PEAJ I PALI LTININCS ARTHQUAK NGINEERIN¢Geotechnical Earthquake EngineeringPRENTICE-HALL INIERNATIONAL SERIES IN CIVIL ENGINEERING AND ENGINEERING MECHANICS William J. Hall, Editor Au and Christiano, Fundemearals of Sirucutral Analysis Au and Christiano, Structural Analysis Barson and Rolfe, Fracture and Fatigue Control in Structures, 2/e Bathe, Finite Element Procedures in Engineering Analysis Berg, Elemenis of Structural Dynamics Biggs. Introduction to Structural Engineering Chajes, Structural Analysis. 2/e Chopra, Dynamics of Structures: Theory and Applications io Earthquake Engineering Collins and Mitchell. Prestressed Concrete Structures Cooper and Chen, Designing Steel Structures Cording et al.. The Art and Science of Geotechnical Engineering Gallagher, Finite Element Analysis Hendrickson end Au, Project Management fne Comstruction Higdon clal., Lagineering Mechanics, 2nd Vector Edition Hultz and Kovacs, Introduction in Geotechnical Engineering Humar, Dynamics of Structures Johnston, Lin, and Galarmhos. Rasie Sie! Design, 30 Kelkar and Sewell, Fundamentals of the Analysis and Design of Shell Siructures Kramer. Geotechnical Karthquake Fngineering MacGregor, Reinforced Concrete: Mechanics and Design, 2/e Mehta and Monteiro, Concrete: Structure, Properties and Materiais, 2f Melosh, Structural Engineering Analysis by Finite Elements Meredith et al.. Design and Planing af Engineering Systems, Ife Mindess and Young, Concrete Nawy. Prestressed Concrete Nawy. Reinforced Concrete: A Fundamental Approach, 2e Pieffer, Solid Waste Management Popov, Engineering Mechanics of Solids Popov, Introduction ia the Mechanirs af Salidy Popov, Mechanics of Materials, 2/e Schneider and Dickey, Refaforced Macomry Design, 2/0 Wang and Salmon, Introductory Structural Analysis Weaver and Johnson, Strucisral Dynamics by Finite Elements Wolf, Dynamic Soil-Structure Interaction Wray, Measuring Fagineering Praporties of Soily Yang, Finite Elemeni Structural AnalysisGeotechnical Earthquake Engineering Steven L. Kramer Uni itv of Washinaton University of Washington Prentice-Hall Internaticual Series in Civil Engineering and Engineering Mechanics Prentice Hall Upper Saddle River, New Jersey 07458Library uf Comgress Catuloping-in-Publication Bata Krasner, Steven Lawrence, Geotechnical cartkouake cogincering ( Steven L. Kramer. D. em, — (Freier Holl ieil enginnoring and mechanics seuies) Includes bibliographical references and index. ISBK (-13-374944.6, 1, Earthquake engincering 2. Bagincering geology. I. Title. IL Series TARS4.6,K72 1996 624.1762) 95-3020 cP Acquisitions editor: BILL STKNOUIST Production editor: FRWIN ZUCKER Cover director: JAYNE CONTE, Buyer: JULIA MEFHAN Fatitorial assistant: MEG WEIST 51996 by Prentice-Hall, Inc. Simon & Schuster /A Viacom Company Upper Saddle River, NJ 07458, All rights reserved. No part of this book may be reproduced, in any Jorn orby any means, without pecmission ia writing trom the publisher, The author and publisher of chis hook have used their best efforts in preparing this hock. ‘These effets include the development, reseetch, and testing of the thoeries and programs to determine their effectiveness. The aundvor and publisher make no warranty of any kind, expressed or implied, with repnrd to these programs or the decamentation contained in this book, The author und publisher shall aot be lable in any event for incidental or consequential daniages dia ous tiv with, us arising Out o4. Ine furnishiag, perterrmance, or use of these prorams, Prinied in the United States of Auncrica 09876543 ISBN O-13-374943-b Premice-Fall International (UK) Limized, London Prentice-Hall of Anstalia Pty Timited. Sydney Prentice Hall Canada fae., Toronto Preutice-Hall Hispanoamicricana, $.4., Mexico Prentice-Hall ol India Private Limited, New Delhi Prentice-Hall f Japan, In. Tekye Simon & Sch Faitora Prentice-Hall de Brasil, Lida., Rie de Janeiro Peo, Lid, SingaganeTo my parentsContents Preface Introduction to Geotechnical Earthquake Engineering Ll 12 13 Introduction 1 Background i Seismic Hazards 2 43.4 Ground Shaking, 2 13.2 Stracteral Hazards, 3 133 Liquefaction, 5 134 Landslides, 9 135 Retaining Structure Failures, 1 13.6 Lifeline Hazards, 1 13.7 Tsunami and Seiche Hazards, 13 Mitigation of Seismic Hazards 14 Significant Historical Earthquakes 14 ayviii Content Scismotoay and Earthquakes i 21 Introduction 18 2.2 Internal Structure of the Earth 18 22.2 Seismic Waves, 19 2 Internal Structire, 20 23 Continental Drift and Plate Tectonics 23 23.4 Plate Tectonics, 24 2.3.2 Plate Bowxdaries, 20 24 Faults 33 24.1 Fault Geometry, 33 24.2 Fault Movement, 34 25 Elastic Rebound Theory 36. 25.4 Relationship to Earthquake Recurrence, 39 2.4.2 Relatienship te Tectonic Environment 41 25.3 Seismic Moment, 42 2.6 Other Sources af Seismic: Activity 42 27 Geometric Notation 43 228 of Bartha 44 29 Size of Earthquakes 29.4 Burshqueke Intensity, 45 2.9.2 Barthquake Magnitude, 46 2.9.5 Barthyuake Energy. 50 2.10 Summary 51 Strong Ground Motion 3.1 32 3.3 Tntroduction 54 Strong-Motion Measurement 56 3.21 Seismographs, 56 3.2.2 Data Acquisition and Digitization, 59 3.24 Strong-Morion Processing, 6f 3.24 Strong-Morion Instrument Arrays, 62 3.2.5 Strong-Motion Records, 64 Ground Motion Parameters 65. 331 Amplinde Parameters, 66 3.3.2 Frequency Content Parameters, 70 3.3.3 Duraiion, 79 3.3.4 Other Measures of Grownd Motion, 82 33.5 Discussion, 8454 5.6 Contents ix 34 Estimation of Ground Motion Parameters 84 34.7 Magnitude and Distanee Effects, 85 3. Development of Predictive Relationships, 86 Estimation of Amplitude Parameters, 88 Estimation of Frequency Content Parameters, 91 Estiraation of Duration, 95 Estimation of Other Parameters. 95 35 Spatial Vatiability of Ground Motions 100 3.6 Summary 10°. 4 = Seismic Hazard Analysis 106 41 Introduction 106 42 ion and Gvaiuation of Barthquake Sources LU/ Geologie Evidence, 107 L22 Tectonic Evidence, 115 4.2.3 Historical Serwmicity, 13 424 tustiumentod Seismicity, 114 Deterministic Seismic Hazard Analysis 114 Probabilistic seisrme Hazard Analysis 117 44.1 Earthquake Souree Characterization, HS 442 Predietive Relationships, 126 44.3 Terapotal Uncertainty, 127 444 — Probabitity Computations, 129 45 Summary 138 5 Wave Propagation 143 Antroduction 143 Waves in Unbounded Media 144 32.4 One-Dimensional Wave Propagation, 144 52.2 Three-Dimensianal Ware Propagation, 40 Waves ina Semi-infinite Body 156 TIT Rayleigh Wey 532 Love Waves, 162 533 Higher-Mode Suxface Weves. 164 5.3.4 — Dispersion of Surjace Waves, 164 SFT Phisce and Cres Waves in a Layered Rody 165 S47 One Dimensional Case, Muterial Bouniory in an nfinie Kod, 10> 542 Three Dimensional Case: Inclined Waves, 170 Attenuation of Suess Waves 174 53.1 Material Damping, 175 5 Ruddliution Damping, (79 Summary 180Content Dynamic Soil Properties 186 61 Introduction 184 62 Representation of Stress Conditions by the Mohr Circle 185 6.2.2 Principal Stresyes, 187 6.2.2 Stress Paths, 188 63, Measurement of Dynamic Soil Properties 191 G34 Fieid Fests, 19 6.3.2 Laboratory fests, 215 6.3.3 Interpretation uf Observed Ground Response, 226 6.4 Stress-Strain Behavior of Cyclically Loaded Soils 228 6.4.1 Some Baste Aspects of Particulute Matter Behavior, 228 64.2 Equivalent Linear Model, 230 845 Cycle Nontnear Moaets, 240 6.4.4 Discussion, 244 65 Strength ot Cyclically Loaded Soils 244 6.5.1 Definitions of Failure, 244 6.5.2 Cyclic Strength. 245 6.53 Monotonic Strength, 246 bo Summary 248 Ground Response Analysis 25¢ TL Introduction 254 72 One-Dimensional Ground Response Analysis 255 7.2.1 Linear Approach. 256 7.2.2 Nonlinear Approach, 275 7.2.3 Comparison af One-Dimensional Ground Response Analyses, 279 73 ‘Two-Dimensional Ground Response Analysis 280 731 Pynamic Finite-Elemeni Analysis, 281 7.3.2 Equivalent Linear Approach, 284 72% Nonlinoar Appeoach, 286 7.3.4 Other Approaches to Two-Dimensional Ground Response Problems, 286 FAS Comparison of Two-Dimensional Ground Response Analysev, 20) 74 Three-Dimensional Ground Response Analysis 291 A} Bquivatent Linear Finite Blemom Approach, 292 742 Nonlinear Finite-Flement Approach, 292 2 Shear Boam Approach, 292 7.44 Comparison of Three-Dimensional Ground Response Analyses, 294 1S Soil+Structure Latevaction 294 7.5.4 Hlusivation of Soii-Structure Tateraciion Bffects, 295 7.5.2 Methods of Anatysis, 300 76 Summary 303Contents xi a Local Site Effects and Desian Ground Motions 208 8.1 Inireduction 308 8.2 Effects of Local Site Conditions on Ground Motion 309 82.4 Evidence from Theoretical Ground Response Analyses, 309 $2.2 Rvidence from Measured Amplification Functions, 310 8.2.3 Evidence from Measured Surface Motions, 312 82.4 Compilations of Data on Local Site Effects, 317 82.5 — Effects of Surface Topography and Basin Geometry, 319 83 Design Parameters 323 8.3.2 Design Karthquakes, 324 83.2 Design Spectra, 325 84 Development of Design Parameters 327 8.4.4 Site-Specific Development, 327 84.2 Code-Based Development, 328 8.5 Development of Ground Motion Time Histories 340 85.1 Medification of Acival Ground Motion Records, 340 83.2 Time-Domain Generation, 341 85.3 Frequency-Domain Generation, 744 85.4 Green’s Function Techniques, 343 &5.5 Limitations of Artificial Ground Matians, 245 86 Summary 345 Liquefaction 348 348 -iquefaction-Related Phenomena 349 G2) Plow Liquefaction, 349 922 Cyclic Mobility, 349 9.3 Evaluation of Liquelaetion Hazards 350 9.4 Liquefaction Susceptibility 351 ¥44 Historical Criteria, 332 94.2 Geologic Criteria, 353 Y43 Compositional Criteria, 354 944 State Criteria, 355 9.5 Initiation of Liquefaction 361 95.2 Flow Liquefaction Surface, 361 9.5.2 Influence of Excess Pore Pressure, 366 95.3 Evaluation of Initiation of Liquefaction, 368 9.6 Effects of Liquefaction 397 9.6.4 Alieration of Ground Motion, 398 9.6.2 Development of Sand Boils, 400 96.3 Sottlemen:, 402 9.6.4 — Instability, 408 97 Summary 417xii 16 WwW Contents Seismic Slope Stability 423 10.1 introduction 423 10.2 Types of BartuquakeInduced Landslides 424 10.3 Farthquake-lnduced J.andshde Activity 426 10.4 Evaluation of Slope Stability 429 10.5 Static Slope Stability Analysis 430 10. Limit Equilibria Analysis, 430 10.5.2 Stress. Deformation Analyses, 433 10.6 Seismic Slope Stability Anal 10.0.1 Analysis of Inertial Instability, 433 10.6.2 Analysis of Weakening fastabitity, 450 10.7 Summary 462 Seismic Design of Retaining Walls 466 Wd Introduction 466 11.2 Types of Retaining Walls 466 113 Types of Retaining Wall Failures 467 ihe Static Pressures om Reiaiving Walls 469 114.1) Raakine Theory, 469 1142 Cotlomb Theory, 472 1143 Logarikmie Spiral Method, 474 1144 Stress-Depormation Anatysts, 176 LLS Dynamic Response of Retaining Walls 477 110 Seismic Pressures on Retammg Walis 4// 1161 Yielding Walls, 478 11.6.2 Nonvielding Waits. 434 11.6.3 Effects of Water on Wall Pressures, 486 116.4 Finite-Elemeni Analysis, 489 iL? Seismic Displacements of Relaming Walls 489 ALFA Richards—Plms Mettexd, 489 17.2) WhitmanLiao Method, 192 4173 Finite-Element Analysiv, 493 IL Seismic Design Considerations 494 ESRD Grnwity Walls, 494 11.8.2 Cantilever Walls, 495 JL&3 Braced Walls, 495 4184 Reinforced Soil Wally, 500 WS Summary 503Contenes 12 Soil Improvement for Remediatinn af Sei: so¢ 12.1 Introduction 506 12.2 Denstitcation‘Lechniques 507 12.24 Vibro Techniques, 508 122.2 Dynamic Compaction. 