous inversion tend to counter improvements to the data misfits inthe other two steps. Thus, to further reduce the misfits the Moho is fixed and PmP and P areinverted in two stages to solve for velocities in the lower crust and upper mantle, respectively.In the first stage, PmP traveltimes are iteratively inverted to solve for lower crustal velocitiesuntil a stopping criterion is satisfied. This serves to map (most of) the remaining PmPtraveltime residuals into velocity variations. In the final stage, P traveltimes are iterativelyinverted to solve for upper mantle velocities until a stopping criterion is satisfied. This servesto map (most of) the remaining P traveltime residuals into velocity variations.Other approaches to modelling the deep structure are, of course, possible. For example,one alternate procedure, simple layer stripping, consists of (i) using PmP to solve alternatelyfor lower crustal velocities and depth to Moho; (ii) fixing lower crustal velocities and depth139Invert Pg for upper crustalvelocities. Construct starting velocitymodel for inversion of deeper phases.Construct 2-D surface belowwhich Pg rays do not penetrate(upper-lower crust boundary)________Invert Pn for upper mantle velocitiesUpdate velocity model.NoI’iisto))iIIgr_Invert PmP for lower crustalvelocities. Update velocityInvert PmP and Pn for MohoUpdate Moho depth model.Invert Pn for upper mantleUpdate velocity model.NoYesRNIS1rieIIinw(Ivcreasilig?I“Quasi-simultaneous”inversion procedureInvert PmP for lower crustalvelocities. Update velocity model.NoPini’SI4)I)I)illei’ite iioilsat ist ied ?—nishedFigure 4.13 Flowchart of the 3—D modelling algorithm used in the real data analysis. Insetfigure of schematic velocity model illustrates the five basic elements comprising thecomplete model: three 3—D velocity models and two 3—D surfaces.140to Moho; and (iii) using P to solve for upper mantle velocities. Synthetic tests showed thatthis was less effective at recovering structure than the procedure outlined above.1415 ANALYSIS OF 3—D DATA5.1 IntroductionIn this chapter the procedures outlined in Chapter 4 are used to develop a 3—D velocitymodel for the southwestern Canadian Cordillera. A description of the fan data and otherdata used in the analysis is presented first (along with a 2—D model for line 4), followedby details of the modelling procedure and a discussion of model resolution and uncertainty.This is followed by a presentation of the results in various display formats (horizontal andvertical slices, 1—D profiles), including results obtained using an alternate (1—D) startingmodel. The results are compared with other geophysical data and the chapter concludes witha discussion and summary.5.2 3—D DatasetFigure 5.1 shows the location of shot points and receiver sites used in the 3—D analysis.Data from the following shot-receiver combinations are used:1. In-line data recorded along lines 1, 2 and 3. The line 1 and 3 datasets are describedin Chapters 2 and 3. McLean (1994) presents the line 2 dataset. The data recordedby receivers to the west of SP 5 and data from SPs 8 and 9 (Fig. 1.2) are not usedin the analysis.2. Fan data from the seven large shots (SPs 1, 3, 5, 4, 7, 12, 17) recorded along lines 1, 2and 3. Record sections are shown in Appendix A (Figs. A.1—A.12).3. Data recorded along the three short interior lines (4, 5 and 6). Figures A.13—A.16 showall data recorded along line 4.4. In-line data recorded along the southeastern end of line 10 of SCoRE ‘90 (O’Leary etal. 1993). Data from SPs 46, 47 and 48 recorded to the southeast of SP 48 are included.The arrival times of three phases were inverted to determine 3—D crustal and uppermantle structure: Pg, PmP and P. The three phases are exemplified on the record section for142350__2002>-150100X (km)Figure 5.1 Shot points and receivers used in 3—D analysis of data. All map-view results aredisplayed using the rotated (7.6°) Mercator coordinate system shown here. Table 5.1 lists thenumber of data recorded along each line. 1MB, Intermontane Belt; CB, Coast Belt;FRF, Fraser River Fault; HF, Harrison Fault; K, Kamloops; V, Vancouver.SP 12 recorded on line 1 (Fig. A.6). Fan data picks were made from various display formatsincluding record sections plotted as a function of shot-receiver azimuth and shot-receiveroffset. The latter, which can be used for oblique shot-receiver geometries for far-offset tracesis more useful for trace-to-trace correlation. Azimuthal plots of the data with normal moveout300250143PhaseLine Calc./Est.10 Total unc. (ms)Pg 1559 956 819 340 47 34 208 3963 55/75PmP 1000 410 355 112 11 5 10 1903 80/110Pa 589 305 179 66 14 15 0 1168 80/100Total 3148 1671 1353 518 72 54 218 7034Elevation corrections were applied to the observed traveltimes to effectively strip offtransit times through the near-surface layer. This layer, as imaged on the 2—D in-line modelsfor lines 1—4 and 10 is typically 0.5—2 km thick, comprising highly variable and relativelylow velocities compared to underlying structure. Because this layer is typically only 1—2grid cells thick the high vertical gradients common in this layer and strong velocity contrastsacross the base of this layer cannot be accurately represented by the finite-difference velocitymodel. To reduce errors in the finite-difference-calculated time field associated with theseproblems, elevation corrections were applied to all receiver sites along lines 1—4 and 10 andall shot sites. Although elevation corrections were not applied to sites along lines 5 and 6,these sites account for only a small number of data (Table 5.1) and the errors associated withthis should be negligible. For a receiver or shot along a given line the correction has theeffect of moving the receiver or shot vertically downwards to the base of the first layer ofthe corresponding 2—D model. The correction for a receiver moved from a depth z to thebase of layer 1 at depth z1 (Fig. 5.2) is1.t=(z—zo)( 1• — 1 “ (5.1)\VjSlflS v2tanOjcorrections applied were also used. This format is useful for trace-to-trace correlation of thereflection phase. Table 5.1 displays the number of data recorded along each line.Table 5.1 Number of data recorded along each line used in 3—D analysis.Final column shows calculated uncertainties using method of section 2.5 and estimateduncertainties taking into account other error sources (see section 5.4.2).Line Line Line Line Line Line1 2 3 4 5 6144where 9 is the ray take-off angle, and v1 and v2 are the velocities in layer 1 and beneathlayer 1. Corrections for each phase were made assuming a constant v2 and 0. v1 valueswere obtained from the 2—D in-line models.zo12Figure 5.2 Illustration of elevation correction applied to data used in 3—D analysis. Receiver (solidtriangle) is moved from the surface (depth Zo) to the base of layer 1 (depth zi; opentriangle). 9 is the ray take-off angle. The correction is given by equation 5.1.5.3 Description of Fan-Shot and Other DataRecord sections for the principal fan shots (i.e., SPs 1, 3, 5, 4, 7, 12 and 17 recordedalong lines 1, 2 and 3) are shown in Figs. A.1—A.12. Data quality, as on the in-line recordsections, is variable. Not all phases are present or recognizable on all sections because oflow signal-to-noise ratio or inappropriate offsets. Also, in some cases arrivals which maybe discernible were not used because of very large uncertainties in the pick times. Otherarrivals not confidently identified as Pg, or P were not used.5.3.1 Corner Shots (SPs 1, 3 and 5)First arrivals across the three corner shot sections (Figs. A.1—A.3) are P with arrivaltimes typically of —7 seconds. The arrivals are emergent across most of the line 2 (SP 3;Fig. A.2) and line 1 (SP 5; Fig. A.3) sections. Exceptions are the far-offset traces on bothsections (27O km at the north end of line 1 and 24O km at the northeast end of line 2).P arrivals across the line 3 (SP 1; Fig. A.1) section, which has a minimum offset of 310km. are comparatively more impulsive. The noisy traces between 2030_2100 azimuth on thissection correspond to receiver sites located in the metropolitan Vancouver area.PmP is clearly evident on the lines 1 and 2 record sections. It is most coherent on line 1where it appears between 8.5—10 s across most of the section. It disappears at larger offsets145(<48° azimuth). PmP on line 2 is less coherent. It arrives at 8—9 seconds between 280—340°azimuth, but again disappears for larger offsets (340° azimuth). PmP is not observed online 3 probably because of the large minimum offset (310 km) and high noise level.5.3.2 Shots at Line Midpoints (SPs 4, 12 and 17)Each of the shots were recorded along two lines: SP 4 along lines 1 and 2 (Figs. A.4 andA.5), SP 12 along lines 1 and 3 (Figs. A.6 and A.7), and SP 17 along lines 2 and 3 (Figs.A.8 and A.9). First arrivals across the sections are a mixture of Pg and P, both of which arepresent on all sections. The distinction between the two is not definitive in the vicinity ofthe PgPn crossover point, which on the in-line sections typically occurs at -.-170—180 km.In general, first arrivals greater than about 7 s are considered to be P. First arrivals can befollowed across the sections, though not necessarily from trace-to-trace. There are some shortgaps where picks could not be made because of high noise levels, e.g., line 3—SP 17 (Fig.A.9) between 222°—236° azimuth. PmP arrivals are observed on all sections but are mostcoherent on line 1—SP 12 (Fig. A.6) where they can be followed across the entire section.5.3.3 Centre Shot (SP 7)This shot was recorded along all three lines and, overall, produced the best signal-to-noise ratio of all the fan shots (Figs. A. 10—A. 12). First arrivals are clearly visible across allthree sections, comprising Pg and P, except on line 3 (Fig. A.12) where P is absent becauseof insufficient shot-receiver offset. PmP arrivals are also present on all the sections but aremost coherent on line 1 (Fig. A. 10) where they can be followed across the entire section,including the northern end of the line where, at offsets of ‘250 km. they arrive at s9•5 S.5.3.4 Lines 4, 5 and 6The three short lines interior to the triangle recorded data from all fan shots except SP 5which was recorded only along line 5. In addition, lines 4 and 6 recorded data from SP 20.Line 4 also recorded SP 21. Data from these small (200 kg) shots have a poor signal-to-noise146ratio relative to the larger fan shots. The dataset from lines 5 and 6 contribute only a smallnumber of arrival times to the overall inversion dataset (see Table 5.1). A more substantialnumber of Pg, PmP and P arrival times are obtained from the eight line 4 record sections(Figs. A.13—A.16). The in-line record sections for the two end shots (SPs 7 and 17; Fig.A.13) contain clear Pg arrivals. PmP is observed on SP 7 but because of the small maximumoffset of these sections (80 km) only a small number of picks were possible. A simple 2—Dupper crustal model for line 4, based on an inversion of Pg arrivals from SPs 7 and 17 usingthe method described in section 2.5, is shown in Fig. 5.3.P is observed clearly across the SP 1 record section and at the north-east end of SP 4.First arrivals on the remaining sections are Pg which appear most clearly on SP 12. PmParrivals are present on all sections (except SP 1) and are most coherent on SP 12 between800_1000 azimuth.5.4 Modelling the 3—D Data5.4.1 Starting ModelThe starting velocity model was constructed from the three 2—D in-line profiles for lines1, 2 and 3 as described in section 4.5.1. The starting Moho depth model was also constructedfrom the three in-line profiles, i.e., a set of points representing depth to Moho along each line(in regions constrained by ray coverage) were gridded and smoothed. Figure 5.4a shows thestarting Moho model and the location of depth points used to construct the model. Prior tothe inversion of PmP traveltimes for both lower crustal velocities and Moho depth, velocitiesbelow the Moho were stripped off and replaced with velocities immediately above the Mohoextended downwards with a zero gradient. This ensures that the first arrival traveltimes atthe reflecting surface are from the downgoing wave and not from waves that have turnedbeneath the reflector (see section 4.4).147I I I0 20 40 60 80 100DISTANCE (km)I I I I I I I3.0 4.3 5.6 5.7 5.8 5.9 6.0 6.1 6.2 6.3velocity (km/s)Figure 5.3 Modelling results for line 4 from inversion of 119 Pg and P arrivals. (a) Two-point raytracing diagram. (b) Comparison of observed (short vertical lines with height equal to the uncertaintyof the pick) and calculated (dots) traveltimes. Rms traveltime residual is 51 ms; normalized x2 is4.1. (c) Velocity model. White region around border is unconstrained by ray coverage.rQ., ‘..,t,VI’7(b)ci.SP 1760 20 40 60 80distance (km)(c)148350X(km) X(km)• • I I 1I’r—T—.31.8 32.4 33.0 33.5 34.1 34.7 35.3 35.8 36.4 37.0 10.5 11.4 12.3 13.3 14.2 15.1 16.0 17.0 17.9 18.8reflector depth (km) boundary depth (km)Figure 5.4 (a) Starting Moho depth model for 3—D inversions. 3—D surface was constructed fromthe three 2—D Moho depth models for lines 1, 2 and 3. Heavy broken lines representlocation of constraint points along the 2—D profiles that were gridded to derive startingmodel. (b) Upper-lower crust boundary defined by maximum depth penetration ofupper crustal rays. 1MB, Intermontane Belt; CB, Coast Belt; FRF, Fraser River Fault;HF, Harrison Fault; V, Vancouver. A, shot points; black and white lines representapproximate location of receiver sites (see Fig. 5.1 for exact receiver site locations).As in the synthetic tests of section 4.5.1, the finite-difference model dimensions arefl X fly x n = 221 x 301 x 56 with a grid node spacing of 1.2 km (Fig. 4.9), or 264 x 360 x 66km. Dimensions of the Moho model are n x n, = 221 x 301 (264 x 360 kin).5.4.2 Modelling Upper Crustal VelocitiesAs described in section 4.3 the slowness perturbations were smoothed at two differentstages of each iteration. The dimensions of the smoothing operators were chosen to allowa good fit to the data (i.e., approximately to within the estimated average uncertainties ofthe data) and avoid excessive smearing of the velocity anomalies along the ray paths. The(a)300250200>-1501000 F’n ii,’0 50 100 150 200 C) 50 10 150 200 250149average calculated uncertainty of the traveltime picks is 55 ms (Table 5.1). This estimate doesnot include errors associated with site mislocations, non-linear instrument clock drift, andelevation corrections that were applied to strip off the transit times through the 1—2 km thicklow velocity near surface layer. Including these effects, a more reasonable estimate of theaverage uncertainty is 75 ms. Table 5.1 also lists more reasonable estimates for the averageuncertainties of P and PmP picks. A large range of smoothing operator dimensions weretested, including fixing the size for all iterations and reducing the size for higher iterationnumbers.In the first smoothing operation (gridding) of the final inversion the slowness perturbationat each grid node was calculated as the average of all the öu terms in a 20 x 20 x 2 cubicgrid cell volume (24 x 24 x 2.4 km3) centered at the grid node. In the second operation thegridded Su terms were smoothed with a 21 x 21 x 3—point MA filter.A total of 3963 Pg traveltimes (Table 5.1) were used in the inversion. The rms traveltimeresidual for the starting model was 186 ms. Figure 5.5 shows the misfit plotted againstiteration number for ten iterations. After the first few iterations the primary features of thefinal model are present and the rms traveltime residual levels off. With subsequent iterationsthe velocity model is only slightly refined and the residual decreases very slowly. Afterten iterations the misfit was 88 ms. This was considered close enough to the estimatedaverage uncertainty of 75 ms and the procedure was stopped. Many more iterations wouldbe required to reduce the error level to 75 ms and would result in only minor changes tothe velocity model. Table 5.2 lists the starting and final (after ten iterations) rms traveltimeresiduals for each shot point. Figure 5.6 displays the starting and final misfits for the sevenlarge shot points accounting for about 60% of the data. Figure 5.4b shows the upper-lowercrust boundary constructed using the method described in section 4.7. Velocities above thisboundary were fixed in subsequent inversions.1500.200.05111111 I I ill0 2 4 6 8 10Iterat.ion numberFigure 5.5 Rms traveltime residual versus iteration number for modelling of upper crustal velocities.Table 5.2 Starting and final (after ten iterations) rms traveltime residuals for eachshot for the inversion of Pg. N, number of data. Bottom row shows overallstarting and final rms misfit and total number of data.starting finalshotpoint misfit (ms) misfit (ms) N1 127 95 1623 206 103 2954 149 79 3995 118 51 817 268 89 56012 272 98 50417 150 84 4062 83 78 1776 94 92 11111 83 44 7613 145 120 4714 153 86 7015 241 119 6716 87 64 18418 95 84 16519 104 98 13720 143 86 12921 85 81 18546 154 69 7847 98 74 7548 302 92 55186 88 39635.4.3 Modelling Deeper StructureThe first stage of the quasi-simultaneous inversion was the inversion of PmP traveltimes151Figure 5.6 Example misfits from the inversion of Pg traveltimes for upper crustal velocities.Starting (a) and final (b) (after ten iterations) traveltime misfits for the seven large shot points.Spikes along the track for each shot point represent the misfit at each recording site along lines 1, 3and 2. “Station” numbers 1—277 are sites along line 1 from north to south; 278—488 are sites alongline 3 from east to west; 489—7 16 are sites along line 2 from southwest to northeast. The rmstraveltime residual for each shot are given along the upper left axis (see also Table 5.2).