540 12.2.3 Blasting, 512 12.24 Compaction Grouting, 513 42.2.5 Areal Extent of Densification, S14 12.3 Reinforcement Techniques 515 i231 Stone Columns, 515 12.5.2 Compaction Piles, 316 42.3.3 Drilled Inclusions, 516 Grouting and Mixing Techniques 517 4244 Grouting, 518 1242 Mixing, 919 12.5 Drainage Techniques 521 12.6 Verification of Soil Improvement 322 T2612 laboratory Tasting gues 72.6.2 In Site Testing Techniques, 523 12.6.3 Genphysical Fosting Techniques, $23 12.7 Other Considerations 524 12.8 Summary $24 A Vibratory Motion 527 Ad introduction 527 Al Types of Vibratory Motion 527 Alt Simple Harmonic Motion, 528 A22 Trgonomeiric Notation for Simple Harmonic Motion, 529 A23 Other Measures of Motion, 533 AS lourter Series 536 AG] > Trigonomemic Form, 536 A32 Exponential Form, 539 AB3 Discrete Fourier Transform, 544 A34 Kast Pourier Transform, 541 ABS — Power Spectrum, 342 B Dynamics of Discrete Systems 543 BL Introduction 543 B2 Vibrating Systems $44 B33 Single-Degree-of-Treedom Systems 544xiv Content B4 Equation of Motion for SDOF System 545 BA] Equation of Motion: External Loading, 545 B42 Equation of Motion: Vibration of Supports (Base Shaking), 547 BS Response of Linear SDOF Systems 547 B51 Undamped Free Vibrations, 548 BS5.2 Damped Free Vibrations, 551 B53 Response of SDOF Systems to Harmonic Loading, 553 B54 — Response af SDOF Systems to Periodic Loading, 561 BSS Response of SDOF Systems to General Loading, 564 B6 Damping 567 B61 Viscous Damping, 567 B62 Other Measures of Energy Dissipation, 569 8.6.3 Complex Stiffness, 570 B7 Response Spectra 571 B.8 —— Respanse of Nonlinear SOF Systems to General Loading $72 B81 Increraensal Equation of Motion, 572 RAD Niwmevical Inegration, S74 BO Multiple-Degree-of-Freedom Systems 375 B.9.2 Undamped Free Vibrations, 577 B93 Mode Superposition Method, 579 B94 — Response Spectrum Analysis, S81 BYS Discussion, 382 Probability Concepts 58: Cd Introduction 583 C2 Sample Spaces and Events 383 C3 Axioms of Probability 584 cA Probabilities of Events 585 cs Random Variables 588 C.6 Expected Values and Standard Deviations 589 C7 Common Probability Distributions 590 C72 Uniform Distribution, 590 C72 Normal Distribution, 591 C73 Lognormal Distribution, 504 feferences Sat index 64:Preface Compared to most disciplines of civil engineering. geotechnical earthquake engineering is quite young, While the damaging effects of earthquakes have been known for centuries, the Strong contribution of soils to the magnitude and pattern of earthquake damage was not widely appreciated until relatively recently. Following damaging curthquakes in 1964 in Niigata, Japan and Alaska, and spurred by the growth of the nuclear power indnciry in the 1960s and 1970s, the ficld of geotechnical earlhquake engineering has grown rapidly. Although much remains to be learned, the field has matured to the paint where generally accepted theories and analytical procedures now cxist for many important problem The purpose of this book is to introduce the reader to the concepts. theories, and pro- cedures of geotechnical earthquake engineering. It is intended for use as a text in graduate courses on geotechnical carthquake engingering and as areference hook for practicing eng’ neers. Recognizing that geotechnical earthquake engineering is a broad, multidisciplinary field, the hook draws from seismology. geology. structural engineering, risk analysis, and other technical disciplines. The book is written at a level suitable for students with knowledge equivalent to that of a senior (fourth-year) civil engincering student, The student should have had basic courses in soil mechanics, structural cngineering. and hydrantics: iatndictery cov 2 geology and probability/statistics would also be helpful. Many graduate students will havexvi Preface had courses in structural dynaraics or soil dynamics by the ime they begin study of geo- é al cartinyuake engineering. For those readers without pnor exposure, introductions to the nomenclature and mathematics of dynamic systems, structural dynamics, and prob- ability are presented in three appendices. ORGANIZATION The subject matter falls into two main categories. The appendices and the first six chapters pr'scnt fundamental principles of seismology, ground motion, dynamics, and soil behavior. Applications of these principles to the practical problems most commonly encountered in geotechnical earthquake enginccring practice are presented in the last six chapters, Chapter | introduces the reader to the types of damage that can accur during earth- quakes and to the problems they present to geotechnical earthquake engineers. Basic con- cepts of carthquake seismolugy and the terminology used to describe earthquakes and their effects are described in Chapter 2. Chapter 3 describes pround motion measurement. the parameters used to characterize strong ground motion, and methods for prediction of those Parameters. Deterministic and probabilistic seismic hazard analyses are presented in Chap- ter 4. Chapter 5 introduces the reader ty wave. propagation, beginning with simple one- dimensions] body waves in homogencous materials and extending to surface waves and ainltidimensional, layered systems. The properties of soil that control their wave propaga- tion behavior are described in Chapter 6. Field and laboratory techniques for measurement of these propertics are also described, Chapter 7 presents methods for analysis of ground response during earthquakes, begin- ning with one-dimensional ground response analysis and moving through two- and three- dimensional dynemic response analyses. Both frequency- and time-domain approaches are described. Chapter 7 concludes with an introduction to the basic concepts and etfects of sail- structure interaction. The effects of local soil conditions on ground motions and earthquake damage are described in Chapter 8. Chapter 8 also introduces the concept of design ground motions, and how they are obtained from site-specific analyses and from building codes, Chapter 9 deais with liquefaction—it begins with a conceptwal framework for understanding various liquefuction-related phenomena and then presents practical procedures for evalua- tion of liquefaction hazards, Seismic stability of slopes is covered in Chapter 10. and seismic Cesign of retaining structures in Chapter 11. Chapters 10 and 11 address their respective topics initially from pseudo-static and theu from permanent displacement standpoints. Chapter 12 introduces commonly used soil improvement techniques for mitigation of seis- mic hazards, PEDAGOGY ‘This book is the first to deal explicitly with the topic of geotechnical earthquake engineer- ing. During its preparation, a great deal of time and effort was devoted to decisions regard ing content and organization. The final form naturally reflects my own preference, but the text has been reviewed hy many engin . Preparation of the text also involved a great deal of interpretation of information from a from both acadamia and professionalPreface xvii wide variety of sources. While the text reflects my own interpretation of this information, it is heavily referenced to allow readers to explore background or more detailed information on various geotechnical earthquake engineering topics. A couple features are noteworthy. Two ground motions from the Loma Prieta earth- quake, one from a rock outerop and one from the surface of a nearby deep soil deposit, are used to illustrate a number of concepts throughout the book. Differences in the amplitudes, frequency contents, and durations of the motions are emphasized in Chapter 3. The reasons for these differences later become apparent in Chapters 7 and 8. The book also emphasizes the use of transfer Functions, particularly in the solution of ground response problems. The transfer function approach helps students form a more complete understanding of ground response—in the frequency domain as well as the time domain. With the advent of com- puter programs such as MATLAB, MathCad, and Mathematica, the Fourier analyses required in the transfer funclion approsch are quite simple; students usc MATLAB exten- sively in my soil dynamics and geotechnical carthquake engineering courses. ‘the book contains worked examples and homework problems. The example prob- Jems are intended to illustrate the basic concepts of the problems they address; Lo allow the results to be checked, a number involve calculations carried out to more significant figures than the accuracy of the procedures (and typical input data) would justify. Many of the important problems of geotechnical earthquake engineering, however, do not lend them- selves to the type of short, well-defined homework problem that is readily placed in a book. My preference is to assign longer, ptoject-oriented assignments bascd on actual case histo- sles, and Erecommend that the homework problems in this book be supplemented by such assigninents. Asin many other fields, the nse of units in geotechnical earthquake engineering is neither uni form nor consistent. The current stale of knowledge in geotechnical earthquake engineering has resulted from advances in a variety of technical fields and a variety of countries, many of which customarily use differeat units, Fortunately, most conform 0 to relatively standard met- ric or Rritish cystoms: Rather than attomy er book uses dual units. In recognition of their origins, the most common units for each quantity is listed first with the. alternative fallowing in parentheses. The approach is intended to allow all readers ( proceed through the matcrial without stopping to convert (mentally or other- wise) from one set of units ta another. To encourage familiarity with both sets of units, some example and homework problems ate specified in meuic units and some in British units. ACKNOWLEDGMENTS A number of people have helped dircctly and indirectly in the preparation of this book. The professional and academic portions of my career fave benefited greatly from a number of people that I have worked closely with, particularly (in chronological order) Bill Houston, Tom Tejima, ILB. Scud, Jue Mabuucy, and Bod Hiv. Their assisiance, advice, and encouragement has taken many forms, and T am grateful to each.xviii Preface As this book evolved from a collection of lecture notes and handouts, it was contin- y improved Uy conupents and suggestions from many students im my soul dynamics and nechnical extthquake engineering courses. Their assistance is greatly appreciated. Tam also grateful to many colleagues who provided constructive critical reviews of different parts of the book. including Dr. Donald G. Anderson, Dr. Juan Baez, Mr. David Baska, Dr, Gopal Biswas, Prof. Ross W, Boulanger, Dr. C.B. Crouse, Prof, Emeritus William J. Hall, Ms. Karen Henry, Prof. Carlton L. Ho, Prof. William D. Kovacs, Prof. Roberto T. Leon, Prof. Crogiiy R. MacRa, Dr. Lelio 1. Mejia, Dr. Robert ryke, Frof, Peter K. Kobertson, Prot. Raj Siddharthan, Prof. Stewart Smith, Prof, Timothy D, Stark, and Prof. George M. Vurkiy yal Dacl uxaue: suggestions that improved the guality af the book. Prot, Geotirey R. Martin and Prof. T. Leslie Youd reviewed substantial Portions of the book; their efforts are pwticularly appreciawd. Finally, (am most grateful to my wife, Diane, and to my daughters, Katie and Megan cat » BOK involved several years of iong Working hours: | could not have done it without their cheerful understanding and cnconragement. ation ulIntroduction to Geotechnical | Earthquake Engineering 1.1 INTRODUCTION Earthquake engineering deals with the effects of earthquakes on people and their environ ment and with methods of reducing those effects. It is a very young discipline, many of its most important developments having occured in the past 30 to 4 vear Barthaual neering 1s a very broad ticld, drawing on aspects of gcology, s ismology, geotechnical enginecring, structural engi neering, risk analysis. and other technical fields. Its practice also requires considcration of social, economic, and Political factors. Most earthquake engincers have entered the field from structural engineering or geotechnical engineering back- Srounds, a fact that is reflected in the practice of earthquake engineering, This book covers geotechnical aspects of earthquake engineering. Although its primary nical engineering students and Practitioners, it contains a great deal of information that should be of interest to the structural engincer and the engineering seismologist. geotceh- 1.2 BACKGROUND The study of earthauakes dates hack en acvo.ds of earthquakes in China date as fax back as 3000 years. Japanese records and records from the caster Med- iterranean region go back nearly 1690. years. Tn the United States the histurival record of2 Introguction to Geotechnical Earthquake Engineering Chap. 1 earthquakes is much shorter, about 350 years. On the seismically active west coast of the United Sites, earthquake records go back only about ZOU years. Compared with the mil- lions of years over which earthquakes have been occurring, umankind’s experience with car thyuckes is very brief. Today, hundreds of millions of people throughout che world live with a significant tisk to their lives and property from earthquakes. Billions of dollars of public inirastructare are continuously at risk of earthquake damage. The health of many local, regional, andeven aaiivual ccomomies are aiso at risk from earthquakes, Lhese t1sks are not unique to the United States, Japan, or any other country. Earthquakes are a global phenomenon and a glo- Lal probiem, Earthquakes have occurred for millions of years and will continue inthe future as they have in the past. Some will occur in remote, undeveloped areas where damage will be neg- ligible. Others will occur near densely populated urban arcas and subject their inbabitants aud iv iuftasteucture they depend on to strong shaking. 11 28 impossible to prevent earth- quakes from occurring, but it is possible to mitigaic the effects of strong earthquake shak- ing: Wo reduce loss of life, imuntes, and damage. 1,2 SEISMIC HAZARDS A number of naturally ovcurring everits. such as earthquakes, hurricanes, tornados, und floods. are capable of causing deaths, injuries, and property damage. These natural hazards cause tremendous damage around the world cach year, Hazards associated with earthquakes are commonly referred to as seismic hazards. The practice of earthquake engineering involves the identification and mitigation of seismic hazards. The most important seismic hazards are described in the following sections. 1.3.1 Ground Shaking When an earthquake occurs. scismic waves radiate away from the source and travel rapidly through the carth’s crust. When these waves teach the ground surface. they produce shaking thar may last from seconds to minutes. The strength and duration of shaking at a particular site Gepends on the size and location of the earthquake and on the characteristirs af the cite Ar sites near the source ol'a large earthquake, ground shaking can cause tremendous damn age. In fact, ground shaking can be considered tn he the most important of all seismic ha ards becanse al! the ether hazards are caused by ground shaking. Where ground shaking levels are low, these other seismic hazards may be low or nonexistent. Strong ground shak- ing, however, caa produce extensive damage from a variety of seismic hazurds. Although seistnic waves travel through rock over the overwhelming majority of their trip from the souree of an earthquake to the ground surface, the final portion of that wip is often through soil. and the characteristics of the soil ean grearly infmence the nature af chak- ing al the ground surface, Soil deposits tend to act as “filters” to seismic waves by attenu- ating motion at certain frequencies ane amplifying it at others. Since soil conditions often vary dramatically over short distances, levels of ground shaking can vary significantly within a stnall area. One of the mos! important aspeers of geotechniral casthqnsike engin: ing practice involves evaluation of the cffects of local soil conditions on strong ground motion. In this book. Chapter 3 presents merhads for quantifying the mést importantSec. 1.3 Seismic Hazards 3 characteristics of strong ground motions, and Chapters 4 through 7 provide the hack ground and techniques for site-spectBie ground motion prediction. 13.) COL Waza ‘Without doubt the mos! dramatic and memorable images of earthquake damage are those of structural collapse. From the predictable collapse of the unceinforced masonry and adobe structures in which many residents af underdeveloped areas of the world live (Figure 1.1) to the surprising destruction of more modern construction (Figures 1.2 to 1.4), structural dam- age is the leading cause of death and economic loss in many earthquakes. However, struc- tures need not collapse to cause death and damage. Falling objects such as brick facings and parapets on the outside of a structure or heavy pictures and shelves within a structure have caused casualties in many eatthquakes, Interior facilities such as piping, lighting, and stor- age systems can also he damaged during o: . Over the years, considerable advances have been made in earthquake-resistant design of structures, and seismic design requirements in building codes have steadily improved. As carthquake-resistant design has moved from an emphasis on structural strength to emphases on both strength and ductility, the need for accurate predictions of ground motions has increased. In current design practice, the geotechnical earthquake enginecr is often respon- sible for providing rhe strmetural engi fa this book, Chapter 8 describes the effects of local soil conditions on ground motions and pro- vides guidance for the develnpment of site-specific design ground motions. sign gro} Figere 1.1 Damage to buildings in Huaras, Peru following the 1970 Peru earthquake. The edobe structures in the foreground were destroyed, but the reinfoveed concrete suucture ju ue Lavkyruural suffered ine damage (photo by G, Platker, courlesy of USGS).Figure 1.2 Collapsed portion of the reinforced concrete Hospital Jaarez in Mexico City following the (988 Mexico caithquake (photo by E.V. Leycncecker, courtesy of BERI). Higure 13 Effects of volumn failures at Olive View Hospital in the 1978 San Fernando earthquake, Collapse of the canopy in the foreground pinned che ambulances beneathSee. 1.3 Seismic Hazards 5 Figure Lt Reinfued euusiute cola a Olive View Hospital following the 1971 San Fernando carthquake, Insufficiest a cansverse reiatorcement was unable to ak “4 : provide adequate confinement (courtesy of oe ( Unu 1.3.3 Liquefaction Some of the most spectacular examples of carthquake damage have ocurred when soil deposits have lost their strength and appearcd to flow as fluids. Tn this phenomenon, tezmed liquefaction, the strength of the soil is reduced, often drastically, to the point where it is , stable. Because it Unly uecurs iu saiuraicd soiis, Tig uefaction is most commonly obscrved near rivers, bays, and other bodies of water, The term liquefaction actually cncumpasses several celated phenuinena. Flow fail ures, for example, can cecur when the suength of the soil drops below the level needed to maintain stability under static conditions. Plow failures are thecefore driven by static grav- itational forces and can produce very large movements. Flow failures ave caused the col- ( } no (Figure 1 ar and other slopes, aud dhe failure of foundations (Figure 1.6) ‘The 1971 San Fetnando earthquake caused a flow failure in the upstreath slope of the Lower San Fenando Dam (Figure 1.7) thal nearly breached Os dau, Thousands could have been: killed in the residential arca immediately below the dam, Laieral spreading is a related phe- nomenon characterized by incremental displacements during earthquake shaking. Depend- ing on the number and strength of the stress pulses that exceed the strength of the soil, lateral eprea a bat tauge ‘uu usgiigibic w quite Targe. Laceral = spreading is quite common near bridges, and the displacements il produces can damage the abutments, foundations, and superstructures uf bridges (Figures 1.8 and 1.9). Finally, theFigure 4.5. Liquefaction (ailure of Sheffield Dam following the 1925 Se conthquake (K. Steinbrugge collection; courtesy of FERC. Univ, of California). Figure 1.6 Liquefaction-induced bearing capacity failures of the Kawagishi-che apartment buildings (ollovein the L964 Niigata earthquake (ccurtesy of USCS)Sec. 1.3 Seismic Hazarcls ? Figure £7 Lower San Fernando Dam following liquefaction faiture of its upstecam slope in the 197] San Fernando earthquake (K. Steinbrugge collection: courtesy of ERC, Univ, of California), figure LS tfect of lateral spreading on a small bridge in Japan following the 1952 ‘Tokarhi-Oki earthquake. Laterel sproading of the soil at the abutment buckled the bridge deck (X Steinkrugge collection; courtesy of EERC, Univ. of Califunnia)8 Introduction to Geotechnical Earthquake Engineering Chap. | Figure 1.9 The Showa Bridge following the 1964 Niigata earthquake, Tatcral spreading caused bridge pier foundations to move and rotale sufficiently for simply supported bridge spans to fall ¢courtesy of US phenomenon of ievel-ground liquefaction does not involve ku ge lateral displar cuca Lut is cusily identified by the presence of simd boils (Figure 1.10) produced by groundwater rushing Ww the surface. Although not particularly danuging by themselves, sauud boils indi- cate the presence of high groundwater pressures whose eventual dissipation can produce subsidence and damaging differential sewlements, Liqucfaction is a complicated phenomenon, but research has progressed to the point where an integrated framework of understanding can be Ueveluped. Chapicr 9 uf tis Louk presents the basic concepts with which the susceptibility, triggering conditions, and effects of all liquetaction phenomena can be understood, together with practical procedutes for evaluation of liquefaction hazards. 1.3.4 Landsl jes Strong earthquakes often cause landslides. Although the mayority of such landslides are small, earthquakes have also caused very large slides. In a number of unfortunate cases, earthquake-induced landslides have buried cntire towns and villages (Figure 1.11), More. commonly, earthquake-induced landslides cause damage by destroying buildings, or dis- rupting bridges and other constructed facilities (Figures !.12 and J.13). Many earthquake- induced tandstides result from liquefaction phenomena, but many others simply represent the failures of slopes that were marginally stable under static conditions. Various types of seismic slope failures, their frequency of occurrence, and procedures for their analysis are described in Chapter 19.Sec. 1.3 Seisrnic Hazards 9 igure 1.10 Sand boil in rive ficld following the 1964 Niigata earthquake (XK. Steintrugge collection; courtesy of FERC, Univ. of California}. 