<C.4-Ic:3c(a)<C.(b)152for lower crustal velocities. In the first smoothing operation the slowness perturbation at eachgrid node was calculated as the average of all the 5u terms in a 30 x 30 x 4 cubic grid cellvolume (36 x 36 x 4.8 km3) centered at the grid node. In the second operation the gridded6u terms were smoothed with a 31 x 31 x 5—point MA filter. The larger smoothing operatorsfor the lower crust reflect the sparser ray coverage and poorer vertical resolution suppliedby the PmP traveltime constraints. A total of 1903 PmP traveltimes (Table 5.1) were used inthe inversion. The rms traveltime residual for the starting model was 254 ms. Figure 5.7adisplays the starting misfits for the seven large shot points (accounting for 86% of the data).In each step of the simultaneous inversion of PmP and P traveltimes for Moho depth thereflector perturbation model was smoothed with a 21 x 21—point (24 x 24 km2) MA filter.A combined total of 3071 PP and P traveltimes were used in the inversion (Table 5.1).The rms traveltime residual for the starting model, which includes the updated lower crustalvelocities from the first stage of the quasi-simultaneous inversion procedure, was 193 ms.The third stage of the quasi-simultaneous inversion was the inversion of P traveltimesfor upper mantle velocities. Smoothing operator sizes were 30 x 30 x 2 (36 x 36 x 2.4 km3)and 31 x 31 x 3 for the two smoothing operations. A total of 1168 P data were availablefor the inversion; however, because the raypaths for some rays did not cross the Moho, notquite all of the data were used in each iteration. The rms traveltime residual for the startingmodel, which includes the updated lower crustal velocities and Moho depth model from thefirst two stages, was 186 ms for 1115 data. Figure 5.8a shows the starting misfits for theseven large shot points accounting for 89% of the data.After three iterations of the quasi-simultaneous inversion procedure, the rms traveltimeresiduals were 135 ms and 116 ms for PP and P, respectively. Further iterations did notdecrease the residuals. Since the residual for PmP was significantly larger than the estimatedaverage uncertainty (110 ms; Table 5.1), the velocity-Moho model after three iterations wasused as a starting model for the inversion of PmP for lower crustal velocities (keeping theMoho fixed). Smoothing operator dimensions were not changed. After ten iterations the153rFigure 5.7 Example misfits from the inversion of PmP traveltimes for lower crustal velocities (anddepth to Moho). Starting (a) and final (b) (after three iterations of the quasi-simultaneous inversionprocedure and ten iterations in which PmP were inverted for velocities only) traveltime misfitsfor the seven large shot points. See Fig. 5.6 for other plotting information.(a)4(b)1544-Figure 5.8 Example misfits from the inversion of P traveltimes for upper mantle velocities (anddepth to Moho). Starting (a) and final (b) (after three iterations of the quasi-simultaneous inversionprocedure and six iterations in which P were inverted for velocities only) traveltime misfitsfor the seven large shot points. See Fig. 5.6 for other plotting information..24.(a)4.(b)155Table 5.3 Starting and final (after ten iterations) rms traveltime residuals foreach shot for the inversion of PP. N, number of data. Bottom row showsoverall starting and final rms misfit and total number of data.starting finalshotpoint misfit (ms) misfit (ms) N1 318 88 853 160 101 2394 149 99 1535 367 107 1117 272 103 35912 239 114 34117 284 127 3462 314 109 7415 69 44 2716 238 123 7318 413 120 1519 204 65 1720 191 105 1421 109 80 3948 215 112 10254 109 1903rms traveltime residual was 109 ms. The final PmP misfits for the seven large shot pointsare shown in Fig. 5.7b.The “fine-tuning” of the lower crustal velocities degraded the rms traveltime residual ofthe P data to 125 ms, requiring further modelling of P traveltimes. After six iterationsthe residual was 98 ms. The final P misfits for the seven large shot points are shown inFig. 5.8b. Tables 5.3 and 5.4 list the starting and final rms traveltime residuals for eachshot point for PmP and P, respectively.5.5 Model Resolution and UncertaintyOther inversion techniques, e.g., least squares, provide information on model resolutionand parameter uncertainties through a model resolution matrix and covariance matrix. Theprocedure used here does not provide a quantitative measure of resolution since these matricesare not available. Nevertheless, it is possible to compute resolution kernels (for the linearized156Table 5.4 Starting and final (after six iterations) rms traveltime residuals for eachshot for the inversion of P. N, number of data. Bottom row shows overallstarting and final mis misfit and total number of data.starting finalshotpoint misfit (ms) misfit (ms) N1 211 110 3833 147 91 2234 220 83 855 255 100 1367 131 88 7812 112 96 6317 138 93 712 184 103 276 137 64 3318 144 95 69186 98 1168problem) which illustrate the degree to which the model at a grid node is derived from the truemodel. Hole (1992) describes the calculation of resolution kernels and presents an examplefor a synthetic 3—D survey showing the streaking of the resolution kernels along rays whichpenetrate a small box (two cubic grid cells) centered at the grid node. Unfortunately, theresolution kernels are time-consuming to calculate and difficult to display and do not offersubstantial insight into the overall resolution of a model.A more useful, though qualitative, estimate of resolution and uncertainty can be gaugedfrom synthetic tests, using different starting models (see section 5.6.1, p. 181), varyingthe spatial smoothing operator sizes and looking at ray coverage (Hole et al. 1993a). Themost important factors governing spatial resolution are survey geometry and size of thesmoothing operators. Estimates for absolute uncertainty must also take systematic errorsources (discussed in section 5.4.2) into account. As pointed out by Hole et al. (1993a),relative differences in velocity between neighboring regions are likely to be better resolvedbecause they are less sensitive to systematic errors.The resolving power of the data can be estimated from synthetic tests using a checker157board velocity model (Hearn and Ni 1994). In these tests, noise-free synthetic data aregenerated using a velocity model comprising (in plan view) alternating high and low velocitysquares and a linear vertical velocity gradient. Data are calculated for the same source-receiver combinations as used in the real analysis and are inverted using a 1—D startingvelocity model. Results for a test using Pg arrival times and two different true models areshown in Fig. 5.9. In both cases the squares were assigned alternating velocities of 6.0and 6.3 km/s at the top of the model, beginning with 6.0 km/s in the southwestern corner.Velocities were extended downward with a gradient of 0.02 si. In the first model squareswere 25 km x 25 km, and in the second were 50 km x 50 km. The starting 1—D model inboth cases began at 6.15 km/s at the top, extended down with a gradient of 0.02 s_i. At thedepth of the slices shown in Fig. 5.9 the velocities of the true models are 6.16 and 6.46 km/s.Figure 5.9 Resolution tests for Pg using checkerboard velocity models with squaresof dimension (a) 25 km x 25 km, and (b) 50 km x 50 km (indicated by the whitegrid lines). At the depth of the slices shown the true velocities alternate between 6.16and 6.46 km/s, beginning at 6.16 km/s in the lower left square.__200>-6.10 6.15 6.20 6.25 6.30 6.35 6.40 6.45 6.50 6.55 6.10 6.15 6.20 6.25 6.30 6.35 6.40 6.45 6.50 6.55velocity (km/s) velocity (km/s)158Fig. 5.9a shows the results for the model with 25 km x 25 km squares obtained byusing a 10 x 10 x 2 cubic grid cell smoothing volume to grid the 5u terms, followed byone pass of an 11 x 11 x 3—point MA filter to smooth the gridded 6u terms. A range ofsmoothing operator sizes were tried, however, these values gave the optimal results for thiscase. The results show that the data are not capable of resolving 25 km x 25 km featuresthroughout the study area.Results for the model with 50 km x 50 km squares (Fig. 5.9b) were obtained using a 20x 20 x 2 cubic grid cell smoothing volume followed by one pass of a 21 x 21 x 3—point MAfilter. In general, the alternating pattern of low-to-high velocities is well resolved throughoutthe middle of the triangle. Although velocities in each cell vary by up to 0.3 km/s, the peakvalues are generally in error by only 0.05 km/s. The smoothing operator dimensions used inthis test are identical to those used in the real data analysis (see section 5.4.2) and thus it islikely that upper crustal velocities imaged by the real data represent an average of the truePg velocities over a region roughly 50 km x 50 km. Table 5.5 lists the estimated spatialresolution and absolute model uncertainty for regions of the model that are constrained byray coverage. The estimates are based on the results of this example and other synthetic tests.Table 5.5 Estimated lateral and vertical resolution and absolute model uncertainty. Absoluteuncertainties apply to velocities at a specific node in a region constrained by ray coverage.Estimated lateral Estimated vertical Estimated absoluteRegion resolution (km) resolution (km) uncertaintyupper crust 50 4.8 0.3 km/slower crust 60 9.6 0.35 km/supper mantle 60 4.8 0.3 km/sMoho 25 n.a. 3 km5.6 ResultsHorizontal slices through the final 3—D velocity model are shown in Figs. 5.10 and 5.11.159Only those portions of the velocity model considered to be reasonably well constrained areshown. Uncoloured regions represent grid node locations that are further than one half thelateral smoothing operator size from a ray. Although not shown, the velocity model doesextend smoothly into these unconstrained regions as indicated by the velocity contours.The slices show a fairly high degree of lateral inhomogeneity. Table 5.6 lists the averagevelocity within the study area and standard deviation for the eight slices of Figs. 5.10 and5.11. The most prominent feature, in both size and magnitude, is the zone of relatively highcrustal velocity in the south-central region of the model. This feature is present, thoughsmall, at 2.8 and 7.6 km depth centered at (x,y) = (125,30) km, with velocities of 0.3—0.4km/s greater than the mean at these depths. At deeper depths the dimensions of this featureincrease. Deep in the lower crust it comprises a rectangular zone in the southwest cornerof the model extending eastwards to the vicinity of the Fraser River Fault. To the west itextends to at least x=50 km; the data do not constrain the deep structure west of this.The principal feature of the slice at 37.6 km (Fig. 5.1 id) is the relatively high (8.0km/s) upper mantle velocities beneath the southern half of line 2 and at the southern end ofline 1, and relatively low (7.7 km/s) velocities beneath most of line 3 and near the north endof line 1. The lower crustal velocities (7.0 km/s) in the Coast Belt are present in Fig. 5.lldbecause this slice cuts through the Moho. P velocity throughout most of the interior of thetriangle is between 7.7 and 8.0 km/s. The average P velocity at this depth is 7.85 km/s.A set of l—D velocity profiles, taken at the locations shown in Fig. 5.12, is shown inFig. 5.13a-k along with average 1—D velocity model for the study area (Fig. 5.131). Theprofiles illustrate the range of crustal and upper mantle variation throughout the study area.The average l—D velocity model is very simple: crustal velocities increase almost linearlywith a gradient of 0.02 s_i from 6.0 km/s near the surface to 6.7 km/s at the base of thecrust (34.5 km depth). The average upper mantle velocity increases from 7.8 to 7.9 km/s at42.5 km depth with a gradient of 0.015 s_i.1605.80 6.01 6.22 6.43 6.64 6.85 7.05 7.26 7.47 7.68 7.89 8.10velocity (km/s)Figure 5.10 Horizontal slices through the final 3—D velocity model at depths of (a) 2.8 1cm,(b) 7.6 km, (c) 12.4 km (through the upper-lower crust boundary) and (d) 18.4 km.Unconstrained regions are shown as uncoloured. CB, Coast Belt; 1MB, IntermontaneBelt; FRF, Fraser River Fault; HF, Harrison Fault; V, Vancouver._—. 200E>-P—. 200E>-7 / 1100 150X(km)150x acm)161350300250—200E>-1501005007.47 7.68 7.89 8.10Figure 5.11 Horizontal slices through the final 3—D velocity model at depthsof (a) 23.2 km, (b) 28.0 km, (c) 32.8 km and (d) 37.6 km. Slices (c) and(d) cut through the Moho. See Fig. 5.10 for other information.150X (km) X (km)I I I I I I5.80 6.01 6.22 6.43 6.64 6.85 7.05 7.25velocity (km/s)162Table 5.6 Average velocity within study area and standard deviation at depths of slices shown inFigs. 5.10 and 5.11. Average velocity shown for 32.8 km is average of crustal velocities only.Average velocity shown for 37.6 km is average of upper mantle velocities only.StandardAve. Velocity DeviationDepth (km) (kmls) (kmls)2.8 6.18 0.097.6 6.23 0.0912.4 6.34 0.1018.4 6.42 0.1323.2 6.56 0.1528.0 6.67 0.1232.8 6.71 0.1337.6 7.85 0.08Figure 5.14 shows velocities at the base of the upper crust, top and bottom of the lowercrust, and top of the upper mantle. These are not horizontal slices, rather they are velocitiesimmediately above and below the upper-lower crust boundary and Moho. Note that velocitiesare very poorly constrained at the base of the upper crust (Fig. 5.14a). This is due to themethod used to construct the upper-lower crust boundary (see section 4.7) which, because ofthe smoothing, tended to over-estimate the maximum turning depth of the Pg rays at manylocations. The principal features seen in the horizontal slices are also seen in Fig. 5.14. Atthe top of the lower crust velocities are lowest in the north and highest in the southwest.A second small high velocity anomaly is located near the line 2—line 10 intersection point.Velocities at the base of the lower crust display a very complicated pattern with valuesranging from 6.4 to 7.0 km/s. A dominant feature is the relatively low velocity (—‘6.6 kim’s)zone subparallel to the CB-IMB boundary (Pasayten Fault) near the south-eastern corner ofthe triangle. This feature is also observed on the horizontal slices at 28 and 32.8 km depth1633503002502001501005000 50 100 150 200 250X (km)Figure 5.12 Location of 1—D velocity profiles shown in Fig. 5.13. Heavy black line representslocation of vertical slice in Fig. 5.17. Shaded region illustrates the width of the swath usedto construct the ray density plot of Fig. 5. 17d (and other ray density plots).(Figs. 5.llb and c). Anomalously low velocities at the top of the upper mantle (‘—‘7.6 kmls)occur near the north end of line 1 and between lines 2 and 4 at —‘(150,150).An indication of how the ray coverage varies with depth is shown in Fig. 5.15.Percentage-wise, the best coverage occurs near the top of the upper crust, followed bythe top of the upper mantle. Coverage is poorest at the base of the upper crust and inthe lower crust. Although significantly more PP rays propagate through the lower crustin comparison to P rays through the upper mantle, the PmP ray paths are much closer tovertical and thus penetrate only a few cells at each level in the model. The P raypaths aremore nearly horizontal and each ray penetrates many cells at the same depth level.164-10o]\’’d’—1 00]\)cI1-20 -20-30 -30-400 1) ao1N_%10 -102--20 -20.1-ICl)D-30 -30#-40 -40I1:h7e- -30 -30 a4-40-4080-10678-20-30-4000--20-30-E -40 -0-10 H-20 --40g678I6781-F86 7I....102-.C.:.::..cia)V-3040I678-10 ave-20-30-20-30-406---CB-IMB8-408velocity (kmls)Figure 5.13 (a—k) 1—D velocity profiles at the locations shown in Fig. 5.12. The curves areshown as continuous but are not necessarily constrained at all depths (e.g., velocityreversals in the upper mantle). (1) Average 1—D velocity profile for study area (solidline), for the Coast Belt (CB) and the Intermontane Belt (1MB).165350300250— 200E>-150100500350300250_—_200E>-1501005005.80 6.01 6.22 643 6.64 6.85 7.05 7.26 7.47 7.68 7.89 8.10velocity (km/s)Figure 5.14 Velocities at (a) base of upper crust (immediately above upper-lower crustboundary) (b) top of lower crust (immediately below upper-lower crust boundary)(c) base of lower crust (immediately above Moho) (d) top of upper mantle(immediately below Moho) See Fig. 5.10 for other information.0 50 100 150 200 250 0 50 100 150 200 250X(km) X(km)166010Upper-lowercrust boundarya20-a.:*j..::.::———————‘,..— Moho400 10 20 3040 50 60% st.udy area sampledFigure 5.15 Percentage of the study area sampled by rays plotted against depth. Percentage iscalculated as the number of cells penetrated by at least one ray divided by the total number ofcells. Study area is defined as triangle formed by SPs 1, 3 and 5. Shaded regionsshow the range in variation of the upper-lower crust boundary and Moho.Figure 5.16a shows the final Moho depth model. Uncoloured regions represent gridnodes further than 12 km (one half the smoothing operator size) from a point in Fig. 5.16c.Although not shown, the reflector model does extend smoothly into these unconstrainedregions (as indicated by the contours). Depth is converted to two-way vertical traveltimein Fig. 5.16b. Figure 5.16c shows the location of the PmP reflection points and P Mohointersection points constraining depth to Moho. The points are most dense east of the CB1MB boundary and are relatively sparse west of the Harrison Fault and along line 3. Overall,however, the coverage is good and provides a well defined Moho image. The principalfeature of the Moho model is the greater depth beneath much of the Coast Belt. Depthto Moho throughout most of the Intermontane Belt is 33—36 km. The minimum stronglyconstrained depth is s32 km near the north end of line 1. In contrast, depths are greater than36 km beneath most of the Coast Belt. Exceptions are the northernmost part of the Coast Beltin the study area, the region in the south that is (roughly) east of the Fraser River Fault, and30167the southwestern corner of the triangle where the data suggest a thinning of the crust to .-..‘33km. The maximum (strongly constrained) depth is s39 km in the south-central Coast Belt.The map of TWTT to the Moho in Fig. 5.16b shows many of the same features as thedepth map. Throughout most of the Intermontane Belt the TWTT varies between 10.6—11.4s. TWTT in most of the Coast Belt is 11.6—12.2 s with the same exceptions noted above.Four Lithoprobe reflection lines (88—11, 12, 15, 18) overlie regions where the TWTT is fairlywell constrained. Four other lines (88—13, 14, 16, 17) are near (<50 km) the well constrainedarea. Table 5.7 shows a comparison of TWTTs from Fig. 5.16b and the interpreted reflectionsections of Cook et al. (1992) and Varsek et al. (1993). Note that differences in the valuesmay be associated with differences in the reflection Moho, which is placed at the base of azone of reflectivity, and the refraction Moho. Overall the TWTTs agree fairly well, althoughthe detailed structure (e.g., dip direction) is not always consistent. The poorest agreementis along line 88—11 with a discrepancy of -‘0.6 s. The best agreement is along line 88—18,which crosses the Fraser River near SP 7. The TWTTs agree very closely and the directionof dip on the Moho from east to west is consistent. Based on an extrapolation of the TWTI’contours in the southwestern corner of the model, the estimate of 10—12 s for the Moho timedepth beneath line 88—16 by Varsek et al. (1993) seems reasonable. The contours suggesta thickening of the crust to the north.Figure 5.16d shows the standard deviation of the dz terms used in the final iteration.The values were obtained by dividing the region into a grid of 6 x 6 km cells and takingthe standard deviation of the dz terms lying within each cell, provided the cell containedmore than one term. The standard deviation is a rough measure of the consistency of thedata, small values indicating that the data within a cell require a similar depth perturbation.The values are also a measure of the uncertainty in the reflector depth from errors in thedata. The true uncertainty will be much larger because of errors in the velocities above andbelow the Moho. The majority of the standard deviations are less than 1 km and there is noobvious trend in the distribution of the values.168350E>_150200E>-150Figure 5.16 (a) Depth to Moho. (b) Two-way vertical traveltime to Moho from ahorizontal plane at 1 km above sea level. Purple lines show location of Lithoprobereflection lines (88—11 — 88—18). (c) Location of PmP reflection points (blue) and PMoho intersection points (yellow) constraining solution in (a). (d) Standard deviation ofdz terms in 6x 6 km cells. See Fig. 5.10 for other information.300250-I00310 32.1 33.2 34.4 355 366 37,8 389 400reflector depth (km)350300250990 18.22 1U55108711201152118512.17 12.50two.way traveltime (s)100150Xtkrn)150X(km)169Table £7 Comparison of 3—D model and reflection-interpreted TWTI’ to Moho. Times in bracesare taken from 1—D starting model results (section 5.6.1). Times are measured to same datum (1km). TWTTs for line 88—11 are from Cook Ct al. (1992); others are from Varsek et al.(1993). Moho, where identified was placed at the base of a zone of reflectivity.Range of TWTRange of TWTT from reflectionLine from 3-D model (s) interpretations (s) Comments on reflection interpretations88-11 11.0 - 11.4 10.5- 10.8 Moho well defined; gently dipping from west to east(11.1-11.3)88-12 11.2 - 11.5 10.8 - 11.6 West side down moho ramp or step; Moho well defined(11.2-11.4) at eastern end88-18 11.2 - 12.0 11.2 - 11.8 West side down Moho ramp or step; Moho fairly well(11.3-11.8) defined across section88-13 11.3- 11.8? 11.2- 12.0 Fairly sharp decrease in time depth at sw end of line;(10.4-11.0) Moho very well defined across most of sectionmostlyunconstrained88-17 11.8? 11.4 - 11.7 Gentle westward shallowing of Moho; fairly well(10.8-11.2) constrained across sectionunconstrained88-14 11.7? - 11.8? 11.2? - 11.5? Moho poorly defined(10.6-10.8)unconstrained88-15 11.6 - 12.0 ? No good evidence for Moho(10.5-11.1)88-16 10.4? - 11.6? 10 - 12 Moho not well imaged/difficult to interpret; Varsek et(10.0?-10.2?) al. (1993) describe possible trajectoriesunconstrainedFigure 5.17a shows a vertical slice through the starting velocity model along a north-south profile through the middle of the study area (Fig. 5.12). The final model is shown inFig. 5.17b, with unconstrained regions uncoloured. Note the large unconstrained zones inthe upper crust near the middle of the model and the middle crust at the south end. The maindifferences between the starting and final model include significantly higher lower crustalvelocities in the middle of the final model and lower velocities in the lower crust north of170200 km. The final model shows a zone of relatively low lower crustal velocity beneath theCB-IMB boundary. Otherwise, velocities in the lower crust below 20—25 km are significantlyhigher south of 200 km. Depth to Moho increases from 33 km in the north to 38 km at thesouth end. Figure 5.17c shows the difference between the final and starting model. The rangeof velocity perturbations seen on this slice (±0.2 km/s) is typical of the other slices presentedbelow. Figure 5.17d is a representation of the ray density along a swath 24 km wide, i.e., thelateral dimensions of the smoothing operator used for modelling the upper crust, centered onthe profile (see Fig. 5.12). Coverage in the upper crust is fairly poor, although where thereis coverage the density is fairly high. Coverage in the lower crust is better but, in general,the ray density is lower. This slice has good upper mantle ray coverage.Vertical slices along profiles close to lines 1, 2 and 3 are shown in Fig. 5.18. Thecorresponding ray density plots are shown in Fig. 5.19. Ray coverage is very poor at thebase of the upper crust along the slice near line 1 (Fig. 5.19a). Coverage throughout the restof the model is very good. There are many gaps in coverage along the slice near line 2 (Fig.5.19b), the largest occurring in the lower crust southwest of 175 km. The slice near line 3is fairly well constrained throughout, with gaps occurring at the base of the upper crust atthe east end and at the base of the lower crust between 90—140 km (Fig. 5.18c).The slices in Figs. 5.18a and c can be compared to the final models for lines 1 and 3determined from 2—D ray-trace based inversion and forward modelling shown in Figs. 2.19and 3.9. Any comparison, however, must consider the very different modelling approachesused. The 2—D modelling included a greater number of phases (primarily more reflectionphases) and incorporated amplitude information. The 2—D models are parameterized in termsof distinct layers which allow for significant velocity discontinuities across layer boundariesand permit detailed modelling of structure along boundaries. Velocities between nodes aredetermined by linear interpolation, and thus for deeper structure with widely spaced nodesthe models can appear quite simple. (Figure 2.1 8a shows the parameterization used for line1, including all boundary and velocity nodes solved for by inversion and forward modelling.)17101020a -30-D40-c 20 -0.a)-400-j-10j-0cD_-D-40-0.30 -0.23 -0.15 -0.08 0.00 0.08i velocity (km/s)0.c -20-0a)-30--D-400.15 0.23 0.3013 17 21number of raysFigure 5.17 (a) Vertical slice through starting model along profile through SPs I and 4 (see Fig.5.12 for location). (b) Vertical slice through final model. See Fig. 5.10 for colour velocity scale. (c)Difference between final and starting models. (d) Ray density near the slice. Plot shows count of raysin swath shown in Fig. 5.12. CB-IMB, Coast Belt-Intermontane Belt boundary (Fraser River Fault).0 -200distance (km)I I I1 5 9 25 29 331721020-40--20c, 30V-40Figure 5.18 (a) Vertical slice through a profile near line 1. (b) Vertical slice through a profile near line2. (c) Vertical slice through a profile near line 3. Nomogram at bottom right shows locations of slicesin panels a—c. See Fig. 5.10 for colour velocity scale. FRF, Fraser River Fault; HF, Harrison Fault.-40distance (km)173-20-G) -30--40 -- I•._.-•i. •_IrL4-: ?____50 100 150I I I I I I3 5 7 9 11 13I I I15 17 19number of raysFigure 5.19 Ray density plots for (a) slice in Fig. 5.18a, (b) slice in Fig. 5.18b,and (c) slice in Fig. 5.1 8c. See Fig. 5.18 for other information.i .1SW HF 08-1MBI- IiNE0-20-ciG) -30-V-40-0-20-cici -30--40-W200HF FRF250 300E50 100 150 200distance (km)1The 3—i) modelling technique allows velocity discontinuities to occur only across a somewhatarbitrarily located upper-lower crust boundary and the Moho. Even with the large lateralsmoothing operators used, the 3—D models can display more detailed velocity variationsbecause of the simple parameterizations used in the 2—D modelling. It is important to174consider, however, the differences in resolution between the 2— and 3—D models (see Tables2.3, 3.3, and 5.5).The differences between the 2—D and 3—D models are most apparent in a plot of thedifference between models (not shown). A comparison of the results for line 1 shows thatthe major differences are higher velocities (by --0. 1—0.25 km/s) throughout the lower crustof the 3—D model between 90—180 km, and lower velocities (by —0.1—0.3 km/s) at the baseof the lower crust north of 200 km and at the top of the upper crust north of 260 km.The differences between the models for line 3 are much smaller. The most significantdifferences occur in unconstrained portions of the 3—D model at the base of the lower crust.Velocities at the top of the lower crust between 170—200 km in the 3—D model are --0.1km/s lower than the 2—D model, enhancing the already significant difference in velocitiesacross the Harrison Fault.The most significant differences between the line 2 model in Fig. 5.18b and the modelof McLean (1994) are higher velocities in the middle crust between 90—150 km resulting inthe pull-up of the 6.5 and 6.6 km/s contours beneath the Harrison Fault, and deeper Moho(by 1—3 km) throughout most of the Coast Belt.Figure 5.20b—d shows slices through three near-parallel east-west profiles separated by-.‘40 km; ray density plots are shown in Fig. 5.21b—d. These slices, together with Fig. 5.18chighlight cross-strike variations in velocity across the CB-IMB boundary and Harrison Fault.The model in Fig. 5.20d shows many of the same features seen in Fig. 5.18c, 25 km to thesouth, e.g., crustal velocities to a depth of --20 km are lower to the east of the Harrison Fault.East of the CB-IMB boundary (Pasayten Fault) upper crustal velocities increase to 6.3—6.4km/s whereas lower crustal velocities decrease to 6.5 km/s. The Moho on this slice shallowsrapidly to the east, beginning at the Harrison Fault, from 39 to 35 km depth. To the north,along the slice shown in Fig. 5.20c, lower crustal velocities are significantly lower (<6.7kmls) and show a similar pattern of decreasing to the east of the CB-IMB boundary. Upper175crustal velocities between the Harrison Fault and CB-IMB boundary are also relatively lowalong this slice. Further north, along the slice shown in Fig. 5.20b, the CB-IMB boundary(Fraser River Fault) is associated with a depression of the 6.7 km/s contour; lower crustalvelocities increase to both the east and west of this boundary and reach values higher thanpresent along the slice in Fig. 5.20c. Mid-crustal velocities (between 10—20 km depth) arerelatively high east of the CB-IMB boundary.Figure 5.20a is a crooked line slice that follows three Lithoprobe reflection profiles(88—13, —12 and —11; see Fig. 5.1 6b for locations). Note the lack of constraints in the lowercrust beneath line 88—13, in the upper crust beneath lines 88—12 and 88—11, and beneath theentire eastern 15 km of 88—11. This slice indicates that upper crustal velocities are higherbeneath 88—12 and that deeper crustal velocity structure does not vary significantly beneaththe three reflection lines. Upper mantle velocity is highest beneath 88—13 and, of the three,this slice probably does show the sharpest transition in reflectivity at the interpreted Moho.The relatively poor resolution of the 3—D model, however, does not allow for a meaningfulcomparison of detailed features in the interpreted reflection sections.Figure 5.22a shows a slice taken along a profile coincident with line 10; the ray densityplot is shown in Fig. 5.23a. Velocities from the 2—D model of O’Leary et al. (1993),determined from ray-trace based inversion and forward modelling, for the southeastern endof line 10 are shown for comparison in Fig. 5.22b. Velocities from this model were not usedto construct the starting model, but line 10 data from the three southeastern shot points wereused in the 3—D inversion. A comparison shows that upper crustal velocities in both modelsincrease to the northwest. Lower crustal velocities in the 3—D model are significantly higher,although the very high velocities shown at the southeastern end of the line are outside theregion of ray coverage in the model of O’Leary et al. (1993). Depth to Moho on the 3—Dmodel is greater (37—39 km versus 35.5—37.5 km) and upper mantle velocity, constrained bydata from one shot on the model of O’Leary et al. (1993), is lower (7.8—7.9 km/s versus8.15 km/s).176FRF88-12.2O-D1o-2000distance (km)Figure 5.20 (a) Vertical slice through a non-linear profile that follows Lithoprobe reflection lines88—11, 12 and 13 (see Fig. 5.1 6b for locations). (b) Vertical slice between SPs 6 and 2. (c) Verticalslice between SPs 11 and 18. (d) Vertical slice along y = 50 km. Nomogram at bottom rightshows locations of slices in panels a—d. See Fig. 5.10 for colour velocity scale.-30]0 50 100 150 20017710--20 -.