1.3.5 Retaining Structure Failures Anchored bulkheads, quay walls, aud uther retaining structures ace frequently damaged in earthquakes. Damage is usually concentrated in waterfront areas such as ports and harbors (Figure 1.14). Because such fuvilitics are ufien essential for the movement of goods upon which local economies often rely, the business losses associated with their failure can ge far d the costs of opaiy os ssvousiuction, The seismic design of retaining structures is covered in Chapter HL. 1.3.6 Lit ine Hazards A network of facilities that provide the services required for commerce and public health can be found in virtually any developed atez. These networks, which include electrical power and teiecommunications, wansporlation, water and sewage, oil and gas distribution, and waste storage systems, have collectively come to be known as lifelines. Lifeline systems inay include power plants, transmission towers, and buried electrical cables; roads, bridges, harbors, and airports; water treatment facilities, reservoirs and elevated water tanks, and bur- ied water distribution systems; liquid storage tanks and buried oil and gas pipelines; and municipal solid waste and hazardous waste landfills. Lifeline systems and the facilities that comprise them provide services that many take for granted but which are essential in modern industrial arcas. Lifeline failures not only have severe economic consequences but can also adversely affect the environment and quality of life following an earthquake.Figure 111 Village of Yungay, Peru, (2) before and (b) after being buried by a giane inasiide in the 1940 reruvian eariiquake. Tite saiue pair nurs coe visible at the side of both pholographs, The landslide involved 59 million cubic meters of material that exentuailly covered an area of some RON sare Kilometers, About 25.000 peoplé were killed by this landslide, over 18.000 in the villages of Yungay and Ranrabires (K. Steinbrucee collection: courtesy oF FERC. Univ, of California).Figure 1.12 A wing of Government Hill School in Anchorage, Alaska, straddled the head searp of the Government Hill landslide in the 1964 Good Friday ceathquake (K. Steinbrugge collection; courtesy of EERC, Univ. of California). Figure 113. Farhquake-indluced landslide along railroad tracks near Olympia, Washington (photo by G.W. Thorsen). 4112 Iniroductian to Geotechnical Earthquake Engineering Chap. | Vigure 1.14 Failure of » quay wall on Rokko Tsland in Kobe, Japan in the 1995 Hyoge- Ken Nanbu curthquake (photo by 8. L. Kramer), of repairing facilities directly damaged by earthquake shaking. The 1989 Loma Prieta and. 1994 Northridge car quakes caused economic losses estimated at $8 billion and $20 billion ini the state of California alone. These Losses had severe local and regional repercussions but had only tinor cffeets on most U.S. citizens. The 1972 Managua earthquake, on the other hand, caused losses of $2 billion, 40% of Nicaragua's gross national product that year. The 4 3 of recumsuiuctiva pro unemployment, and eventuzlly contributed to the destabilization of the Nicaraguan govern- meat, Moxe receutly, ie Fyogo-Keu Nanbu earthquake devastated the eity of Kobe, Japan; total damages have been estimated in excess of $100 billion. Lifeline failures can also hamper cmergeney response and rescue efforts immediately following damaging earthquakes. Most of the damage in the 1906 San Francisco earth- quake, for example, was caused by a fue that could not be fought properly beeause of bro- ken water mains, Bighty-three yeurs latcr, television allowed the world to watch another fire in San Trancisco following he Loma Picta earthquake. These fires were caused by broken natural gas pipes, and again, firefighting was hampered by broken water mains. The Loma Pricta carthquake also caused the collapse and near collapse of several elevated highways and the collapse of a portion of the San Francisco—Oakland Buy Bridge. 1 portatio Lifelines Laused giidiock davughuut tie aia, Sor still out of service five years afier the earthquake.Sec. 1.3 Seismic Hazards 13 1.3.7 Tsunami and Seiche Rapid vertical seafloor movements caused by fault rupture during earthnnakes can produce long-period sea waves called ssunamis. In the open sea, tsunamis travel great distances athigh speeds but are difficult to detect—they usually have heights of less than 1 m and waveleagthe (the distance between cresis) of several hundred Kilometers, As a tsunami approaches shore, however, the decreasing waier depth causes its speed to decrease and the height of the wave lo increase, in some coastal areas, the shape of the seafloor may amplify the wave, producing a nearly vertical wall of water thut rushes far inland and causes devastating daraage (Fi gure. 1.15), The Great Hoe! Fokaido~Nonhaido tsunaini killed 30,000 people in Japan in 1707, The 1960 Chilean earthquake produced a tsunami that not only killed 300 people in Chile, but also Killed 61 people in Hawaii and, 22 hours laier, 199 people in distant Japan (lida et al 1967). Earthquake-induced waves in enclosed bodies of water are called vaich: Typ Caused by long-period seismic waves ihat match the natural period of oscillation of the Water in a lake or reservoir, seiches may be observed at great distances from the source of ‘an carthquake. ‘The 1964 Good Friday earthquake in Alaska, for exaraple, produced dam. aging waves up to 5 ft high in Takes in Louisiana and Arkansas (Spaeth and Rerkman, 1967) Another type of seiche can be formed when fauiting causes permanent vertical displace- anents within a lake or reservoir, In 1959, vertical fault movement within Hebgon duced a seietang motion that alicrnately overtopped Hebgen Dam and exposed the lake bottom adjacent to the dam in 1959 (Steinbrugge and Cloud. 1967) Figure 1.15 Tecra! damage in Kodiak, Alaska, following the 1964 Good Friday carthemake (courtesy of USCS),1” tnrreduction to Geotechnical Earthquake Engineering Chap. 1 4.4 MITIGATION OF CEISMIC HAZARDS Ultimately, the goal of the earthquake engineer is to mitigate seismic hazards. For new con- struccion, hazard mitigation is ombedded in the process of earthquake-resistant design Details of earthquake-resistant design of structures are beyond the scope of this book, but some aspecis of earthquake loading of structures are described in Chapter 8. Barthquake- resistant design of slopes, dams, embankmenis, and retaining structures is based on topics. presented in Chapters 9 to 11. Mitigation of existing sessmic hazards is also very iuperiant. ‘The important topic of remediation of soil deposits for seismic hazard mitigation is covered in Chapter 12. 1.5 SIGNIFICANT HISTORICAL EARTHQUAKES Earthquakes occur almost continuously around the world. Fortunately, most are so small that they cannot ha felt Only a very small percentage of earthquakes are large enough to cause noticeable damage, and a small petcentage of those are large enough to be considered major earthquakes. Throughout recorded history, some of these major earthquakes can be regarded as being particularly significant, either because of their size and the damage they produced or hecanse nf whar scientists and engineers were able to learn from them. A partial list of significant earthquakes, admittedly biased toward U.S. earthquakes and earthquakes with significant geotechnical earthquake cngincering implications, is given in Table 1-1 TABLE 1-1. Significant orical Earthauakes Date Location Magnitade Deaths Comments TROBC. Chine One of the first relizble written avequnts of a sireng eartiquake: priducec witlespreael danas wet of Xian in Shaanxi Province AD. 19 Italy Biatcon years of frequent carthquakes culminating with the eruption of Mr. Vesuvius, which buried the city of Porngeit 303 India 1804000 Wideypread damage: many killed in collapse of earthen homes 1356 China go (est) 530.000 Occurred in densely populated region near Xin; produced invaaws of Landslides, which killed inhabitants of soft rank cones in hillsides: death estimate of questionable aceuracy 1758 Portugal 86 60,000 Lisbon earthquake: first scientific description of catthquake offects 1783 laaly 50,000 Calabria earthquake; first scientific commission [or earthquakeSec. 1.5 Significant Historical Earthquakes 1 TABLE 1-1 Significant Historical Earthquakes (continued) IRLL-1812 Missouri F5,73.78 Several ‘Thee Large earthquakes in tess than lwo tmonihs in New Maand area: felt all seross contral and eastern United States 500 Cuch earthquake; first well- documented observations of feuding 1357 California 83 1 Fort Tejon earthquake; one of the largest earthquakes kaowa to have been produced by the San Andreas Foul; favlt euptured for 250 mites (400 Kany with up 9 30 €¢ (9 mi offset 1872 California &5 27 Owens Valley earthquake: one of the strongest ever to have occurred in the United States 1886 South Carolina 70 110 Strongest earthquake to strike cast coast oF United States: produced significant quefaction 1819 India 1906 California 74 708 t gicat Cintlaquahe iu sinihe densely populated area in United States; produced up t0 21 117 m) offset in 270-mite (430-km) rupture of San Andreas Feult; anost damage cansed by tire; extent of ground shaking damage comelated to grnlagio postearthquuke investigetion 1908, Italy 18 83,000 Messina and surrounding area devastated; Italian government appointed engineering comsnission Thal recommended structures be designed far equivalent static lasoral Isade 1923 Japan 79 9,000 Kanto carthquake: caused major damage in Tokyo-Yekobama ara, much due Lo fire i Tokye and tsunami in coastal regions: strongly influenced subsequent design in Japan 1925 Calfimnia 63 ta Babasa eaithyaaie, caused liquefaction feiture of Shetfield Dam: led 10 first eaplicie provisions Yor earthquake resistance in U.S. building codes16 TABLE 1-1 Significant Historical Earthquakes (continued) Introduction to Geotechnical Earthquake Engineering Chap. 1 1933 1940 1959 1960 top W964 1967 ro 1975 196 Celifernie Califomia Montana Chile Alaska Jepen Venezuela Californie, Chine China 63 RW a 5 73 78 120 266 oo 1,300 70,000 Considerable building damege; schools partionlarly hurd hit, with ‘many childrca killed and injured; led to yroatcr seismic design requirements in building codes, porticaladly for public school Lulkdings Lorge giownd displacements along Tinperial Fanit nrae BI Centro: frst important aceclerogeam for engineering purposes was tecorded Hebgen Take carthquake: fwulting within reservoir produced large seiche that overtopped cunt Gant Probably cho largest earthquake ever rewuided ‘The Good Friday earihquake: caused sever damage due (0 liquefaction and maay ewrthquake-indueed landslides Widespread liquefaction caused eslensive damage to buildings. bridges, and port Facilities in Nigata; along with Good Friday earthquake ia Alasko, spurred jntense interest in the phenomenon of liquefnotion Caused collapse of relatively new structures in Caracas: Mustrared pifects ot Local soul conditions on gyound motion and damage: ‘San Ternancle carthgualkes produced. several exainples of Liquefaction, including reur collapse of Lower San Kernando Dam; caused collapse of several buildings and highway bridges: many structural lessons learned, particularly regarding need for spirat reinforcement of eonerete columns; many strong motion records obtained Lvzcuation following successful predietion saved thousands at lives jn Hatches. Liaoning Province Thyught to be the most deadly garthquake in history, desinoyed cirg of Tangshan, Hebei Province: not predicted: death estimate of questionable accuracySec. 