--40f0 500-10- -20-30-400-20ci-30-40W 88-13FAF88-12I. I .1.88-11 E— tp_ —150 200wI I50 100 50 100 15050 100 150 200distance (km)Figure 5.21 Ray density plots for (a) slice in Fig. 5.20a, (b) slice in Fig. 5.20b,(c) slice in Fig. 5.20c, and (d) slice in Fig. 5.20d. See Fig. 5.19 for scale.Figure 5.22c shows a vertical slice parallel to line 2 and 70 km to the southeast; the raydensity plot is shown in Fig. 5.23b. It passes through Lithoprobe reflection profile 88—18which crosses the Fraser River Fault in the vicinity of SP 7, and is coincident with line 4.Velocities from the 2—D model for line 4 (from Fig. 5.3c) are reproduced in Fig. 5.22dfor comparison. As with line 10, the 2—D line 4 velocities were not used to construct the3—D starting model but the data recorded along this line (from both in-line and fan shots)were used in the inversion. Both models show upper crustal velocities increasing to the178--20C,-D-4010---D--20G)30____-D-40distance (km)Figure 5.22 (a) Vertical slice between SPs 6 and 21 (coincident with line 10). (b) Vertical slicethrough 2—D velocity model of O’Leary et al. (1993) between SPs 46 and 48. (c) Vertical slicebetween SPs 4 and 17 (coincident with line 4 at the northeast end). (d) Vertical slice through2—D velocity model for line 4 taken from Fig. 5.3c. (e) Vertical slice between SPs12 and 3 (coincident with line 5 near the middle). Nomogram at bottom right showslocations of slices in panels a, c and d. See Fig. 5.10 for colour velocity scale.0 50 100 0CB-IMB Line 4I I50SP 710040-0 20 40 60 80(C)0 50 100 150(e)1790-200-30-40- -20.10-40NWSP 120•’••-10j-200V-4050 100 150distance (km)Figure 5.23 Ray density plots for (a) slice in Fig. 5.22a, (b) slice in Fig.5.22c, and (c) slice in Fig. 5.22e. See Fig. 5.19 for scale.southwest. Upper crustal velocities are relatively lower between the Harrison Fault and CB1MB boundary. Deep crustal velocities are significantly higher southwest of the HarrisonFault. The CB-IMB boundary is associated with a depression of the 6.6 km/s contour tothe base of the crust.Figure 5.22e is a vertical slice paralleldensity plot is shown in Fig. 5.23c. Theto line 10, about 50 km to the northeast; the raycentral part of this profile is roughly coincident50HF100CB-IMB180with the Pasayten Fault (CB-IMB boundary) and line 5. The principal feature of this sliceis the higher velocities in the lower crust at the northwest end of the line. Upper crustalvelocities are relatively low around SP 7, and highest towards SP 3.5.6.1 Results for a 1—D starting modelTo test the reliability of the results, the modelling procedure described in section 5.4was repeated using 1—D starting models for the upper and lower crust, upper mantle, and ahorizontal starting Moho. The 1—D velocities were taken from the average 1—D model forthe study shown in Fig. 5.131. Starting Moho depth was 34.5 km. The same number ofiterations was used to model Pg and P. One additional iteration (i.e., four iterations) wasused in the quasi-simultaneous inversion step. The final rms traveltime residuals were nearlyidentical to those listed in section 5.4.Representative results for the final model are shown in Fig. 5.24 and can be compared toresults for the 3—D starting model in Figs. 5.lOd, 5.1 ib, 5.14d and 5.16a. Differences in thefinal models are negligible near the surface (12.4 km) with close correspondence betweenlocal high and low velocity anomalies. The differences become slightly more pronouncedwith increasing depth (Fig. 5.24a-c). Overall, there is still good agreement in the locationof local anomalies but the magnitude of the velocities tend to differ by a greater amount.For example, the relatively low velocities in a zone sub-parallel to the CB-IMB boundarynear the southeastern corner of the triangle (Figs. 5.llb, c and 5.14c) are present but not asprominent on Fig. 5.24c. Also, the presence of the relatively high lower crustal velocitiesin the southwest is consistent with the 3—D starting model results presented above. Maps ofdepth to Moho (Fig. 5.16a and 5.24d) show similar features, including a significantly greaterdepth throughout most of the Coast Belt. One major difference is the much shallower depthwithin the Coast Belt beneath line 2 (33—34 km for the 1—D starting model case versus 36—37km). A comparison of TWTI’s to Moho with those interpreted by Cook et al. (1992) andVarsek et al. (1993) suggests significantly poorer agreement for several of the reflection181profiles in comparison with the results for the 3—D starting model (Table 5.7). For thisreason, the model determined using the 3—D starting model is preferred.5.7 Comparison With Other Geophysical Data5.7.1 Magnetotelluric DataJones et al. (1992b) present a study of regional variations in crustal conductivity inthe southern Canadian Cordillera. Their interpretation is based on MT soundings along twoE-W oriented profiles, roughly in the locations of the vertical slices shown in Figs. 5.20aand 5.20d, but extending much further to the east. Combining the results from the twoprofiles into one depth section, they find a correlation between lateral variations in lowercrustal conductivity and the major morphogeological belt boundaries. Across the Coast andIntermontane Belts they interpret a highly resistive (>1000 m) upper crust extending toa depth of 15 km overlying a moderately resistive (150 2m) middle crust extending toabout 20 km. The lower crust of the Coast Belt is significantly more conductive (30 mversus .-s 150 m) than that of the Intermontane Belt. They interpret the variation in lowercrustal conductivity to be mainly due to varying porosity and/or salinity of lower crustal porefluids. A similar transition in lower crustal conductivity (‘.-10 m versus >200 m) wasfound across the Slocan Lake Fault, which separates the Omineca and Foreland Belts in thesoutheastern Cordillera (Jones et al. 1988). Seismic refraction data (Zelt and White 1994)shows that the higher conductivity of the Omineca Belt corresponds to significantly higherlower crustal velocities beneath SCoRE ‘90 line 9 (6.6 km/s versus 6.3 km/s). As they note,higher conductivity (together with relatively high heat flow) would normally be associatedwith reduced velocities and thus the seismic results may be suggesting that mechanismscontrolling lower crustal velocity, conductivity and heat flow may vary from the OminecaBelt to the Foreland Belt.To the west, the southern profile of Jones et al. (1992b), at roughly y =50 km inFig. 5.lOa—c, crosses over the high velocity lower crustal zone in the southern part of the182Figure 5.24 Representative results for a 1—D starting model. See Fig. 5.10 for colour velocityscale. Horizontal slices through the final 3—D velocity model at depths of (a) 18.4 km (comparewith Fig. 5.lOd) and (b) 28.0 (compare with Fig. 5.1 lb). (c) Velocities at top of upper mantle(immediately below Moho); compare with Fig. 5.1 4d. (d) Depth to Moho; compare with Fig. 5.1 6a.E>-..-.. 200E>-150X(km) X(km)183velocity model. This feature extends eastward to approximately the Fraser River Fault. Thesituation is thus analogous to that studied by Zelt and White (1994) with higher velocitylower crustal rocks of the southern Coast Belt correlating with higher conductivities relativeto Intermontane Belt values.A westward transition to higher lower crustal velocities across the Fraser River Fault isnot, however, a consistent trend throughout the model. Near the northern profile of Joneset al. (1992b) at roughly y =175 km in Fig. 5.lOa—c there is a transition to slightly lowervelocities west of the fault. A more detailed study of MT data along four short profiles acrossthe Fraser River Fault by Jones et al. (1992a) showed that the lower crust of the Coast Beltis consistently more conductive than that of the Intermontane Belt. However, there appearsto be no consistent trend in velocity, indicating that if one mechanism is responsible for theincrease in conductivity, e.g., either a small increase in porosity or salinity of pore fluids, itdoes not seem to play a significant role in determining velocity.5.7.2 Heat Flow DataComparing a map of heat flow for the southern Cordillera (Fig. 3 of Majorowicz et al.1993) with depth to Moho (Fig. 5.16a) shows that increased crustal thickness is associatedwith lower heat flow, suggesting that a significant proportion of total heat flow is sub-crustalin origin. In the northeast, where crustal thickness is a minimum (33 km) heat flowvalues are > 80 mW/rn2. Depth to Moho is 34—36 km throughout most of the rest of theIntermontane Belt where heat flow averages 70—80 mW/rn2. In most of the Coast Belt, wherecrustal thickness is >36 km, heat flow values are sparse but average 60—70 mW/rn2. Thecorrelation is not valid in the southwestern corner where Moho depth decreases to ‘-p33 km.In this location a decrease in heat flow to mW/rn2 is associated with the subducting Juande Fuca plate (Lewis et al. 1992). A subcrustal origin for a significant proportion of the totalheat flow would be expected if an upcurrent of mantle convection were to flow, at least until184recently, beneath this region as suggested by Gough (1986). The present location of mantleupflow, however, is probably to the east, beneath the Omineca Belt (Majorowicz et al. 1993).5.7.3 Gravity DataIt is beyond the scope of this study to attempt a detailed reconciliation of the velocitymodel and observed gravity field. Instead, a qualitative comparison is presented. A Bougueranomaly map for the study region is shown in Fig. 5.25a. The primary features are low valuesin the —100 to —150 mgal range throughout most of the area. This is slightly higher than theaverage value for the Coast Mountains of —140 to —160 mgal (Stacey 1973). The relativelystrong negative anomalies correspond, in general, to granitic batholiths. For example, thelarge anomaly (<—150 mgal) centered near (125,150) corresponds to Late Cretaceous plutonswhich intrude the Bridge River, Shuksan and Cadwallader terranes. A smaller anomaly at(200,160) is due to the Guichon Creek batholith (GCB), for which the gravity signature wasstudied in detail by Ager et al. (1973). Much of the region between the very low anomalyvalues is covered by volcanic rocks where the Bouguer values are typically 20 to 30 mgalmore positive (Stacey 1973). Parallel to the coastline there is a well-defined transition fromlow values (-140 mgal) in the Coast Belt to near 0 mgal along the coastline and + 10 mgal overVancouver Island. Stacey (1973) interpreted the transition to higher values over VancouverIsland as arising from either very dense and very thick (up to 70 km) crustal material andrelatively dense mantle material or low density mantle material and a very thin (<20 km)crust. More recent work, including several studies described in this thesis (e.g., Spence et al.1985; Drew and Clowes 1990; Dehler; 1991) has shown these interpretations to be incorrect.Dehier (1991), working with a more extensive gravity dataset, modelled the transition as achange from relatively dense crustal units of Wrangellia to less dense plutonic rocks of theCoast Belt. Her gravity model indicates the geometry of the transition comprises a thin (<4km) layer of low density Coast Belt plutonic rocks overlying Wrangellia for the first 30 kmeast of the Wrangellia-Coast Belt contact, thickening abruptly to a depth of at least 20 km.185The strongest anomaly in the study region is an eastward finger-like extension of relativelyhigh anomaly values in the southwestern corner where values range from —100 to —10 mgal.This feature overlies known surface exposures of Wrangellia and extends eastward overundifferentiated Jura-Cretaceous plutonic rocks and part of the Harrison terrane, terminatingnear the Harrison Fault.A comparison of velocities (Figs. 5.10 and 5.11) and gravity shows that regions ofrelatively low Bouguer values do not necessarily correspond to low velocities. The low valueregions may be caused by relatively shallow features (e.g., batholiths) which, depending ontheir depth extent, may not be well sampled by rays in regions other than those near thereceivers. For example, the GCB at (200,160) is funnel shaped and has an average depth of6 km, though the central core, with dimensions that are unresolvable by the refraction data,extends to at least 12 km (Ager et al. 1973). The GCB is associated with relatively lowvelocities (<6.1 kmls) at a depth of 2.8 km (Fig. 5.10 a). However, other zones of evenlower velocity, e.g., along the northern end of line 2, are not associated with gravity lows.The large gravity low centered near (125,150) is associated with relatively low velocities(‘.‘6.0 km/s) immediately beneath line 2 where velocity constraints are the strongest, but thelow velocities do not extend southeastwards in conjunction with the gravity low.The strongest correlation between gravity and velocity occurs in the southwestern cornerof the study area where relatively high Bouguer values are found over a region of relativelyhigh middle and lower crustal velocity (Figs. 5.10 and 5.11). A local gravity high, just eastof SF 4 at x=125 km. correlates well with relatively high velocities in the upper crust (Fig.5.lOa—c). Associating the high velocity crustal rocks with high density offers the simplestexplanation for the relative gravity high.Variations in crustal thickness do not appear to be associated with Bouguer anomalies(compare Fig. 5.25a with Fig. 5.16a). In particular, there is no obvious signature associatedwith the thicker crust throughout most of the Coast Belt, suggesting that this large-scalefeature may be roughly isostatically compensated. Similarly, upper mantle velocity (Figs.186Bouguer anomaly (mgal)Figure 5.25 (a) Bouguer anomaly map. (b) Terrane map of study region.350300250200>-150100500Terranes: / / ::::wNI:i=‘:7 (b) / I1IHCD - Cadwaflader / I ktffl!jjjjlCK=Clnfliwack I ‘ sMIHHA=HanisonKO=KootenayMT=MethowQN = Quesnellia I •SR=Shuksan‘SM — Slide Mountain ‘.J IW=Wrangellia j /m — undifferentiatedinetamorpinc rocksr.—r4 ç_J ,/ As •‘I-50 100 150X (km). I I t’’i== . . . .-200 -179 -158 -137 -116 -94 -73 -52 -31 -10200 250 0 50 100 150 200 250X (km)1875.lld and 5.14d) is also not strongly correlated with gravity. A zone of anomalously lowvelocity (7.6 km/s) at (200,260) does correlate with a local Bouguer low; however, thislocation also correlates with a surface cover of Early Jurassic plutons. The zone of highlower crustal velocity in the southwestern part of the study area is underlain by relativelylow velocity (<7.8 km/s), and presumably low density, upper mantle material. The localgravity high here, however, indicates that the dense crustal material dominates the gravitysignature.5.8 Discussion of ResultsComparing the results shown in Figs. 5.10 and 5.11 with a terrane map (Fig. 5.25b)shows that, overall, there is not a strong correlation between variations in velocity and terraneboundaries. This is not surprising given the relatively poor resolving ability of the data (seeTable 5.5) which places limits on the scale of features that can be meaningfully interpreted.Many of the terranes are smaller than the estimated lateral resolution of the model. Also, thelarge smoothing operators used in the modelling blur any sharp velocity contrasts that mayexist across geological boundaries. The CB-IMB boundary does not appear to separate crustof distinctly different velocity along its entire length. Features that may be associated withthis boundary include: (i) an increase in upper crustal velocity eastward across the boundaryin the south and corresponding increase to the west across the Harrison Fault (Figs. 5.lOa-d);(ii) an increase in upper mantle velocity across the boundary in the south (Fig. 5.1 id); (iii)a relatively low velocity channel near the boundary in the deep crust (Figs. 5.1 ib, c and5.14c); and (iv) an increase in Moho depth in the central part of the model which also seemsto be associated with the Fraser River Fault in the south (Fig. 5.16a). Table 5.8 lists theaverage velocity at different levels in the crust and upper mantle and average depth to Mohofor the Coast and Intermontane Belts. Average crustal velocities are slightly higher in theCoast Belt, with the difference in velocities between the two belts becoming larger withincreasing depth. Average upper mantle velocity is marginally lower in the Coast Belt.188Table 5.8 Average velocities at various levels in the crust and upper mantle and averagedepth to Moho in the Coast and Intermontane Belts. Averagea is the average for theentire crust. Combined’ gives the average values for the entire study area.The Central Coast Belt detachment (see Fig. 5.26 for location) represents the surfacetrace of the suture between the Insular and Intermontane superterranes. The velocity modeldoes not resolve a consistent discontinuity across the length of this boundary at any depth.In the south, the 3—D velocity model is consistent with the 2—D interpretation of line 3data, imaging a significant decrease in middle and lower crustal velocities to the east in thevicinity of the Harrison Fault.In the 2—D analysis of line 10 refraction data, O’Leary et al. (1993) used the line 3interpretation to distinguish between two different interpretations of reflection data along lines88—14 and 88—17 (J.M. Journeay in Monger and Journeay 1992; Varsek et al. 1993) (see Fig.5.1 6b for location). The two interpreted models for the collision zone between the Insularsuperterrane and North America differ significantly. J.M. Journeay (in Monger and Journeay1992) proposed crustal delamination, with the Insular superterrane (Wrangellia) displacedalong east-vergent faults over rocks of the southeastern Coast Belt. Varsek et al. (1993)proposed crustal wedging, with the Insular superterrane interfingering the Intermontanesuperterrane. The primary difference is that the interpretation of Varsek et al. (1993) requiresthe entire crust of Wrangellia to extend at least as far east as line 10 (Harrison Fault). In themodel of J.M. Journeay (in Monger and Journeay 1992), Wrangellia beneath line 10 existsaverageupper crust(<10 km)velocities(louis)6.19averagemiddle crust(10-20 km)velocities(km/s)6.40averagelower crust(>20 km)velocities(kmls)6.68averageavelocity ofcrust (lou/s)6.44averageupper mantlevelocities(lou/s)7.83averageMoho depth(km)36.7Coast beltIntermontane 6.17 6.35 6.58 6.38 7.90 34.6beltcombined” 6.18 6.38 6.64 6.42 7.87 35.9189X (kni)Figure 5.26 Map showing possible extent of lower crust of Wrangellia (shaded region) based onseismic refraction, reflection and gravity data. Heavy broken line in southwest corner (uWr) markseastern limit of upper Wrangellian surface exposures. BKFS, Bralorne-Kwoiek Creek Faultsystem; CBTS, Coast Belt Thrust System; CCBD, Central Coast Belt detachment; HF,Harrison Fault; OLF, Owl Lake Fault; TLF, Thomas Lake Fault.only as a narrow wedge between overlying Coast Belt terranes and plutonic assemblages,and underlying rocks of the southwestern Coast Belt. The similarity of the line 10 velocitystructure with line 3 velocities east of the Harrison Fault led O’Leary et al. (1993) to favourthe model of J.M. Journeay (in Monger and Journeay 1992). If the high velocity middle andlower crust in the southwest does represent Wrangellia, then the 3—D results confirm thischoice. Further, the results suggest that deep Wrangellian crust does not extend significantlyeast of SP 5 except in the south, close to line 3. If it does, its velocity is significantly lower tothe north. This is roughly consistent with the interpretation of J.M. Journeay (in Monger andJourneay 1992) which places the eastern limit for the presence of lower Wrangellia at roughlymidway between reflection lines 88—16 and 88—14 (see Fig. 5.16b for location). East of this,Wrangellia has been delaminated along east-vergent thrust faults and overlies rocks of thesouthwestern Coast Belt. This interpretation is also consistent with the 2—D velocity modelfor line 2 (McLean 1994) which has high velocity (6.7—6.8 km/s) lower crust extending?200>-0 50 100 150 200 250190no further northeast than SP 11. The inferred location of lower Wrangellia as outlinedby relatively high velocities correlates well with the relatively high Bouguer values in thesouthwest. This anomalous zone is an eastward extension of a zone of similar Bouguer valuesthat parallels the west coast where it is almost certainly underlain by Wrangellia. Figure 5.26shows the extent of lower Wrangellia that is consistent with the above observations.The association of high velocity crust in the southwestern part of the study area withWrangellia is reasonable based on the evidence presented here. Alternatively, this feature(and possibly others) may be related to more recent post-early Tertiary (40—0 Ma) tectonismwhich is associated with a change from dextral transpression and transtension to a convergentregime and that, within the Coast Belt, is dominated by Cascade arc magmatism (Mongerand Journeay 1994). Other activity associated with recent tectonics include Juan de Fucaunderplating, uplift in the last 10 Ma (Parrish 1983), and the development of northeasttrending Neogene (extensional?) faults possibly related to underthrusting of the Juan deFuca plate (J.W.H. Monger, personal communication, 1994). The high velocity crust in thesouthwest may be the product of Cascade-related mafic intrusions and magmatic underplating.The existence of a small crustal root beneath the southern Coast Belt has not beendetected by previous geophysical studies. The only previous seismic refraction survey acrossthe region (Berry and Forsyth 1974; see Fig 2.2 for location) could not resolve detailed crustalstructure because of wide receiver and shot spacing. Their interpretation shows the Mohoshallowing from about 28 km at the eastern border of the Coast Belt to 23 km at the westernborder. The thick crust of the 3—D model is, however, consistent with seismic reflectioninterpretations (Cook Ct al. 1992, Varsek et a!. 1993) and, in particular with profile 88—18(Fig. 5.16b) which shows a west side down Moho ramp or step of roughly 0.6 seconds nearthe Fraser River Fault. Reprocessed results for profile 88—18 (Perz 1993), however, suggestthat the Moho is flat across the entire profile.Thickening of the crust beneath the Coast Belt is consistent with other arc settings basedon a compilation of wide-angle surveys by Holbrook et al. (1992). They show an average191depth to Moho of 38 km which agrees well with the average depth beneath the Coast Beltof 36.7 km. The crustal root beneath the southern Coast Belt may be a response to themid-Cretaceous collision of the Insular and Intermontane superterranes or a result of recent(40 Ma) Cascade-related magmatic underplating.Holbrook et a!. (1992) found that arcs have a bimodal distribution of average velocitieswith peaks at 6.1 and 6.6 km/s (middle crust) and 6.8 and 7.0 km/s (lower crust). Incomparison, average velocities for the middle and lower crust of the Coast Belt are6.4 and 6.7 km/s. Although average P-wave velocities alone cannot uniquely determinecomposition these values suggest the presence of schists, metagabbros (greenschist) andpossibly granodiorites in the middle crust. In the lower crust, likely compositions areanorthosite, mafic granulite and possibly schist. Higher velocities in the southern part ofthe study area may indicate the presence of amphibolites and felsic and intermediate garnetgranulites (Holbrook et al. 1992).Depth to Moho beneath the Intermontane Belt averages 34.6 km, somewhat greater thanthe global average of 32 km determined by Holbrook et a!. (1992) for regions where the lastmajor tectonic event was extensional. Again, they show a bimodal distribution for averagevelocities with peaks at 6.2 and 6.6 km/s for the middle crust and 6.7 and 7.3 km/s forthe lower crust. In comparison, average middle and lower crustal velocities are 6.35 and6.6 km/s in the Intermontane Belt. These velocities suggest mid-crustal compositions ofgranodiorite, felsic amphibolite gneiss and schist, and lower crustal compositions of schist,intermediate granulite and anorthosite.5.9 SummaryThe 3—D modelling algorithm outlined in Chapter 4 has been applied to refraction datarecorded on a triangular array in the southwestern Canadian Cordillera. Two primary factorsaffect the resolution and absolute uncertainty of the final model: (i) the survey geometry,which necessitated the use of large smoothing operators, thereby decreasing resolution; and192(ii) the inferred low vertical velocity gradients in the crust which limited sampling by turningrays to the upper 10 km. so that deeper structure could be constrained only by Mohoreflections. Given these limitations, the procedure has produced a reasonably well-constrainedfinal model that is not strongly dependent on the starting model. Major features of the 2—Din-line models for lines 1 and 3 (Chapters 2 and 3) are present in the 3—D model, as expectedgiven that most of the data used in the different analyses were the same (i.e., Pg, andPs). Differences are due mainly to different model parameterizations and resolving ability,additional phases used in the 2—D analyses, and 3—D effects that are properly handled onlyin the 3—D analysis.The 3—D velocity model for the southwestern Canadian Cordillera is characterized by (i)lateral velocity variations that are, in general, not strongly correlated with surface geologicalfeatures or terrane boundaries, (ii) relatively high velocity middle and lower crust in thesouthwest corner of the study region which extends eastward to the Harrison Fault andcorrelates with a strong gravity anomaly, (iii) average upper mantle velocity of 7.85 km/swith localized regions of very low velocity (7.6 kmls) and a broad region of <7.8 km/s in thesouthern Coast Belt, (iv) depth to Moho of 33—36 km in the Intermontane Belt and 36—38throughout much of the Coast Belt, shallowing in the west near the Insular—Coast contact to 33km, (v) correlation between thick crust and low heat flow, suggesting a substantial proportionof heat flow is sub-crustal in origin, possibly associated with (a recent) upflow of mantleconvection. The relatively high velocity middle and lower crust in the southwest may outlinethe extent of lower Wrangellia beneath the southern Coast Belt (Fig. 5.26), andlor recent (40Ma) Cascade-related mafic intrusions. The former interpretation is consistent with the crustaldelamination model of J.M. Journeay (in Monger and Journeay 1992) for the collision zonebetween the Insular superterrane and North America. It is difficult to distinguish recent (40—0Ma) features associated with the Cascade magmatic arc from older, mid-Cretaceous-—-earliestTertiary (100—50 Ma) features associated with the major crust building stage in the westernCordillera. To do this will require further and more extensive integration of geological and193geophysical studies.1946 SUMMARY AND CONCLUSIONS6.1 Analysis of the Spatial Seismic Refraction Recording MethodThis thesis shows that the “spatial seismic refraction recording” method (Kanasewichand Chiu 1985), comprising a triangular array of shots and receivers as described in section1.5, can effectively image two- and three-dimensional crustal and upper mantle velocitystructure. Important questions regarding this method, however, are (z) was this method themost effective use of resources for SCoRE ‘89; and (ii) should future refraction surveysemploy this method, or some variation of the triangular geometry with a large number offan shots. There are, of course, no definitive answers to these questions, since both dependon the specific objectives of the survey. For the purposes of this discussion it can beassumed that one objective is to obtain some indication of 3—D velocity variations within thecrust and upper mantle. Both questions can be reduced to discussions of cost effectivenessand the advantages and disadvantages of 2—D versus 3—D seismic refraction modelling andinterpretation.The most expensive components of most land-based experiments are the explosives anddrilling costs, which typically account for 70% of the budget. Without the fan shots, the1989 survey would have been more conventional, i.e., comprising three long in-line profiles,and cost roughly 30% less. Alternatively, the resources used for the fan shots could havebeen used to deploy and record two additional 350 km-long lines with the same shot-receiverconfiguration as line 1. A network of five lines, if laid out strategically to maximize thenumber of line intersections (ten) and interpreted simultaneously to ensure consistency ofthe 2—D models at the intersections, could be very effective at providing useful 3—D velocityinformation by interpolating between the lines (Zelt 1994a). In practice, the mountainousterrain and shortage of roads in southwestern British Columbia would make the layout of fiveoptimally arranged lines very difficult and expensive. For this reason, the fan shot program,195especially with the addition of the three short interior recording lines, was probably the mostcost effective method of obtaining 3—D information.One limitation of the “spatial seismic refraction recording” method is that the 3—D raycoverage is limited to longer offset raypaths, typically 75 km, that sample deep structure.The lack of shallow coverage may limit the degree to which velocities will be correlatablewith surface geology. To overcome this would require a network of shots and receiversinterior to the triangle, which may not be feasible for land-based experiments due toaccessibility problems and cost constraints.For a survey in which the principal objective is to derive 3—D velocity variations, andin which accessibility is not a serious problem, a more effective use of resources in relationto the SCoRE ‘89 experimental design could be obtained by moving every second receiverand every second small (200 kg) shot point from the perimeter to the interior of the triangle.Moving every second receiver would increase the in-line spacing to 2.4—3.2 km and provides35O receivers, spaced 11 km apart, for the interior of the triangle. Moving the the smallshots would increase the in-line shot spacing to 100 km and provide shot points (inaddition to the large center shot) for the interior of the triangle. The increased receiver andshot spacing along the in-line profiles would result in more poorly constrained 2—D modelsfor the perimeter of the triangle. The 2—D models, however, would still provide very usefulvelocity information and could be used to construct an adequate 3—D starting model. Thedense receiver spacing along the perimeter is not required for the 3—D analysis becauseraypaths from a shot point to two very closely spaced receivers do not provide independentinformation of velocity structure, except very near the receivers. The trade off for poorer inline coverage is much greater areal coverage interior to the triangle and hence more isotropicray coverage at all levels. This would provide for a more strongly constrained 3—D velocitymodel at all depths.Figure 6.1 shows a modification of the SCoRE ‘89 experimental design using the sametrianglular geometry and roughly the same number of receivers and shotpoints, but designed196for improved 3—D coverage, as described above. Receiver spacing along the perimeter istwice as large, and the number of small shot points along the lines has been halved withrespect to the actual design (compare with Fig. 5.1). Approximately 300 receivers on an11 xli km grid are shown interior to the triangle (a precise distribution as shown is notimportant). Six of the small in-line shots have been dispersed throughout the interior ofthe triangle. With the number of recording instruments presently available a survey of thisdesign could, in principle, be recorded in one deployment, although it would require a greatdeal of manpower. This design, however, is impractical for the southern Cordillera and otherareas because of the inaccessibilty of much of the interior region of the triangle. Designs ofthis nature, i.e., a 2—D grid of shots and receivers, are more common and practical in marineexperiments, with ocean bottom seismometers replacing the shot points, and shots from aship steaming along a grid replacing the receiver sites (e.g., Toomey et al. 1990; Zelt 1994b).Future use of the “spatial seismic refraction recording” method will depend on the specificobjectives of a survey. The results of this thesis demonstrate the type of 3--D image andresolution that can be obtained using the analysis technique presented in Chapter 4. Futureenhancements to this recording method such as described above, modifications to the analysisprocedure presented in this thesis, or different analysis procedures may offer significantimprovements in terms of image quality, estimates of model resolution and uncertainty, andthe ability to incorporate constraints.The benefits of a network of intersecting 2—D profiles, if interpreted simultaneously toensure consistency using current methods (Zelt and Smith 1992; Zelt 1994a), in comparisonto the 3—D analysis method presented here include: (i) easy to incorporate amplitudeinformation; (ii) easy to include non-first arrival refracted phases; (iii) easier to testmany models and perform hypothesis testing; (iv) easy to incorporate constraints on bothvelocity/gradients and depths to boundaries from other sources; (v) get formal estimates