1.5 TABLE 1-1 Significant Hictorinat 1985 1989 1994 1995, Significant Historica! Earthquakes W Earthquakes ivuriinued) Mexico BE 9,500 Epicenter atf Pacific ts bat most damage oscucred over 220 miles (360 km) away in Mexico City; illustrated effect of local soil conditions on ground motion Lkcation aul Gamage: subsequent studies Zed 10 better understanding of dytamic Properties of fine-grained sails California MW 63 Loma Prieta earthquake; extensive ground motion amplification and Tiquetaction damage in San California 68 61 Northricige earthquake; ocearred on previensly imknown fault bencath heavily populated area; buildings, bridges, Tifetines extensively dameged; produced extraocdinarily strong shaking at several locations Japan Ao £280 Hyugu-Ken Wanbu earthquake: caused tremendous damage to Kobe, Tapan; widesproad fiquefaction in ceclaimed iancls constructed for port of Kobe: landslides and damage to retaining wails and underground subway2.1 INTRODUCTION The study of geotechnical earthquake engineering requires an understanding of the various provesses by which carthquakes occur and their effects on ground motion. The ficld of seé mology (From the Greek seismos for earthqueke and fogos for science) developed trom a internal suucivis aud Lehavius uf is coils, particularly as they vetaie to earthquake phenomena. Although earthquakes are complex phenomena, advances in seismology have produced a youd understanding of (he mechanisins and sales of vocuneuce of earthquakes in most seismically active areas of the world, This chapter provides « brief introduction to the structure of the carth, the 1casous why carthquakes oceur, and the teni- avlogy used to describe them. More complete descriptions of these topies may be found in a aumber of seismulugy ivais, sult an Guiculcay aud Ridkier (1954), Richics (1958), Bullen (1975), Bath (1979), Bullen and Bolt (1985), Gubbins (1990), and Lay and Wallace (1995), A very icadable descri pion of seistiulogy ain cartlyuakes is given by Bolt (1993), 2.2 INTERNAL STRUCTURE OF THE EARTH Une earth 1s roughly spherical, with an equatorial diameter of 12.740 km (7918 miles) and a polar diameter of 12,700 km (7893 miles), the higher equatorial diametcr being caused bySec, 2.2 Internal Structure of the Earth 19 higher equatorial velocities due to the earth's rntation The earth w eighe come 5.4 x 197! tons 4.9 x 10 kg), which indicates an average specific gravity of about 5.5, Since the spe- cific gravily of surficial rocks is known io be an the ardor of 2.7 lo 3, higher specitic grav- ities are implied al greater depths. One of the first important achievements in seismology was the determination of the internal structure of the earth, Large earthquakes produce enough energy to cause measur- able shaking at points all around the world Ac the different types of through the earth, they are refracted and reflected at boundaries between different layers, reaching different points on the earth's surface hy different paths. Studies of these refrac tions and reflections early in this century revealed the layered structure of the earth and pro- vided insight into the characteristics of each layer 2.2.1 Seismic Waves When an earthquake occurs, different types of seismic waves are produced: body waves and surfare waves. Although seismie waves arc discussed in detail in Chapter 5, Ue brief description that follows is necessary to explain some of the concepts of Chapters 2 to 4. Body waves, which can travel through the interior of the carth, are of two lypes: p-maves and s-waves (Figure 2.1), P-waves, also known as primary, compressional, or lon- gitdinal waves, which they pass. They are analogous to sound waves; the motion of an individual particle that a p-wave travels through is parallel to the dircetion uf Wavel. Like sound waves, P-waves can travel through solids and fluids. S-waves, also known as secondary, shear, or transverse waves, cause shearing deformations as they travel through a material. The motion of an individual particle is perpendicular to the direction of s-wave travel. The direc- tion of porticle mower avus iniy ivy vompuuenis, SV (vertical and raicfaviivu uf te maveials uirouga Compressione Undisturbed medium 2 Undisturbed medium Z \ (o) ee Wavelength sk Deformations prvluccd Uy butiy waves: (2) p-Wave; (hj) SV-wave, brom Farthguakes by Bolt. Copyrisht © 1993 by WH. Freeman and Company. Used with petmission.20 Seismology and Earthquakes Chap. 2 plane movement) and SH (horizontal planc movement). The speed at which body waves travel varies with the stiffness of the materials they uravel through. Since geologic materials are stiffest in compression. p-waves travel faster than other seismic waves and are therefore the first to arrive at a particular sitc. Fluids, which have no shearing stiftness, cannot sustain s-waves. Surface waves result from the interaction between body waves and the surface and surficial Tayers of the earth. They travel along the carth’s surface with amplitudes that decroase roughly exponentially with depth (Figure 2.2). Because of the nature of the inter- actions cequired 1» produce them. surface waves are more prominent at distances farther from the source of the carthquake. At distances greater than about twice the thickness of the earth’s crust. surface waves, rather than body waves, will produce peak ground motions, ‘The most important surface waves, for engineering purposes, are Kayleigh waves and Love waves. Rayleigh waves, produced by interaction of p- and SV-waves wiih the earth’s sur- face, involve both vertical and horizontal particle motion. They are similar, in some respects. to the waves produced by a rock thrown inte a pond. Love waves result from the interaction of SH-waves with a sofl surficial layer and have no vertical component of par- ticle motion, Undisturiaed medium ‘ \ ®) Wavelength « a Undisterhed macium Z \ tb} Figure 2.2 Deformations produced by surface wavre: (a) Rayleigh waver and (hb Love, wave, From Larthquikes by Bolt. Copyright ©1993 by W.HL. Freeman and Company, Used with permission 2.2.2 Internal Structure ‘The crust, on which human beings live, is the outermost layer of the earth. The thickness of the crust ranges from about 25 to 40 km (15 to 25 miles) beneath the continents (although it may be zs thick as 60 to 70 km (37 to 44 miles} under some young mountain ranges) to as thin as 5 km (3 miles) or so bencath the oceans—only a very small fraction of the earth’ s diameter (Figure 2.3), The internal structure of the crust is complex. but can be represented by a basaltic layer that is overlain by a granitic layer at coutineatal locations. Since it isSee. 7.2 Internal Structure of the Eerth 21 Upper mantie Lower mantle Outer core (liquid) Innes ure Figure 2.3 intemal siructure of the auth, Posed ie the vecaus ur ilies atmosphere, the crust is cooler than the materials below it (Figure 2.4). In addition to being thinner, the oceanic crust is geoerally more uniform and more dense than tne continental crust, A distinct change in wave propagation velocity marks the boundary between the crust and the underlying mansic. This boundary is known as the Mohorovitié discontinuity, o the Moho, named after the seismologist who discovered it in 1909. Although the specific i tS nor Well understood, its role as a reflector and refractor of seis- imic waves is wel] established. The mantle is about 2850 km. (1770 miles) thick and can be Temperature (°F) Figure 2.4. Fstimated variation of temperature below the surface of the earta, (After Verhoozan, 1960.)22 Seismology and Earthquakes Chap. 2 divided into the upper manite (shallower than ahant 450 km (404 miles)) and the fawer manile. No earthquakes have been recorded in the lower mantle, which exhibits a uniform velocily simetire and appears ta he chemically hamngenenns, except near its lower hound ary. The mantle is cooler near the crust than af greater depths but still has an average tem- perature of about 4000°F. As a result, the mantle materials are in a viscous, semimolien state. They behave as a solid when subjected to rapidly applied stresses, such as those asso- ciated with caicmic wavec, hut enn slowly flow like.a fluid in response ta Long-term streeses The mantle matcria! has a specific gravity of about 4 to 5. The autor core, or Hiquid core, is some 2260 km (1400 miles) thick Asa Tiquid, itean. hol transmit s-waves. As shown in Figure 2.5, the s-wave velocity drops to zero at the core— mantle boundary, or Gusenberg disranriniry: note also the precipitous drop in p-wave velocity. The outer core consists primarily of molicn iron (which helps explain ils high spe- cific gravity af to 19). The iner care, or valid care. isa very dense (specific gravity mp ta about solid nickel-iroa material compressed under emendous pressures ature nf the inner core is estimated to be relatively uniform ar aver JNN°R The temper- Chust i _Mantle i] TE vuter core | Inner core Velocity (km/sec) Density (gem?) ‘ ' Figure 2.5 Variation of p- and s-wave ol! i ! velwoity and doa (ster 9 1000 2000 2000 4000 5000 KM — Elby. 1980) Figure 2.6 shows the influcnce of the earth's structure on the distribution of seismic waves during earthquakes, Since wave propagation velocities generally increase with depth, wave paths are usually refracted back toward the earth's surface. An exception is at the core-mantle boundary, where the outer core velocity is lower than the mantle velocity.Sec. 2.3 Continental Drift and Plate Tectonics 23 Inner Outer core core Figure 2.6 Seismic wave paths illesirating reflection atul refraction of seismic: waves from the source (focus) of the earthquake by the different layers of the earth. Note that P-and s.waves can ceuch the earth's surface between O and 103°, hur the liquid nauure of the outer core allows only p-waves tw coach the surface between 143 and 180°. In the. a "only puts xefiewted trom the inter core ean reach the cazth’s sucfee. (From Suninct, 196%.) 2.3 CONTINENTAL DRIFT AND PLATE TECTONICS Alth of of observations of 93 y between ihe wasilines aud goviugy of eastern South America and the western Africa and the southern part of India and northem part of Australia had intrigued scientists since the seventeumth venluny (Gien 1975; Kearvy aud Vine, 1990), the theory thal has come to be known as continental drift was not proposed until the early twentieth century (Taylor, 1910; Wegener. 1915). Wegener, for example, believed that the earth had only oné large continent called Pangaea 200 million years ago, He believed that Py y duificd (Figure 2.7) iniy the present coniigurarion of the continents. A more detailed view of the current similarity of the African and South American cousts is shown in Vigure 2.8. ‘The theory of continental drift did not reccive much atteniion until about 1960, when the current worldwide network of scismographs was able to define earthquake locations accurately, and to confirm that long-term deformations were concentrated in narrow zones ween rolatively intact blocks of crust. Alsu, vaplutation uf tie Ucean floor did not begin in earnest until after World War IL when new techniques such as deep-water echo sounding, seismic refraction, and piston coring became available. The geology of the ocean floor is, young, representing only about 5% of the warth’s history (Gubbins, 1990), and relatively simple. Its detailed study provided strong supporting evidence uf the historical movement aaea broke into pioees gaea broke inte picees become in the carth sciences in a century. dely avcpied aid ackuowictiged as the greatest advance24 Seismology and Earthquakes Crap. 2 Figure 2.7 Wegener's theory of continental drift: (2) 270 million years ago: (b) L350 million years azo: (2) 1 million years azo. (Atier Verney, 1979.) 2.3.1 Plate Tectonics The original theory of continental drift suggested images of massive continents pushing through the seas and across the occan floor. It was well known, however, that the ocean. floor was too strong to permit such motion, and the theory was originally diseredited by most carth scientists. From this background, however, the modem theory of plate iectonicy began to evolve. The hasic hypothesis of platc teclonics is that the earth’s surface. consists ot a number of large, intact blocks called plates, and that these plates move with respect toSec. 