ofresolution and uncertainty; (vi) easier to visualize results; and (vii) can obtain 3—D informationby interpolating between lines. Disadvantages of 2—D interpretations are: (i) 3—D effects can197350300250200E>-150100500X (km)Figure 6.1 Hypothetical experimental design for improved 3—D recording in the southwesternCordillera. The design as shown incorporates the same triangular configuration as SCoRE ‘89 andcomprises approximately the same number of receivers and shot points. Receiver spacing along theperimeter is 2.4—3.2 km. Receivers in the interior of the triangle are shown on an 11 xli kmgrid. This design would provide better 3—D ray coverage but would be impractical inmany locations because of the inaccessibility of regions interior to the triangle.cause errors; and (ii) interpolation between profiles, depending on their spacing, may notprovide adequate information on 3—D variations (this would be the case for the three SCoRE‘89 lines). Depending on the study region, the first problem may not be considered serious.More work, e.g., synthetic tests, is needed to show the effects of 3—D variations on 2—D0 50 100 150 200 250198interpretations.6.2 Analysis of the 3—D Modelling ProcedureThe effectiveness of the 3—D analysis technique presented in Chapter 4, and applied inChapter 5, is partly determined by the dataset, which in turn is primarily controlled by therecording geometry and the velocity structure of the study region. These important factorslimit the degree of success any algorithm will have in imaging structure. For example, lowvertical velocity gradients in the southern Cordillera limit the depth extent to which refractedarrivals sample the crust. Thus deeper crustal velocities can be constrained only by PmPreflections, which are also the primary constraint for Moho depth. This leads to a trade-offbetween lower crustal velocities and Moho depth and gives rise to non-uniqueness.The advantages of the analysis method presented here over some others are: (i) bothreflected and refracted arrivals are used; (ii) fast and accurate forward modelling; (iii) thedensely sampled model allows for fine resolution, if warranted by the data, by adjustingthe dimensions of the smoothing operators; (iv) simple and fast inversion schemes of bothrefracted and reflected arrivals which eliminate the need to solve a large system of equationsand allows for densely sampled models. Disadvantages include: (i) no formal estimatesof resolution and parameter uncertainties—partly compensated for by synthetic tests suchas described in section 5.6.1; (ii) streaking of the model along raypaths, common for backprojection-type procedures—can be reduced by using appropriately dimensioned smoothingoperators; (iii) the method is valid only for first arrival refractions and reflections—probablynot a serious restriction since for many datasets other types of arrivals usually comprise onlya small percentage of the total data; (iv) method for generating reflections is not valid forvery steeply dipping interfaces (35°)—only a problem if the data warrant the use of verylittle smoothing, since the smoothing operation limits the maximum dip that can be imaged,and the study region does contain steeply dipping faults; and (v) requires equal grid nodespacing in all directions which effectively means that the required vertical resolution governs199the maximum allowed grid size—not a serious problem unless computer memory is limited,but ideally would like to use smaller vertical grid node spacing in comparison to lateralnode spacing. This, however, would require major modifications to the 3—D finite-differencetraveltime code.The algorithm presented in this thesis has produced a fairly detailed 3—D velocity modelfor the southwestern Canadian Cordillera. The model has unconstrained regions which areeasy to identify and result from the dataset used in the inversion. The reliability of the modelcan be gauged from the traveltime misfits, plots of ray coverage, results of tests using alternatestarting models, and knowledge of the amount of smoothing used in the model construction.Correlation, or a lack of correlation, of features in the velocity model with surface geologicalfeatures, geological models, and other geophysical data does not necessarily imply reliability,or a lack of reliability, since many factors can control crustal and upper mantle velocities.6.3 SummaryThis thesis addresses the modelling and interpretation of seismic refraction/wide-anglereflection data recorded in the southwestern Canadian Cordillera. The primary results can bedivided into four separate areas of research (Chapters 2—5): (i) analysis of in-line data alongline 1 for 2—D structure along-strike in the southern Intermontane Belt; ii) analysis of in-linedata along line 3 for 2—D structure across the strike of the southernmost Coast and InsularBelts; (iii) development of an algorithm to invert 3—D wide-angle seismic data using firstand reflected arrivals; and (iv) application of the algorithm to determine the 3—D velocitystructure of the southwestern Canadian Cordillera beneath the SCoRE ‘89 triangle.The analyses of in-line data used an iterative combination of 2—D traveltime inversion(Zelt and Smith 1992) and forward modelling of amplitudes (Zelt and Ellis 1988) to interpretcrust and upper mantle P-wave velocity structure. This procedure, which has subsequentlybeen used by others (O’Leary et al. 1993; Kanasewich et al. 1994), proved effective atproviding well-constrained velocity models with estimates of resolution and uncertainties of200model parameters.The line 1 analysis shows the Intermontane Belt to be characterized by (1) a thin (<2.5km) near-surface layer with velocities of 2.8—5.4 km/s; (ii) an upper and middle crust withlow average vertical velocity gradients and lateral velocity variations between 5.95—6.55km/s; (iii) distinctive lower crust beginning at 25 km depth, approximately 8 km thickwith velocities of 6.15—6.6 km/s at the top and 6.55—7.15 km/s at the base; (iv) depth toMoho that averages 32 km with variations of —1.5 km; and (v) a Moho transition zone ofmaximum depth extent 3.5 km. overlying an upper mantle of average velocity 7.85 km/s.varying between 7.6—7.95 km/s. Where the refraction line obliquely crosses a Lithoprobedeep seismic reflection profile, there is good agreement between the interpreted reflectionsection and the 2—D velocity model. In particular, depths to wide-angle reflectors in theupper crust agree with depths to prominent reflection events and Moho depths agree within1 km. From this comparison, the upper and middle crust probably comprise the upper partof the Quesnellia terrane. The lower crust from the refraction interpretation does not showthe division into two components, parautochthonous and cratonic North America, that isinferred from the reflection data, indicating that their physical properties are not significantlydifferent within the resolution of the refraction data. Based on these interpretations, thelower lithosphere of Quesnellia is absent and presumably was recycled in the mantle. Ata depth of -- 16 km below the Moho, an upper mantle reflector may represent the base ofthe present lithosphere.The line 3 model for the southernmost Coast Belt and eastern Insular Belt is characterizedby large lateral variations in velocity. The most significant of these variations is a decreasein upper and middle crustal velocities to the east of the surface trace of the Harrison Faultwhich likely represents the transition from Insular to Intermontane superterrane crust. Thisinterpretation is consistent with present geological models which place the suture betweenthe two superterranes less than 20 km east of the Harrison Fault. Velocities at the baseof the upper crust average 6.4 and 6.2 km/s west and east of the fault, respectively. Mid-201crustal velocities average 6.6—6.9 km/s to the west and 6.35—6.45 km/s to the east of thefault. Velocities in the lower crust also decrease slightly to the east. Other features of thevelocity model include: (i) a thin near-surface layer with velocities between 2.5 and 6.1kmls; (ii) upper crustal thickness of 12.5 km, thinning to 8 km at the eastern boundary ofthe Western Coast Belt (WCB); (iii) high velocity (6.6—6.9 km/s) mid-crustal layer west ofthe Harrison Fault extending to 21 km depth; (iv) high velocity (6.75—7.1 kmls) lower crust;(v) low velocity gradient upper mantle with depth to Moho at 34—37 km beneath most of theCoast Belt, decreasing to 30 km beneath the eastern Insular Belt; much less than previousestimates. The inferred crustal velocity structure beneath the WCB is consistent with thethree layer conductivity structure for this area. The association of high resistivities with theupper crust suggests that the upper 8—12 km represents the massive cover of plutonic rockswhich characterizes the WCB. Middle and lower crustal velocities beneath the WCB areconsistent with Wrangellian velocities beneath Vancouver Island, suggesting Wrangellia mayextend at depth eastwards as far as the Harrison Fault.An algorithm for the inversion of wide-angle seismic data to determine 3—D velocitystructure and depth to reflecting interfaces was developed to analyze the large SCoRE ‘89fan-shot dataset. The algorithm is based on further developments of the first arrival traveltimeinversion procedure of Hole (1992) which includes (i) forward modelling of traveltimes usinga 3—D finite-difference algorithm; and (ii) a simple velocity model parameterization for theinversion which eliminates the need to solve a large system of equations. The Hole (1992)algorithm was extended to allow (i) fast and accurate forward modelling of reflection times;(ii) the inversion of reflection times to solve for depth to a reflecting interface and/or velocitystructure; (iii) the inversion of first arrival traveltimes to solve for depth to a refractinginterface; and (iv) layer stripping. The complete modelling algorithm uses Pg to constrainupper crustal velocity structure, PmP to constrain lower crustal velocity structure and depthto Moho, and P to constrain upper mantle velocities and depth to Moho.The 3—D velocity model for the southwestern Canadian Cordillera is characterized by (I)202significant lateral velocity variations at all depths that do not, in general, strongly correlatewith surface geological features or, with one major exception, gravity anomalies; (ii) higheraverage crustal velocities in the Coast Belt in comparison with the Intermontane Belt; (iii)relatively high velocity middle and lower crust in the southwest corner; (iv) average uppermantle velocities of 7.85—7.9 km/s with localized regions of very low velocity (<7.6 km/s)and a broad region of <7.8 km/s in the southern Coast Belt; (v) depth to Moho of 33—36 kmin the Intermontane Belt and 36—38 km throughout most of the Coast Belt, shallowing to thewest to 33 km near the Insular-Coast contact. These results are, given the poorer resolutionof the 3—D model and differences in modelling techniques, consistent with the 2—D resultsdescribed above. The zone of high velocity middle and lower crust in the southwest correlateswith a strong relative gravity high and may outline the extent of lower Wrangellia (Insularsuperterrane) in the southern Coast Belt, i.e., to the east as far as the Harrison Fault andnorth to .‘49.5° N latitude. This interpretation favours the crustal delamination model of J.M.Journeay (in Monger and Journeay 1992) for the Insular-Intermontane superterrane collisionzone. Alternatively, the high velocity zone may be a younger feature outlining the extent ofmafic intrusions associated with recent Cascade arc magmatism. A correlation between thickcrust and low heat flow suggests that a significant portion of total heat flow is sub-crustalin origin, possibly associated with (a recent) upflow of mantle convection as suggested byGough (1986).Figure 6.2b shows a simple schematic interpretation of the present crustal structure alonga southwest-northeast oriented (cross-strike) profile (Fig. 6.2a) across the Coast and westernIntermontane Belts. Geologic boundaries are based on results presented in this thesis andother geological and geophysical information. Velocities shown are the average velocities ina ±25 km wide swath surrounding the profile (shaded box in Fig. 6.2a). In the southwest,beneath the Western Coast Belt, the crust comprises Wrangellia (Insular superterrane) plusgranitic rocks. The eastemmost surface exposures of Wrangellia occur about 50 km fromthe western end of the profile (uWr in Fig. 6.2b). Results from the 3—D analysis suggest203that along this profile deep Wrangellian crust may not extend much further east than this,too. Below 10 km depth, the eastern boundary of Wrangellia shown in Fig. 6.2b (solid line)is consistent with this conclusion and the crustal delamination model of J.M. Journeay (inMonger and Journeay 1992). If the association of Wrangellia with relatively high velocitiesin the deep crust is incorrect, e.g., the high velocity crust may be a more recent feature dueto Cascade arc magmatism, then the refraction results do not rule out the crustal wedgingmodel of Varsek et al. (1993). The black and white broken line in Fig. 6.2b depicts theeastern limit of deep Wrangellia in their interpretation.To the east of the uWr line, the shallow crust comprises voluminous Jura-Cretaceousplutons that probably intrude Wrangellia. The same plutonic rocks intrude the Harrisonterrane to the east, linking it to Wrangellia by at least 165 Ma (Monger 1991). The Harrisonterrane extends westward to approximately the Thomas Lake Fault, the basal thrust of theHarrison allochthon, and westernmost fault of the Coast Belt Thrust System (CBTS), a westvergent contractional belt that formed along the eastern margin of the Insular superterranein early-Late Cretaceous time (Journeay and Friedman 1993). Other major faults in thissystem include from west-to-east: (I) the Owl Lake Fault, which extends northwestwardfrom the dextral Harrison Fault; (ii) the Central Coast Belt detachment (CCBD), which isthe floor thrust to overlying terranes of the Eastern Coast Belt; and (iii) the Bralorne-KwoiekCreek Fault system (BKFS), which marks the eastern boundary of the CBTS (J.M. Journeay,personal communication, 1994). These structures roughly demarcate several small Coast Beltterranes. The Harrison terrane is present as far east as the Owl Lake Fault. The Cadwalladerterrane is probably present west of the Owl Lake Fault, and as far east as the CCBD. Betweenthe CCBD and BKFS lie rocks primarily of the Shuksan terrane which correlate with rocksof the Easton terrane in the Northwest Cascades System (Monger 1991). The Bridge Riverterrane is present between the BKFS and Fraser River Fault. Bridge River rocks are alsopresent east of the Fraser River Fault further south in the Eastern Cascades, having beenoffset by 80—190 km of Eocene dextral motion. Although not shown on the cross section20401020G) 30404- ECB-+ 1MB5.80H 6.00-6.10-6.25 --6.40 2.-6.506.757.007 cn Cl)I •4J’J %7.85-8.10Figure 6.2 (a) Location map for cross section (heavy black line) in b. Velocities in the crosssection represent average values within the shaded rectangle. uWr, eastern limit of upperWrangellian surface exposures. (b) Schematic cross section across the Coast and westernIntermontane Belts. WCB, Western Coast Belt; ECB, Eastern Coast Belt; 1MB, Intermontane Belt.Faults depicted are: CBTS, Coast Belt Thrust System; TLF, Thomas Lake Fault; OLF, Owl LakeFault; CCBD, Central Coast Belt detachment; BKFS, Bralorne-Kwoiek Creek Fault system;FRF, Fraser River Fault. Terrane abbreviations: Wr, Wrangellia (“+g” represents thedominant granitic rocks emplaced in Wrangellia); HA, Harrison; CD, Cadwallader; SH,Shuksan; BR, Bridge River; Qn, Quesnellia. Other abbreviations: PNA, parautochthonousNorth America; +, Jura-Cretaceous plutons; M, Moho (heavy broken line).SWI WCBX (km)(a)Line 3Liun;10I—CBTS— IFRFOLF CCBD LLine 1.I.100 150distance (km) (b)205of Fig. 6.2b, terranes west of the CCBD are intruded by mainly late Middle Jurassic tomid-Cretaceous (165—95 Ma) plutons; terranes to the east are intruded by younger, mostlymid-Cretaceous to early Tertiary (95-46 Ma), plutons.The Coast Belt terranes east of the Thomas Lake Fault are probably relatively shallowfeatures, detached from their basements and stacked along thin-skinned thrust faults (Journeayand Friedman 1993). Individual terranes are not resolved in the refraction models. Basedon interpretations of reflection data, they may be no thicker than -‘10 km (Varsek et al.1993; J.M. Journeay in Monger and Journeay 1992). A reinterpretation of one of thereflection profiles (Perz 1993), however, indicates that thrust faults of the CBTS may soleout substantially deeper (20—25 km depth; J.M. Journeay, personal communication to M.J.Perz in Perz 1993). Thus, the terranes may extend down into the region denoted “CB?”in Fig. 6.2b. In this scenario, the zone marked “Wr?” (above the broken white line) maycomprise Wrangellia.East of the dextral Fraser River Fault, the crust consists primarily of Quesnellia, one offour terranes of the Intermontane superterrane. A zone of low velocities at the top of the crustimmediately east of the Fraser River Fault outlines the Middle Jurassic Mount Lytton plutoniccomplex and Early Jurassic Guichon Creek batholith. Based on reflection interpretations(Cook et al. 1992), the lower lithosphere of Quesnellia is not present in the vicinity of line1; presumably it was removed by subduction erosion and/or crustal delamination. Instead,the relatively high velocity deep crust in the Intermontane Belt comprises parautochthonousNorth American rocks, and possibly a very thin layer of highly strained North Americancrust. West of the Fraser River Fault, below the Coast Belt terranes, and east of Wrangellia,the deep crust may comprise highly metamorphosed rocks of the Intermontane superterrane,and possibly deeper elements of the overlying shallow terranes.The present crustal structure described above and depicted in the cross section of Fig.6.2b is the product of processes that probably span a time interval from the early Mesozoicto the present. In the Intermontane Belt, removal of Quesnellia’s lower lithosphere by206erosion may have begun in the Late Triassic (prior to the accretion of the Intermontanesuperterrane to North America) as the oceanic Cache Creek terrane was being subductedbeneath it. The disrooting continued in the Middle Jurassic by crustal delamination asIntermontane rocks were thrust eastwards over parautochthonous North American material(J.W.H. Monger, personal communication, 1994). The lower lithosphere of Quesnellia waspresumably recycled in the mantle. A similar process may have delaminated the lithosphere ofWrangellia as the Insular superterrane collided with the Intermontane superterrane in the EarlyCretaceous. Alternatively, the two superterranes may have wedged together, with Insularmaterial interfingering the Intermontane superterrane at the collision zone. Contractionassociated with continuing convergence between the two superterranes into the early Tertiaryled to crustal thickening. The CBTS formed in response to convergence in the early-LateCretaceous. Terranes within the CBTS were detached from their basements and stacked alongthin-skinned thrust faults, also resulting in a large amount of crustal thickening, the effects ofwhich have been mostly removed by erosion during subsequent uplift. The small crustal rootpresent in the Coast Belt may, in part, be a remnant signature of this contractional activity.Plutonism, another mechanism of crustal thickening, occurred throughout the entireperiod of convergence associated with the accretion of the two superterranes. Terranes in theWestern Coast Belt are intruded by mainly late Middle Jurassic to mid-Cretaceous plutons.In the Eastern Coast Belt, the intrusives are of mid-Cretaceous to early Tertiary age. Theireffects are observed as the extensive surface plutons in the region. Their depth extent is notstrongly constrained, but appears to be less than 10 km.A change from a contractional regime to one of extension, transtension, and strike-slipfaulting beginning in the Early Eocene is represented in the cross section of Fig. 6.2b by theFraser River Fault sytem, The Fraser River and associated faults offset structures by 80—190km of dextral motion in the Middle and Late Eocene (Misch 1977; Monger 1985; Colemanand Parrish 1991; Friedman and van der Heyden 1992).The most recent phase of crustal evolution, within the last 40 Ma, has been dominated207by activity associated with the Cascade magmatic arc. This includes uplift of the CoastMountains, the development of a northeast-trending fault system between 25—14 Ma whichpossibly controlled the emplacement of older volcanic and plutonic rocks, the effusion ofyoung (4—0 Ma) volcanic rocks in the Garbaldi volcanic belt, and arc magmatism (Mongerand Journeay 1994). The manifestation of effects associated with these younger processesis not readily apparent. High velocities in the southern Coast Belt may be the result of32 Ma Cascade-related mafic intrusion (J.W.H. Monger, personal communication, 1994).The present depth to the Moho in the southwestern region of the study area also may becontrolled by recent tectonism.The cross section in Fig. 6.2b depicts a simple interpretation of the present crustalstructure in the Coast and Intermontane Belts. Behind the simplicity, however, lies a complexseries of events and processes which have been active for an extended period of time. The2— and 3—D velocity models presented in this thesis provide constraints on the crust andupper mantle velocity structure of the southwestern Canadian Cordillera and, therefore, helpto better understand the processes that led to its development. They provide part of the database on which realistic geodynamical models of tectonic processes (Brown 1993; Beaumontet al. 1994) will be founded.208REFERENCESAger, C.A., T.J. Ulrych, and WJ. McMillan, 1973. A gravity model for the Guichon Creek Batholith,south-central British Columbia. Canadian Journal of Earth Sciences, 10: 920—935.Aid, K., A. Christoffersson, and E.S. Husebye, 1977. Determination of the three—dimensional seismicstructure of the lithosphere. Journal of Geophysical Research, 82: 277—296.Amato, A., G. Biella, and R. de Franco, 1991. Velocity structure of the Vulsinian volcanic complex(Latium, Italy) from seismic refraction data and three-dimensional inversion of travel times. Journalof Geophysical Research, 96: 517—535.Beaumont, C., P. Fullsack, and J. Hamilton, 1994. Styles of crustal deformation in compressionalorogens caused by subduction of the underlying lithosphere. Tectonophysics, 232: 119—132.Bennett, G.T., R.M. Clowes, and R.M. Ellis, 1975. A seismic refraction survey along the southernRocky Mountain Trench, Canada. Bulletin of the Seismological Society of America, 65: 37—54.Benz, H.M., R.B. Smith, and W.D. Mooney, 1990. Crustal structure of the northwestern Basin andRange Province from the 1986 program for array seismic studies of the continental lithosphericseismic experiment. Journal of Geophysical Research, 95: 21823—21842.Berry, M.J., W.R. Jacoby, E.R. Niblett, and R.A. Stacey, 1971. A review of geophysical studies inthe Canadian Cordillera. Canadian Journal of Earth Sciences, 8: 788—801.Berry, M.J., and D.A. Forsyth, 1975. Structure of the Canadian Cordillera from seismic refractionand other data. Canadian Journal of Earth Sciences, 12: 182—208.Braun, J., 1993. Three-dimensional modeling of compressional orogenies: thrust geometry andoblique convergence. Geology, 21: 153—1 56.Cassidy, J.F., and R.M. Ellis, 1993. S-wave velocity structure of the Northern Cascadia subductionzone. Journal of Geophysical Research, 98: 4407—4421.erven9, V., I. Molotkov, and I. Penik, 1977. Ray method in seismology. Charles UniversityPress, Prague.Chander, R., 1977. On tracing seismic rays with specified end points in layers of constant velocityand plane interfaces. Geophysical Prospecting, 25: 120—124.Chandra, N.N., and G.L. Cumming, 1972. Seismic refraction studies in western Canada. CanadianJournal of Earth Sciences, 9: 1099—1109.Clowes, R.M., 1990. Lithoprobe—multidisciplinary studies of continental evolution: an examplefrom western Canada. In Proceedings, Pacific Rim Congress ‘90, Australasian Institute of Miningand Metallurgy, Gold Coast, Queensland, 6—12 May 1990, Australasian Institute of Mining andMetallurgy, Parkville, Victoria, Australia, III: 83—97.209Clowes, R.M., M.T. Brandon, A.G. Green, C.J. Yorath, A. Sutherland Brown, E.R. Kanasewich,and C.S. Spencer, 1987. Lithoprobe—southern Vancouver Island: Cenozoic subduction compleximaged by deep seismic reflections. Canadian Journal of Earth Sciences, 24: 31—51.Clowes, R.M., F.A. Cook, A.G. Green, C.E. Keen, J.N. Ludden, J.A. Percival, G.M. Quinlan, andG.F. West, 1992. Lithoprobe: new perspectives on crustal evolution. Canadian Journal of EarthSciences. 29: 1813—1864.Coleman, M.E., and R.R. Parrish, 1991. Eocene dextral strike-slip and extensional faulting in theBridge River terrane, southwest British Columbia. Tectonics, 10: 1222—1238.Coney, P.J., D.L. Jones, and J.W.H. Monger, 1980. Cordilleran suspect terranes. Nature, 288:329—333.Cook, F.A., J.L. Varsek, and R.M. Clowes, 1991. Lithoprobe reflection transect of southwesternCanada: Mesozoic thrust and fold belt to mid-ocean ridge. In Continental lithosphere: deepseismic reflections. Edited by R. Meissner, L.D. Brown, H. Durbaum, K. Fuchs, and F. Seifert.American Geophysical Union, Geodynamics Series, 22: 247—255.Cook, F.A., J.L. Varsek, R.M. Clowes, E.R. Kanasewich, C.S. Spencer, R.R. Parrish, R.L. Brown,S.D. Carr, B.J. Johnson, and R.A. Price, 1992. Lithoprobe crustal reflection cross section of thesouthern Canadian Cordillera 1: Foreland Thrust and Fold Belt to Fraser River Fault. Tectonics,11: 12—35.Crickmay, C.H., 1930. The structural connection between the Coast Range of British Columbia andthe Cascade Range of Washington. Geological Magazine, 67: 482—491.Cumming, W.B., R.M. Clowes, and R.M. Ellis, 1979. Crustal structure from a seismic refractionprofile across southern British Columbia. Canadian Journal of Earth Sciences, 16: 1024—1040.Davis, E.E., and T.J. Lewis, 1984. Heat flow in a back-arc environment: Intermontane and OminecaCrystalline Belts, southeastern Canadian Cordillera. Canadian Journal of Earth Sciences, 21:715—726.Dehler, S.A., 1991. Integrated geophysical modelling of the northern Cascadia subduction zone.Ph.D. thesis, Department of Geophysics and Astronomy, University of British Columbia, Vancouver.Drew, J.J., and R.M. Clowes, 1990. A re-interpretation of the seismic structure across the activesubduction zone of western Canada. In Studies of laterally heterogeneous structures using seismicrefraction and reflection data. Edited by A.G. Green. Geological Survey of Canada, Paper 89—13,p 115—132.Ewing, T.E., 1980. Paleogene tectonic evolution of the Pacific Northwest. Journal of Geology, 88:619—638.210Friedman, R.M., and P. van der Heyden, 1992. Late Permian U-Pb dates for the Farwell and northernMount Lytton plutonic bodies; Intermontane Belt, British Columbia. In Current research, Part A.Geological Survey of Canada, Paper 92-la, pp. 137—144.Fuis, G.S., and W.D. Mooney, 1990. Lithospheric structure and tectonics from seismic-refraction andother data. In The San Andreas Fault system. Edited by R.E. Wallace. United States GeologicalSurvey Professional Paper 1515, pp. 206—236.Fuis, G.S., and 0. Plafker, 1991. Evolution of deep structure along the Trans-Alaska Transect,Chugach Mountains and Copper River Basin, southern Alaska. Journal of Geophysical Research,96: 4229—4253.Gabrielse, H., and C.J. Yorath, 1989. DNAG #4. The Cordilleran orogen in Canada. GeoscienceCanada, 16: 67-83.Gabrielse, H., and C.J. Yorath (editors), 1992. Geology of the Cordilleran Orogen in Canada,Geological Survey of Canada, Geology of Canada no. 4 (also Geological Society of America, TheGeology of North America, G—2: 844 p.Gehrels, G.E., and J.B. Saleeby, 1985. Constraints and speculations on the displacement andaccretionary history of the Alexander-Wrangellia-Peninsular superterrane [abstract]. GeologicalSociety of America, Abstracts with programs, 17: 356.Ghosh, D.K., 1991. Nd and Sr isotopic constraints on the timing of obduction of Quesnellia onto thewestern edge of North America. In Southern Cordillera transect workshop. Edited by F.A. Cook.University of British Columbia, Lithoprobe Secretariat, Report 16: pp. 38—55.Gough, D.I., 1986. Mantle upifow tectonics in the Canadian Cordillera. Journal of GeophysicalResearch, 91: 1909—1919.Gough, D.I., and J.A. Majorowicz, 1992. Magnetotelluric soundings, structure, and fluids in thesouthern Canadian Cordillera. Canadian Journal of Earth Sciences, 29: 609—620.Green, A.G., R.M. Clowes, C.J. Yorath, C.S. Spencer, E.R. Kanasewich, M.T. Brandon, and A.Sutherland Brown, 1986. Seismic reflection imaging of the subducting Juan de Fuca plate. Nature,319: 210—213.Gridley, J.M., 1993. Crustal structure of western Washington State. Ph.D. thesis, Department ofGeological Sciences, The University of Texas at El Paso, El Paso.Harris, J.M., S. Lazaratos, and R. Michelena, 1990. Tomographic string inversion [abstract]. Societyof Exploration Geophysics Meeting, Expanded Abstracts, pp. 82—85.Hawley, B.W., G. Zandt, and R.B. Smith, 1981. Simultaneous inversion for hypocenters and lateralvelocity variations: An iterative solution with a layered model. Journal of Geophysical Research,86: 7073—7086.211Hearn, T.M., and F.F. Ni, 1994. P velocities beneath continental collision zones: the Turkish-IranianPlateau. Journal of Geophysical Research, 117: 273—283.Holbrook, W.S., 1990. The crustal structure of the northwestern Basin and Range Province, Nevada,from wide-angle seismic data. Journal of Geophysical Research, 95: 21843—21869.Holbrook, W.S., R.D. Catchings, and C.M. Jarchow, 1991. Origin of deep crustal reflections:Implications of coincident seismic refraction and reflection data in Nevada. Geology, 19: 175—179.Holbrook, W.S., W.D. Mooney, and N.J. Christensen, 1992. The seismic velocity structure of thedeep continental crust. In Continental lower crust. Edited by D.M. Fountain, R. Arculus, andR.W. Kay. Elsevier, pp. 1—43.Hole, J.A., 1992. Nonlinear high-resolution three-dimensional seismic travel time tomography.Journal of Geophysical Research, 97: 6553—6562.Hole, J.A., R.M. Clowes, and R.M. Ellis, 1992. Interface inversion using broadside seismicrefraction data and three-dimensional travel time calculations. Journal of Geophysical Research,97: 3417—3429.Hole, J.A., R.M. Clowes, and R.M. Ellis, 1 993a. Interpretation of three-dimensional seismic refractiondata from western Hecate Strait, British Columbia: structure of the crust. Canadian Journal ofEarth Sciences, 30: 1440—1452.Hole, J.A., R.M. Clowes, and R.M. Ellis, 1993b. Interpretation of three-dimensional seismic refractiondata from western Hecate Strait, British Columbia: structure of the Queen Charlotte Basin.Canadian Journal of Earth Sciences, 30: 1427—1439.Hole, J.A., and B.C. Zelt. Three-dimensional finite-difference reflection travel times. Submitted toGeophysical Journal International.Humphreys, E., and R.W. Clayton, 1988. Adaptation of back projection tomography to seismic traveltime problems. Journal of Geophysical Research, 93: 1073—1085.Hunch, C.A., S.B. Smithson, D.M. Fountain, and M.C. Humphreys, 1985. Seismic evidence ofmylonite reflectivity and deep structure in the Kettle dome metamorphic core complex, Washington.Geology, 13: 577—5 80.Hyndman, R.D., T.J. Lewis, and G. Marquis, 1991. Comment on “Origin of deep crustal reflections:Implications of coincident seismic refraction and reflection data in Nevada”. Geology, 19: 1243.Irving, E., and R.W. Yole, 1972. Paleomagnetism and the kinematic history of mafic and ultramaficrocks in fold mountain belts. In The ancient oceanic lithosphere. Edited by E. Irving. EarthPhysics Branch Publication, 42—3, pp. 87—95.Jones, A.G., R.D. Kurtz, D.W., Oldenburg, D.E. Boerner, and R.G. Ellis, 1988. Magnetotelluric observations along the Lithoprobe southeastern Canadian Cordillera transect. Geophysical Research212Letters, 15: 677—680.Jones, A.G., RD. Kurtz, D.E. Boerner, J.A. Craven, G.W. McNeice, D.I. Gough, J.M. DeLaurier,and R.G. Ellis, 1991. Electromagnetic investigations over the southern Cordilleran Lithoprobetransect: 1991 status report. In Southern Cordillera transect workshop. Edited by F.A. Cook.University of British Columbia, Lithoprobe Secretariat, Report 16: pp. 38—55.Jones, A.G., R.D. Kurtz, D.E. Boerner, J.A. Craven, G.W. McNeice, D.I. Gough, J.M. DeLaurier,and R.G. Ellis, 1 992a. Electromagnetic constraints on strike-slip fault geometry—The FraserRiver Fault system. Geology, 20: 561—564.Jones, A.G., D.I. Gough, R.D. Kurtz, J.M. DeLaurier, D.E. Boerner, J.A. Craven, R.G. Ellis, andG.W. McNeice, 1 992b. Electromagnetic images of regional structure in the southern CanadianCordillera. Geophysical Research Letters, 19: 2373—2376.Jones, T.D., and A. Nur, 1984. The nature of seismic reflections from deep crustal fault zones.Journal of Geophysical Research, 89: 3153—3171.Journeay, J.M., and L. Csontos, 1989. Preliminary report on the structural setting along the southeastflank of the Coast Belt, British Columbia. In Current research, Part E. Geological Survey ofCanada, Paper 89-lB. pp. 177—187.Journeay, J.M., and R.M. Friedman, 1993. The Coast Belt Thrust System: Evidence of LateCretaceous shortening in southwest British Columbia. Tectonics, 12: 756—775.Kanasewich, E.R., and S.K.L. Chiu, 1985. Least-squares inversion of spatial seismic refraction data.Bulletin of the Seismological Society of America, 75: 865—880.Kanasewich, E.R., M.J.A. Burianyk, R.M. Ellis, R.M. Clowes, D.J. White, T. Coté, D.A. Forsyth,J.H. Luetgert, and G.D. Spence, 1994. Crustal velocity structure of the Omineca Belt, southeasternCanadian Cordillera. Journal of Geophysical Research, 99: 2653—2670.Klemperer, S.L., 1987. A relation between continental heat flow and the seismic reflectivity of thelower crust. Journal of Geophysics, 61: 1—11.Klemperer, S.L., and BIRPS Group, 1987. Reflectivity of the crystalline crust: hypotheses and tests.Geophysical Journal of the Royal Astronomical Society, 89: 2 17—222.Lewis, T.J., A.M Jessop, and A.S. Judge, 1985. Heat flux measurements in southwestern BritishColumbia: the thermal consequences of plate tectonics. Canadian Journal of Earth Sciences, 22:1262—1273.Lewis, Ti., W.H. Bentkowski, E.E. Davis, and R.D. Hyndman, 1988. Subduction of the Juan deFuca plate: Thermal consequences. Journal of Geophysical Research, 93: 15207—15225.Lewis, T.J., W.H. Bentkowslci, and R.D. Hyndman, 1992. Crustal temperatures near the Lithoprobesouthern Canadian Cordilleran transect. Canadian Journal of Earth Sciences. 