2.3 Continental Drift and Plate Tectonics 26 Figure 2.8 Statistical spherical [it of several continents using the continental slopes rather than the coastlines. (Atter Bullard ct al., 1965, with permission of the Royal Socicty.) each other. The earth’s crust is divided into six continental-sized plates (African, American, Antarctic, Australia-Indian, Eurasian, and Pacific) and about 14 of subcontinental size (e.g., Caribhean, Cocos, Nazca, Philippine, etc.). The major plates are shown in Figure 2.9, Smaller platelets, or microplates, have broken off from the larger plates iv the vicinity of many of the major plate boundarics but are not shown here. The relative deformation between plates occurs only in narrow zones near their boundaries, This deformation of the plates can occur slowly and continuously (aseismic deformation) or can occur spasm- odically in the form of earthquakes (seismic deformation). Since the deformation occurs predominantly at the boundaries between the plates, it would be expected that the locations of earthquakes would be concentrated near plate boundaries, The map of earthquake epicenters shown in Figure 2.10 confirms this expectation, thereby providing strong support for the theory of plate tectonics. The theory of plate tectonics is a kinematic theory (i.2., it explains the geometry of plate movement without addressing the cause of thal movement}. Something must drive the movement, however, and the tremendous mass of the moving plates requires that theSs9pee tap) “TusUTD Ao voorpne sor oweaa ohm, 6r BNET soystwal pur “Nauseiost ‘SoFpu oN JosuoNrap PLU SaAvLEY TBE oD LO Sire eBoy uno} (Wuojsuan) dys-oxLag Kvepunossed — uppozun) ~~ auoz uoNOnpqng —¥—w— ALL OPOIB HY aid ney aS as pom AS ADT BW auenboryy _ oe pues ebuo|-ocpeuiey, aleid se vous sepuger oN 4 youen Soomon NO a « seurLeyy ~ uous uedep yous 26(R86 “YOR ayy) somd useaag souepunog amp iseusoo soysnbyyea Je Azole oid oy Jo suo an99| op JOM Lamdde sv yp “seus weCHUBTS .0 suayuo|e ayy wANATD! SIO) AL “/AANTE IISSAPINPHOM OT AERA, 2728 Seismology and Earthquakes Chap. 2 driving forces be very large, The most widely accepted explanation of the source of plate movement relies on the requicement of thermomechanical equilibrium of the earth’s mate- Hals. The upper portion of the mantle is in contact with the relatively cool crust while the Jower portion is in contact with the hot outer core. Obviously. a temperature gradient must exist within the mantle (see Figure 2.4). The variation of mantle density with temperature produces the unstable situation of denser (cooler) material resting on top of less dense (warmer) material, Eventually, the cooler, denser materia begins to sink under the action of gravity and the warmer, less dense maicrial begins to rise, The sinking material gradually ‘warms and becomes less dense; eventually, it will move laterafly and begin to rise again as subsequently cooled material begirs to sink. This process is the [amiliar one of convection. Convection currents in the semimolien rock of the mantle, illustrated schematically in Figure 2.11. impose shear stresses on the bottom of the plates, thus “dragging” them in var- ious directions across the surface of the earth, Other phenomena, such as ridge push ar slab pull, may also contribute to the movement of plates (Hager, 1978}. Tactanie plate Spreading ridge hhonindary Subduction zone _.. boundary ee km Figure 2.11 Convection currents in mantle, Near the bottom of the crust, horizontal component of convection certents impose shear stresses on bottom of crust, causing, movernent of plates on earth's surface. ‘The movement causes the plates lo move apart in soue places and to converge in others. (After Nason ot al, 1988.)Sec. 2.3 Continental Drift and Plate Tectonics 29 2.3.2 Plate Boundaries Three distinct types of plate boundaries have been identified, and understanding the move- ment associated with cach will aid in the uaderstanding of plate wetonics. The characteristics of the plate boundaries also inifluence the nature of the earthquakes that occur along them. 2.3.2.1 Spreading Ridge Boundaries In certain areas the plates move apart fram cach other (Figure 2.12) at honndarias known as spreading ridges ot spreading rifts. Molten rock from the underlying mantic rises to the surface where it cools and becomes part af the spreading plates. Tn this way, the plates “grow” at the spreading ridge. Spreading rates range from approximately 2 to 18 em/yr (1 t0 7 in/y2); the highest rates are found in the Pacific Ocean ridges and the lowest along the Mid-Atlantic Ridge. It is estimated (Garfunkel, 1975) that new oceanic crust is currently formed at a rate of about 3.1 kin-/yr (1 2. miles*4yn) worldwide ‘The cm ne, fresh basal, is thin in the vicinity of the spreading ridges. Il may be formed by refatively slow upward movement of magma, or it may ha ajected quickly during seismic activity. Underwater photographs have shown formations of pillow Java and have cven recorded Java eruptions in progress. Volcanic activity, much of which occurs beneath the ocean sur- face. is common in thie vicinily of spreading-ridge boundarics. Spreading ridges can pro- trude above the ocean: the island of Ireland, where valeanic (there are 150 active voleanos), is such an example. ‘The mantle matetial cools after #t reaches the surface in the gap between the spreading Plates. As it cools, it becomes magnetized (remnant magnetism) with a polarity that depends of the direction of the earth's magnetic field at thar time. The magnetic field of the earth is not constant on 2 geological time scale; it has fluctuated and reversed at irregular historical inter- vals, thas imposing magnetic anomalies (revarcale of polarity) in the rack that forme at the Ineraasing age Increasing age Figure 2.12 Spreading ridge boundary. Mogma tises to surface and cools in gap. Fouued by spicedliag pkawes. viagnetic anomalies are shown as sitipes of normal and Teversod magnetic polarity, (Affer Easter, 1971.)30 Seismology and Earthquakes Chap. 2 spreading ridge boundaries. Measurement of the magnetic field in a direction perpendicular Wo 4 spreading riage plate boundary reveals a fructuaung pattern of magnetic intensity, as illustrated for the eastern Pacitic Ocean region in Figure 2.13. These magnetic anomalies have allowed large portions ol tae mayor plates to be dated. Comparison of the ages ot various materials allows identification of the geometry and movement of various plates and has proven invaluable in the verification and acceptance of the theory of plate tectonics. Since the size of the earth remains constant, the creation of new plate material at spreading ridges must be balanced by Lhe consumption of plate matetial at other locations. This occurs ut subduction zone boundaries where the relative movement of two plates is toward cuch other. At tho point of contact, one plate plunges, or subducts, beneath the other, 1 Propagating rift Age (Ms) Chron features: Abandoned spreading conter <> Wandering oitset — Posudofsul 50° 4g” 20" 470" 160" 150" 140" 330° 120° Maguctic aucinalica ia he hands of common magnctic polarity. (After Atwater and Sveringhaus, 1989.)Sec, 2.3 Continental Drift and Piate Tectonics 3 Artive unieannes Sound (e.g. Mt. Rainier) i Cascade Columb Detormed sediments Olympic | — Range Plateau ‘ Mountains Pacific Ocean \ SAG] Oceanic crust a. North American wan de Fuca Juan de Fuca plato Plate Ridge Sy, Not fo scale Rising mages Figure 2.14 Cascadia subéuetion zone off the cousts of Washington and Oregon. The Juan de I'vea plate originates al the Juan de Puca spreading cilge and subducts beneath Ue Nocth American plate. Viepma rising from the deeper part of the subduetian zone has formed a geries of voleanos that run roughly parallel to the subduction zone. Oue of these, Mt, St. Helens, crupted explosively in 1980. (After Noson et al. 1988.) as shown in Figure 2.14, Subduction zone plate boundaries caist off the western coasts Mexico and Chile, souti of the Aleutian Island chain of Alaska, and off the castern coast ¢ Japan. The Cascadia subduction zone off the coast of Washington and British Columbia Subduction zone boundaries are often found near the edges of continents. Becaus the oceanic crust ic generally cold and dense. it sinks under its own weight bencath th lighter continental crust. When the rate of plate convergence is high, a trench is formed ; the boundary between plates. Ln fact, subduction zone boundaries arc sometimes calle wench houndaries. farthauakes are. generated i in the sloping Benioff gone al the interfac the gate of conve gee in slow, sed retionary wedge on top of the crustal rock, thus obscuring th ments accunmlate it in an ac trench, The subducting plate warms and becomes less brittle as it sinks. Eventually, | becomes so ductile as t be incapable of producing earthquakes; the greatest recorded earth quake depth of approximately 700 km (435 miles) supports this hypothesis, Portions of th uoting plate melt, producing magma that can rise to the suface tu fun a liue uf wu anos roughly parallel to the subduction zone on the overriding plate. Whea plates carrying continents move toward cach vtlut, continental collisions ca lcad to the formation of mountain ranges. The Himalayas consist of two crustal layers the have formed av the Australia Indian plate has collided with the Curasian plate, Continente collision of the plates carrying Arica and Evrope are currently reducing the sizeof the Mec veatually load to the fortuation of a Vullisivuriy pe uuuuiaia cay (Mckenzic, 1970). 532 Seismology and Earthquakes Chap. 2 2.3.2.3 Transform Fault Boundaries Transjorm fauits occur where plates move past each other without cresting new crust or consuming old crust. They atc usually found offsetting spreading ridges as illustrated in Figure 2.19, Mhese (ransiorm faults are identiticd by ottsets in magnetic anomalies and, where preserved, scarps on the surface of the crust. Magnetic anomaly offsets defining jrac- ture zones may be observed over thousands of klometers; however, it 1s only the segment of the fracture zone between the spreading ridges thet is referred to as the transform fault AS Hustrated tn bigure 2.19, the motion on the portions ot the tracture zone that extend beyond the wansform fault is in the same direction on either side of the fracture zone: hence there is generally no reiative motion, hese inactive portions of the fracture zone can be viewed as fossil faults that are not producing earthquakes. we _ Spreading ridgos | Paina oe = ,— a < < inactive Active Inactive - fracture - transform 4 fracture - > zone faut zone fa) b) Figure 2.15 42) Oblique and (5) plan views of transform Gaull and adjacent inactive fracture ones, The San Andreas fault, for example, has been characterized as a transform fault (Wil- son, 1968) comnucting the Vent Pacifiv ridge uff the uuast uf Meaicy with the Juairde Fuca fidgc off the coast of Washington state. In reality, the geometry of transform faults is usually quite complex with many bends and kinks, and they are olten divided into a number of fiat: segments, Their depth is typically limited but can extend horizontally over very long dis- . Other Lnyurtant Uausivun faults inlude die Moiagua fauli (which scpasaies the North American and Caribbean plates), the Alpine Sault of New Zealand, and the Dead Sea fault system that coumects the Red Sea lo he Bilis Mountains of Turkey (Keatcy aml Vine, (990), Plate tectouics provides a very useful framework for understanding and explaining movements on the earth’s surface and the locations of earthquakes and volcanoes. Plate tee- suuativt: of mew aid Grmsunytivn of uld cLustal maicrialy int terms of the three types of plule movement illustrated in Figure 2.16. ILdovs not, however, explain all observed tectonic svisiticity. For example, i¢ is Kuown that deruplase earthquakes fcarthiquakes that occur within a plate, away from its edges) have occurred on mest conti- nents, Well-known North American examples ate the series of midplate carthquakes that occurred in the vicihity of New Madrid, Missowi, in 1811-1812, and the 1886 Charleston (Suuth Carolina) cartiquake. The 1976 Tangsiian (China) and 1293 Marathawada (indiay earthquakes are more recent examples of damaging intraplate earthquakes.Sec. 2.4 Faults 33 Spreading ridge boundary Subduction zone boundary Spreading { tinge boundary f rrecnire zone | we s/s " iy § /é < /8 § FP atsionm auk boundary is & * ” .