29: 1197—1214213Majorowicz, J.A., D.I. Gough, and T.J. Lewis, 1993. Electrical conductivity and temperature in theCanadian Cordillera. Earth and Planetary Science Letters, 115: 57—64.McLean, N.A., 1994. M.Sc. thesis, Department of Physics and Astronomy, University of Victoria.(In preparation.)McLean, N.A., and G.D. Spence, 1992. Velocity structure from seismic refraction across the CoastPlutonic and Intermontane Belts [abstract]. Eos Transactions, AGU, Spring meeting supplement,73: 205.McMechan, G.A., and G.D. Spence, 1983. P-wave velocity structure of the Earth’s crust beneathVancouver Island. Canadian Journal of Earth Sciences, 20: 742—752.Misch, P., 1977. Dextral displacements at some major strike faults in the north Cascades. GeologicalAssociation of Canada, Annual Meetings, Program with Abstracts, 2: 37.Monger, J.W.H., 1985. Structural evolution of the southwestern Intermontane Belt, Ashcroft andHope map-areas; British Columbia. In Current research, Part A. Geological Survey of Canada,Paper 85-lA, pp. 349—358.Monger, J.W.H., 1990. Georgia Basin: Regional setting and adjacent Coast Mountains geology,British Columbia. In Current research, Part F. Geological Survey of Canada, Paper 90-iF, pp.95—107.Monger, J.W.H., 1991. Correlation of Settler schist with Darrington phyllite and Shuksan greenschistand its tectonic implications, Coast and Cascade mountains, British Columbia and Washington.Canadian Journal of Earth Sciences, 28: 447—458.Monger, J.W.H., 1993. Canadian Cordilleran tectonics: from geosynclines to crustal collage.Canadian Journal of Earth Sciences, 30: 209—231.Monger, J.W.H., and R.A. Price, 1979. Geodynamic evolution of the Canadian Cordillera—progressand problems. Canadian Journal of Earth Sciences, 16: 770—791.Monger, J.W.H., R.A. Price, and D.3. Tempelman—Kluit, 1982. Tectonic accretion and the origin ofthe two major metamorphic and plutonic welts in the Canadian Cordillera. Geology, 10: 70—75.Monger, J.W.H., and J.M. Journeay, 1992. Guide to the geology and tectonic evolution of thesouthern Coast Belt. Field Guide, Penrose Conference on the tectonic evolution of the CoastMountains orogen, May 1992. Geological Survey of Canada, Vancouver.Monger, J.W.H., and J.M. Journeay, 1994. Basement geology and tectonic evolution of the Vancouverregion, southwestern British Columbia. In Geology and geological hazards of the Vancouver region,southwestern British Columbia. Geological Survey of Canada Bulletin 481. (In preparation.)Monger, J.W.H., R.M. Clowes, D.S. Cowan, C.J. Potter, R.A. Price, and C.J. Yorath, 1994a.Continent-ocean transitions in western North America between latitudes 46 and 56 degrees. In214Phanerozoic evolution of North American continent-ocean transitions. Edited by R.C. Speed.Boulder, Colorado Geological Society of America, Centennial Transect Volume CTV-1. (liz press.)Monger, J.W.H., P. van der Heyden, J.M. Journeay, C.A. Evenchick, and J.B. Mahoney, 1994b.Jurassic-Cretaceous basins along the Canadian Coast Belt: Their bearing on pre-mid-Cretaceoussinistral displacements. Geology, 22: 175—178.Mooney, W.D., and T.M. Brocher, 1987. Coincident seismic reflection/refraction studies of thecontinental lithosphere: a global review. Reviews of Geophysics, 25: 723—742.Oldow, J.S., A.W. Bally, G.G. Ave Lallemant, and W.P. Leeman, 1989. Phanerozoic evolution ofthe North American Cordillera; United States and Canada. In The geology of North America: Anoverview, The geology of North America Volume A, pp. 139—232. Edited by W. Bally and A.R.Palmer, Geological Society of America, Boulder.O’Leary, D.M., R.M. Clowes, and R.M. Ellis, 1993. Crustal velocity structure in the southern CoastBelt, British Columbia. Canadian Journal of Earth Sciences, 30: 2389—2403.Parrish, R.R., 1983. Cenozoic thermal evolution and tectonics of the Coast Range of British Columbia:1. Fission track dating, apparent uplift rates, and patterns of uplift. Tectonics, 2: 601—631.Pavlis, G.L., 1986. Geotomography using refraction fan shots. Journal of Geophysical Research,891: 6522—6534.Perz, M.J., 1993. Characterization of the Fraser River Fault, southwestern British Columbia, andsurrounding geology through reprocessing of seismic reflection data. M.Sc. thesis, Department ofGeophysics and Astronomy, University of British Columbia, Vancouver.Phadke, S., and E.R. Kanasewich, 1989. Seismic tomography to obtain velocity gradients and three-dimensional structure and its application to reflection data on Vancouver Island. Canadian Journalof Earth Sciences, 27: 104—116.Phillips, W.S., and M.C. Fehier, 1991. Traveltime tomography: A comparison of popular methods.Geophysics, 56: 1639—1649.Ross, G.M. and R.R. Parrish, 1991. Detrital zircon geochronology of metasedimentary rocks in thesouthern Omineca Belt, Canadian Cordillera. Canadian Journal of Earth Sciences, 28: 1254—1270.Singh, S., and R.B. Herrmann, 1983. Regionalization of crustal coda Q in the continental UnitedStates. Journal of Geophysical Research, 88: 527—538.Spence, G.D., R.M. Clowes, and R.M. Ellis, 1985. Seismic structure across the active subductionzone of western Canada. Journal of Geophysical Research, 90: 6754—6772.Spencer, C., I. Asudeh, and T. Côté, 1989. SEGY-LDS Version 2.0 format reference document,manual version 1.00, Geological Survey of Canada, 17 pp.215Stacey, R.A., 1973. Gravity anomalies, crustal structure, and plate tectonics in the CanadianCordillera. Canadian Journal of Earth Sciences, 10: 615—628.Thurber, C.H., 1983. Earthquake locations and three-dimensional crustal structure in the Coyote Lakearea, central California. Journal of Geophysical Research, 88: 8226—8236.Thurber, C.H., and K. Aki, 1987. Three-dimensional seismic imaging. Annual Review of Earth andPlanetary Sciences, 15: 115—139.Toomey, D.R., G.M. Purdy, S.C. Solomon, and W.S.D. ‘Wilcock, 1990. The three-dimensionalvelocity structure of the East Pacific Rise near latitude 9°30’N. Nature, 347: 639—645.Um, 3., and C. Thurber, 1987. A fast algorithm for two-point ray tracing. Bulletin of theSeismological Society of America, 77: 972—986.van der Heyden, P., 1992. A middle Jurassic to Early Tertiary Andean-Sierran arc model for theCoast Belt of British Columbia. Tectonics, 11: 82—97.Varsek, J.L., F.A. Cook, R.M. Clowes, J.M. Journeay, J.W.H. Monger, R.R. Parrish, E.R. Kanasewich,and C.S. Spencer, 1993. Lithoprobe crustal reflection structure of the southern Canadian CordilleraII: Coast Mountains transect. Tectonics. 12: 334—360.Vidale, J.E., 1988. Finite-difference calculation of travel times. Bulletin of the Seismological Societyof America, 78: 2062—2076.Vidale, J.E., 1990. Finite-difference calculation of traveltimes in three dimensions. Geophysics, 55:521—526.Wenzel, F., and K.-J. Sandmeier, 1988. Reflectivity method for dipping layers. Journal ofGeophysical Research, 93: 15046-15056.Wheeler, J.O., and P. McFeely, 1991. Tectonic assemblage map of the Canadian Cordillera andadjacent parts of the United States of America. Geological Survey of Canada, Map 1712A, scale1:2,000,000, 2 sheets.White, D.J., and R.M. Clowes, 1984. Seismic investigation of the Coast Plutonic Complex—InsularBelt boundary beneath the Strait of Georgia. Canadian Journal of Earth Sciences, 21: 1033—1049.White, W.R.H., and J.C. Savage, 1965. A seismic refraction and gravity study of the Earth’s crustin British Columbia. Bulletin of the Seismological Society of America, 55: 463—486.White, W.R.H., M.N. Bone, and W.G. Milne, 1968. Seismic refraction surveys in British Columbia,1941—1 966: A preliminary interpretation. In The crust and upper mantle of the Pacific area,American Geophysical Union, Geophysical Monograph 12: 81—93.Wickens, A.3., 1977. The upper mantle of southern British Columbia. Canadian Journal of EarthSciences, 14: 1100—1115.216Zelt, B.C., R.M. Ellis, and R.M. Clowes, 1990. SCoRE ‘89: the Southern Cordillera refractionexperiment: Description of data set. University of British Columbia, Lithoprobe Secretariat,Lithoprobe Report 20, 38 pp.Zelt, B.C., R.M. Ellis, R.M. Clowes, E.R. Kanasewich, I. Asudeh, J.H. Luetgert, Z. Hajnal, A.Ucami, G.D. Spence, and R.D. Hyndman, 1992. Crust and upper mantle velocity structure ofthe Intermontane Belt, southern Canadian Cordillera. Canadian Journal of Earth Sciences. 29:1530—1548.Zelt, B.C., R.M. Ellis, and R.M. Clowes, 1993. Crustal velocity structure in the eastern Insularand southernmost Coast Belts, Canadian Cordillera. Canadian Journal of Earth Sciences. 30:1014—1027.Zelt, C.A., 1994a. 3—D velocity structure from simultaneous traveltime inversion of in-line seismicdata along intersecting profiles. Geophysical Journal International. (In press.)Zelt, C.A., 1994b. Three-dimensional velocity structure from wide-angle finitie-difference traveltimetomography. Annales Geophysicae. 12: C56.Zelt, C.A., and R.M. Ellis, 1988. Practical and efficient ray tracing in two-dimensional media forrapid traveltime and amplitude forward modelling. Canadian Journal of Exploration Geophysics,24: 16—31.Zelt, C.A., and R.B. Smith, 1992. Seismic traveltime inversion for 2-D crustal velocity structure.Geophysical Journal International, 108: 16—34.Zelt, C.A., and D.A. Forsyth, 1994. Modeling wide-angle seismic data for crustal structure:southeastern Grenville Province, Journal of Geophysical Research, 99: 11687—11704Zelt, C.A., and D.J. White. Crustal structure of the southeastern Canadian Cordillera from wide-angleseismic data. (To be submitted to Journal of Geophysical Research.)Zhou, D., A. Hasegawa, and S. Horiuchi, 1992. Tomographic imaging of P and S wave velocitystructure beneath northeastern Japan. Journal of Geophysical Research, 97: 19909—19928.217APPENDIX A Fan and Line 4 Record SectionsThis appendix displays trace normalized record sections for (i) the seven large fan shots(SPs 1, 3, 5, 4, 7, 12, 1) recorded along lines 1, 2 and 3; and (ii) all data recorded along line4 (from SPs 1, 3, 4, 7, 12, 17, 20, 21; see Fig. 5.1 for shot and line locations).2180-4It)—o:-c______________________—It)—co_ _ __________--rr —--—_ _ __—-‘ I I Iit)— 0) It)(s) 9/a-IFigure A.1 Record section for SP 1 into line 3. Data are trace normalized, band passfiltered from 3—15 Hz and plotted with a reducing velocity of 8 km/s. Horizontal scaleis shot-receiver azimuth measured clockwise from north. Nomogram showslocation of shotpoint (large dot) and recording line (heavy line).219z0CI,0CI,-- Q)--C\2- z:.00)rIJL!Eit) C, ‘-I-I(s)9/G—LJJFigure A.2 Trace normalized record section for SP 3 into line 2. SeeFig. A. 1 caption for other information.Ii0IC)CI,0U)CI)11E00CI0)0C’2220________0ct 0____-S2S= 0)= _0.-1sz— o—’_________ ____-— -____.--- -0 -• —-otOW_Ii I U 0)y‘ - lT1TI W -‘ I I I I IC1 0 cc cc(S) 9/a-IFigure A.3 Trace normalized record section for SP 5 into line 1. SeeFig. A.1 caption for other information.221_____ ___ ______0__-_ __ __= =;=— -—‘—-——-—.-..---.-..-..——-——.--v...’-—.-.’--_ _______ _____— -_ _ _—-—----- 1W— 0_= V2 -- 0r -U) *- .— --—‘fVI’ ‘v’-’------—._-—-———-.- —-‘-‘-----------—-“-—-_— I - - .“. L_-— —-—-.- —.— 0 • —ci)__ __ _ _____ _ _ _—0— __41 llPN- -0z C’20(s)/a-Figure A.4 Trace normalized record section for SP 4 into line 1. SeeFig. A.1 caption for other information.222C’20C.,00CoC.,0U]-,.Cl,• a)-•oz-C.,.1-I• N-o-o-C.,0 CD•0-oU)C’3(s) 9/a—iFigure A.5 Trace normalized record section for SP 4 into line 2. SeeFig. A. 1 caption for other information.2230— —_______• -_-p —— -.—_ _ ____ _ _0C\Z- 0-- —- ‘-I-)___ _______________çVV\j__ ____________ _ _ _ ___2rV_____ . V— —-r_CI) —_—-F- — — • —N-w—--ZE_— 0co=F_--_-‘-_ _ __ __ _0°- -z I ‘ I0 CD() 8/a—iFigure A.6 Trace normalized record section for SP 12 into line 1. Grey lines indicate theapproximate location of the three phases used in the 3—D analysis. P appears as thefirst arrival at azimuths 35°. See Fig. A.1 caption for other information.224________0D I — ‘-e -y v-— C..)-c — o.44 p.__ _ _ ________ _ _JwI t- ‘J / -_______________._-( —.---—- --0E-=‘-4-.--I-) __c -- z -- ‘—v 0_=-- -t5c ot - -——--—-‘—‘—.—---. v- --i-— .4_)‘t— ---‘.-‘ —v t i J.WVM W-__ _--—- 0•. =EEEEzE2 -\‘, .- ___ --93 &CPcM = = -flHH:__• • I IC) C..)(s)/a—iFigure A.7 Trace normalized record section for SP 12 into line3. See Fig. A.1 caption for other information.225o-—•.-fU)(s) 9/a—JFigure A.8 Trace nonnalized record section for SP 17 into line2. See Fig. A.1 caption for other information.0-CoCl,0—Cl,0-2C,,to-Q.)-0-COC20C’2226—-0—____ __________-1 0U)C)__ _-I_ _-T —(s) e/u—j1Figure A.9 Trace normalized record section for SP 17 into line3. See Fig. A. 1 caption for other information.2270-Af - -___.—v’ —H______0___ _ ____ _ _— 03—S-,’—--w .-_____ _No -0_CI)-_ _NC Co. =z— —_-__ ____- -——-_ _ _ _ _ ___ _ _ _,_—---_,—_,\ v.v ,—.—--—-‘-- — ‘v.— -‘ —‘-I- JWI --’.--.s—‘. I--v:—__=— 0.J4— .L—-- —-_—____-——,-- v-.’-r-)d —---—‘ — ——-‘ -- --—--—_______— —_0—= 03I I I I I I0 03 0() 9/a—iFigure A.1O Trace normalized record section for SP 7 into line1. See Fig. A. 1 caption for other information.228___________-—=N--)o o--0C)-N1.—I—“\ —-‘- t’- V.-WV-’f’_-—----—-——-—----—-----—---—-——---—-——-——________ _ ____ _ ____-——--——-.‘-.--- r-—- - -LI I I‘-I af) I(s) 9/a—IFigure A.11 Trace normalized record section for SP 7 into line2. See Fig. A. 1 caption for other information.229N__________I tLOI 0U)v vv v V V Vv S-(s) 9/U—JFigure A.12 Trace normalized record section for SP 7 into line3. See Fig. A.1 caption for other information.230T—D/8(s)0 I 0T—D/8(s)0 I 000—.-00oCD000 ,CD00.‘—‘—-—00CDCD C..<C..gCD-cn0..0I0VIV-*CD0QCDCD...—‘I.-.0CD •‘—,s-I00000z00 000,CD000I-’.00) 0 (b 0 -4 000 0) c-I (:10 -4 00013-12-11•CIJE-7-6-185 190Azimuth (deg)Figure A.14 Trace normalized record sections for (a) SP 3 and (b) SP 1 into line 4. Data aretrace normalized, band pass filtered from 3—15 Hz and plotted with a reducing velocityof 8 km/s. Horizontal scale is shot-receiver azimuth measured clockwise from north.Nomograms show location of shotpoints (large dot) and recording line (heavy line).SW Line 4 Shot 31110-9-8-7-6-5-4-cnENEiJT330I!• 1I350325I I INE335 340Azimuth (deg)Line 4 Shot 1345320qj SWI L ILI232Figure A.15 Trace normalized record sections for (a) SP 12 and (b) SP 4into line 4. See Fig. A.14 caption for other information.233LNE Line 4 Shot 12 swj108--64-3-I 111F’1111ti .[1. — T 1 1 i •80 90 100 110 120 130Azimuth (deg)sw Line 4 Shot 4 NEI21011w110 120 130 140Distance150(km)160 170 1809-8-7-6-5-4-3.III I I I I I I •—15 —101’rrI________________—5 0Azimuth (deg)Figure A.16 Trace normalized record sections for (a) SP 21 and (b) SP 20into line 4. See Fig. A.14 caption for other information.4 tsw Line 4 Shot 21 NEII110 120Distance (km)I • I I I I I I I80 90 1002.SWI I I I •130 140 150Line 4 Shot 209-8-7-6-5-4.3-2-NE(b)5 10234

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0052952/manifest

Comment

Related Items