- Subducting plaic Upwelling mantle rocks Figure 2.i8, inerreiationsnpy among spreachag nudge, subduction zone, and transtorm fanlt plate boundaries 2.4 FAULTS Wiiile thi iteory of place wetnics generally assigns the relative movement of plates to one of the three preceding types of plate boundaries, examination on a smaller scale reveals that the moyeruen! at a particular location can be quite complicated, In some regions, plate boundaries are distinct and easy 1o identify, while in others they may be spread out with the edges of ihe plutes broken to form smaller platelets or microplates trapped between the larger plates. Locally, the movement between two portions of the crust will occur on new or preexisting oifsers in the geologic structure of the crust known as farutts. Faults may range in length from several meters to hundreds of Kilometers and extend frou the ground surface to depths of several tens of kilometers. ‘Lhe presence may be obvi- ous, as reflected in surficial topography, or they may be very difficult to detect. The pres- ence of 4 fault does not necessarily mean that earthquakes can be expected; movement can occur aseismically, or the fault may be inactive. The lack of observable surficial faulting, on tie other hand, does nor impiy that earthquakes cannot occur: in Tact, aul rupwute does not reach the earth's surface in most earthquakes, The activity of faults is discussed in more dewail in Chapter 4. 2.4.1 Fault Geometry Standard geologic notation is used to describe the orientation of a fault in space, While the surtace of a large fault may be irregular, it can usually be approximated, at least over short. distances, ava plane. The orientation of the fault plane is described by its sirike and dip. The34 Seismology and Earthquakes Chap. 2 strike of a fault is the horizontal line produced by the intersection of the fault plane and a horizontal pla howd in Vigwie 2.17. The azimuth of the strike (¢.g., N6U"L) 1s used to describe the orientation of the fault with respect to duc north. The downward slope of the fault pline is described by the dip angle. which is the angle between the tauil plane and the horizontal plane measured perpendicular to the sizike. A vettical fuult would have a dip angle of 90°, Fault plane Strike vector Horizontal plane Dip vector Figure 2.17 Geomenie notativn for description of fankt plane orientation, rauit Movement The type of movement occurring on a fault in usually reduced ly Components in the drec- tions of the strike and dip. While some movement in both directions is inevitable, movement in one direction or the other will usually predominate. 2.4.2.1 Dip Slip Movement Fault movement that oceurs primarily in the direction of the dip (or perpendicular to the strike) is referred to us dip slia movement. There are different types of dip slip move ments. classified according to the direction of movement and the dip angle of the fault, Nor- mel flats, illustrated in Vigure 2.18. occur when the horizontal enmponent of dip slip movement is extensionai and when the material above the inclined fault (sometimes relerred to as the hanging wail moves downward relative. to the material balou the fault (che foot wall). Normal faulting is generally associated witi tensile stresses in the crust and results in a horizontal lengthening of the crnst_ When the horizontal component of dip slip movement is compressional and the material above the fault moves upward relative to the material below the fault, reverse Saulting is said to have occured Movement on reverse faults. illustrated in Figure 2.19, results ina horizontal shortening of the crust. A special type of reverse Fault is a sfurust fruciz, which aceurs when the fantt plane hi dip angle. Very large movements can de produced by thrust laulting; the European Alps are an excellent example of theust structure. 2.4.2.2 Strike-Slip Movement Faull movement occurring parallel to the suike is called strike-slip movement. Strike- slip faults are usually nearly vertical ancl can produce Jarge movements. Surike-stip faults are further categorized by the relative direction of movement of the materials on either side of the fault. An observer standing near a right lateral strike-s lip fault would observe the ground onSec.2.4 Faults (b) Figure 2.18 (a) Normal faulting {efter Noson et al., 1988); (b) scamp of the nocmat fault thal produced the 1954 Dixie-Pairviow carthquake in Nevada (K. Steiabrugze collection: onumiesy of ERC, Univ. of Calitomiay, 3536 Seismology and Earthquakes Chap. 2 Figure 2.19 Reverse faulting. Because the dip angle is so sinall, dis reverse fault WOILU probably be classttied as a thrust fault. (After Noson ct al., 1988.) the other side of the fault moving to the right, Similarly, un observer adjacent to a left fateral strike-stip fault would observe the matcrial on the other side moving to the lell. The strike- skp fault shown in Figure 2.20a would be characterized as a left lateral strike stip fault. The San Andreas fault in California is an excellent example of right lateral strike-slip faulting: m tha [906 San Fran offset by nearly 6 ma (20 f1) (Figure 2.20b). Obliqne fault movement (Le., movement with both dip slip and strike slip compu- hents) oficn occurs, The 197] San Fernando earthquake ruptured the ground surface over a length of 15 km (9 miles). ‘The maxirnum vertical displacement (produced by reverse fault movement} was 1.461 (4.8 10, and the maximum horizontal displacement (from left lateral ) was 2.13 m (7.0 ft) (Berlin, 1980). f (Borkin, 1980), ke, several readn and fonces north of San strike-slip mover 2.5 ELASTIC REBOUND THEORY ‘The plaws of the cart are in constant motion, and plate tectonics indicates that the majatity of their relitive movement vecurs near their boundaries. The long-term effects of this move- ment cau be ubycrved in the geologic record, weicn reflects deformations that have occurred over very long periods of time. With the advent of modern electronic distance measuicicn| cquiprent, however, Movements ca also be Observed over much shorter time scales. Figure 2.21 shows a set of survey lines established across the Sun Andtens and Calaveras faults by the California Department of Water Resources and Division of Mines and Geology. The shortening of chords 17 and 19, and the lengthening of 20 and 23, indicate that fault mu yeunem is occuring. Chord 21, whieh lies enurcly cast of the Calaveras fault, shows very little change in length. Ap iclative muvernent of the plates occurs, elashe strain energy is stored in the mate- rials near the boundary as shear stresses inerease on the fault planes that separate the plates, When the shear stress reaches che shear sirength of the rock along the fault, the rock fails and the accumulated strain energy is released. ‘The effects of the failure depend on the nalure of the sock afuug tie fault. Tr itis weak and ductile, waat litle strain cnergy Lhat could be storedSec. 2.5 Elastic Rebound Theory 37 Figure 2.20 (1) Left lateral strike-slip faulting (from Neson ot al., 1988); (b) trees Oliset by stnike-shp savitmg through citrus grove in 1940 Imperial Valley catthquake (courtesy of U.S. Geelogical Survey). will be released relatively slowly and the movement will occur aseismivally. if, on the other hand, the rock is strong and brittle, the failure will be rapid. Rupture of the rock will release the stored energy explosively, partly in the form of heut and partly in the form of the stress Waves that are felt as carthquakes. The theory of elastic rebound (Reid, 1911) describes this38 Seismology and Earthquakes Chap. 2 { Calaveras 1 fault were | be —! * Changes in lanth (centimeters) 1 ] Vigure 2.21 (a) Survey lines actess San Andreas and Calaveras faults in California; (h) whaage in chord lengul (extension positive), (:zom Martkqunbes by Bait Copyright ‘©1903 by W.H. Freermin and Company. Used with permission.) process of the successive buildup and release of strain energy in the rock adjacent to faults. Tis often illustrated as shown in Figure 2.22. The nature of the buildup and release of stress is of interest. Faults are not uniform, either geometrically or in terms of material properties—both strong and weak zones can exist over the surface of a fault. The stronger zones, referred lo as asperities by some (Kan- amori and Stewart, 1978) and barriers by others (Aki, 1979), arc particularly important. The asperity model of fault rupture assumes thai the shear stresses prior Lo an earthquake are not uniform across the fault because of stress release in the weaker zones by creep ar foreshacks. Release of the romaining stresses held by the asperities produces the main earthquake that leaves the rupture surface in a state of uniform stress. In the harrier model. the pre-earth- quake stresses on the fault are assumed to be uniform. When the main earthquake occu stresses are released from all. parts of the fault except lor the stranger barriers; aftershocks then occur as the rock adjusts to the new uniform siress field, Since both foreshocks and aftershocks arc commonly observed, it appears thal some strong zones bchave as asperities and others as harriers (Aki, 1984), The engineering significance of asperities and barriers lies in their influence on ground-shaking characteristics close to the fault. A sile located close to one of these strong zones may experience stronger shaking than a site equally closeSec 2.5 Elastic Rebound Theory 38 ‘Deformaron ot a ductile stick | Lo r L Fracture of a brittle stick ro (b) Figure £22 Lilashe rebound theory of eurthgnakes. lu (a) ihe slow deformation of mick inthe vicinity of a plate boundary results in a buildup of strain enemy in the rock in the same way thet strain energy builds up in a ductile stick deformed as shown gy the 2igh., Te the strength of the rock is exeecded, it will rupture, releasing strain cnorgy in the form o€ vibrations, much as the eneegy in the stick would be released when tan shick hresk« After the carthquake, te ruck is displaced trom its original position. ‘The total relative displacement of the platcs is the sunt of the stip displacement st he Fault and passible displa-caucnts Jue v waiying Uistortion of the edges ef the plates near the fault. (After Foster. RJ., General Geology, Sie, © 1988. Adapted by permission of Prentice Hall, Upper Sadde River, New Jersey. o the fault but farther from a strong zone. At largcr distances from the fault the effects of foult nonuniformity devi cusc. Unocumately, methods for locating these strong zones prior to rupture have not yet been developed, Rupture gencrally progiesses across a fauli as a series of dislocations (some multiple event earthquakes can be thought of as a series of small earthquakes that occur in close spa- and iempoal pruauuity). Suvail earthquakes can be modeled as pomt processes since their rupture surfaces usually span only a few kilometers. Large earthquakes, however, can rupturc over distawes of (cus, or even hundreds, of kuometers, and the nature of ground shaking can be influenced by the characteristics of the rupture process. For example, waves emanate from the fault with different strengths in different directions; such directivity effects can produce azimuthal differences in ground motion characieristics (Benioff, 1955; chew, 1961). Cousuuctive interference of waves produced by successive dislo- cations can produce strong pulses of large displacement calfed jfiag (Figure 2.23) at nearby sites toward whict the ruptute is progressing (Bemoll, 1955; Singh, 1985}. 2.5.1 Relationship to Earthquake Recurrence The theory of elastic rebound implies that the occurrence of earthquakes will relieve Stresses along the portion of 4 tault on which rupture occurs, and that subsequent rupture40 Seismology and Earthquakes = Chap. 2 &) Time 0 mA 0 m 1 m 1 n 2 -. 2 nl 3 3 4 n 4 mn 5 5 Pulses Pulses A Resultant in Direction Away Resultant in Direction Trow Divection uf Fault Rusptuie Uf Fault Rupiure Figure 2.28 Schematic ilfustration of directivity eftect on ground motions at sites toward and away from direction of Jault rupture. Overlappiné of pulses can Fead Lo strong Hing pnise at sites Loward which the Fault ruptares. (After Singh, 1985; used by permission af FER. will not oecur on that segment until the stresses have had time to build up again. The chances of an earthquake occurring on # particular fault segment should therefore be related in some way Lo the time that has elupscd since the last carthquake and, perhaps, to the amount of energy that was released. In a probabilistic scnsc, then, individual carthquakes on a particular fault segment should not be considered as random, independent events, This characteristic is important in seismic hazard analysis. Because earthquakes relieve the strain energy that builds up on faults, they should be more likely to occur in areas where litle or no seismic activity has been observed for some time. By plotting fauli movement and historical earthquake activity along a faulk, it is pos- sible to identify gaps in seismic activity at certain locations along faults. According 1 elas- tic rebound theory, cither the movement is occurring aseismically or strain energy is building in the vicinity of these scistnic gaps. In areas where the latter is known to be the case, seismic gaps should represent the most likely locations for future carthquakes. A num- der of seismic gaps have been identified around the world and large earthquakes have sub- sequently been observed on several of them. The 1989 Loma Prieta earthquake occurred on a segment of the San Andreas fault that had previously been identified as a gap, as shown in Figure 2.24. The use of seismic gaps offers promise for improvement in earthquake predic- tion capabilities and seisrnic risk evaluation.Sec. 2.5 Elastic Rebound Theory at Fransisco San San Juan Bautista Parkfield 50 100 150 200 ISI ANGE (my) Figure 2.24 Cross section of the San Andreas fant from north of San Francisco 1 sovth of Parktield: (a) seismicity in the 20 years prior to the 1989 Lona Prieta earthquake is shown with the Loma Prieta gap highlighted; (b) main shock (open circle) snd aftecshooks of the Loma Pricta earthyuake. Note Wie renatining gaps between San Francisco and Portola Valley and south of Parkticld. (After Housner et al, 1990.) 2.5.2 Relationship to Tectonic Environment Elastic rebound also implies that tectonic environments capable of storing different amounts of energy will produce earthquakes of different size. Consider, for example, the tectonic environment in the vivinity of a spreading ridge plate boundary. First, the crust is thin; hence the volume of rock in which strain. energy can build up is small. Second, the hor- izontal component of the relative plate movement is extensional: hence the normal stress on the fault plane, and with it the rupture strength, is low. ‘Lhird, the rock is relatively warm and ductile, so it will not release strain energy suddenly. Taken together. these factors limit the total strain cnergy that can build up and be suddenly reteased at a spreading ridge boundary. These factors explain the observed absence of very large carthquakes al spreading ridge boundaries. By the time the oceani¢ crust has moved from a spreading ridge to a subduction zone. it has cooled and become much thicker and stronger. Relative movement of the plates is toward cach other, so high compressive normal stresses increase the rupture sina eth on the fault plane, Because subduction zone plate houndaries are inclined, the potential rupture area is large. All of these factors support the potential buildup of very large amonnts af strain energy that, when suddenly released, can produce great earthquakes. In fact, the larg- est recorded carthquakes have been produced by subduction zones, At transform faults, the rock is generally cool and brittle, but large compressive Sitesses do not usually develop because the faults arc often nearly vertical and movement ie typically of « strike-slip nature. Because the depth of trenstorm faulting is limited, the total42 Seismology and Earthquakes Chap. 2 amount of strain energy that can he stored is controlled by the length of rupwure. Very large. earthquakes involying rupture lengths of hundreds of kilometers have becn observed on transform faults, but truly “great” earthquakes may not be possible. 2.5.3 Seismic Moment The concept of elastic rebound theory can be used to develop a uscful measure of the size of an earthquake. The seismic moment of an earthquake is given by Mp = wAD QD where [1 is the rupture strength of the material along the fault, A the rupture area, and D the average amount of slip. The seismic moment is named for its units of force times tengt’a, however, it is more a measure of the work donc by the earthquake. As such, the seismic moment correlates well with the energy released during an earthquake. The seismic moment can be estimated from geologic records for historical earthquakes. or obtained from the tong-period components of a seismogram (Bullen and Bolt, 1985). 2.6 OTHER SOURCES OF SEISMIC ACTIVITY The sudden relaace rey hy rupture of the rack at plata honmdaries is the primary cause of seismic activity around the world. There arc, however, other sources of seismic activity that produce smaller earthquakes that may he important in Incalized regions arihquakes have been associated with voleanic activity. Shallow volcanic eath- quakes may result from sudden shitting or movement of magma. In 1975, a magnitude 7.2 earthquake on the big island of Hawaii produced significant damage and was followed shortly hy an eruption of the Kilanea voleana, The 1080 emption af Mt Sr Helens in sonth. em Washington was actually triggered by a small (M, = 5.1), shallow, volcanic earthquake that triggered a massive landslide on the north slope of the voleann. The unloading of the north slope allowed the main eruption to occur approximately 30 sec later. Volcanic erup- tions themselves can release tremendous amounts of energy essentially at the earth's sur- face and may produce significant ground motion. Seiemic was b om ation af chemical explosives or nuclear devices (Bolt, 1975). Many significant developments in seismology during the Cold Wer years stemmed from the need to monitor nuclear weapons testing activities of other countries. Collapse of mine o cavern coofs, or mine bursts, can cause smabi focal earthquakes, us can large landslides. A 1974 Jandslide involving 1.6 x 10° m3¢0.1 » 10° ye?) of material along the Montaro River in Peru produced seismic waves equivalent to those of a maonitade 4 4 Reservoir-induced earthquakes have been the subject of considerable study and some controversy. Locul seismicity increaved significantly after the filling of Lake Mead behind Hoover Dam on the Nevada—Arizona border in 1935. When the Koyna Dam (India) reser- voir was filled, local shallow earthquakes became common in an area previously thought to have been virtually aseisinic. In 1967, five years after filling of the Koyna reservoir hadSéc. 2.7 Geometric Notation a begun, a magnitude 6.5 earthqnake Killed and injucd une cian 2000 more. Local seismicity has been observed to increase seasonally with seasonal increases in reser- voir level, [n 1975. seven years after the filling of Oroville Dana itr an area of low historical! seismicity in northern California, a swarm of earthquakes culminated in a magnitude 5.7 main shock. Aticr construction of the High Dum, a magnitude 5.6 earthquake occurred in Aswan, Egypt where very little significant seismic activity had been observed in the 3000- year history of the area. In these cases, seiomie activ ity appears iv lave been triggered by the presence of the reservoir, While the effect of the weight of the impounded water is likely to be newligible atthe deprhs of the induced seismic activity, au increase in porewater pressure that migtates as a “pulse” away from the reservoir after filling may have been sufficiant to reduce the strength of the rock to the point where rupture could occur, WT LITT ps GEOMETRIC NOTATION To desoribe the location of an earthquake, it is necessary 10 use accepted descriptive termi- nology. Earthquakes result from rupture of the rock along a feult, and even though the rup- ture may involve thousands of syuare kilometers of fault plane surface, it must begin somewhere. The point at which rupture begins and the first seismic waves ori iginate is called the fo © carthquake (tigre 2.25), From the fovus, the rupture spreads across the faull at velocities of 2 to 3 kuisec (1.2 to 1.9 miles/see) (Bolt, 1989). Although fault repture cant exteuel wo the ground surface, the focus is located al some focal depth (or hypocentrai depth) below the ground surface. The point on the ground surface directly above the focus is called the epicenter. The distance on the ground surface between an observer ot site and the epicenter is known as the epicentral distance, and the distance between # of aud dhe fucus is called the focal distance or hypocentral distance. it hypoceater, uf Ground surface 7 _~ Epiventral distance a Site or observer Focus or hypocenier —@—_ ZIN Figure 2.25 Notation for description of earthquake location,4 Seismology and Earthquakes Chap. 2 2.8 LOCATION OF EARTHQUAKES ‘The location of an carthquake is oficn initially specified in terms of the location of its epi- center. Preliminary cpicenteal location is a simple and straightforward matter, bul refinement of the final lovation can be considersbly more complex, Preliminary location is bused on the relative artival times of p- and s-waves at a set of at least three seismographs. Since p-waves travel faster than s-waves, they will arrive first ata given seismograph, The difference in arrival 1 ifference berween the p ands wave Jifference bevween the p ands wave velocities, and on the distance between the scismograph and the focus of the earthquake, according to a 2.2 14, = Wy ea) fa timc betwoon the first p- ai are the p- and s-wave velocities, respectively. In bedrock, p-wave velocities ate generally 3 to 8 km/sec (1.9 to 5 miles/scc} and s-wave velocities ange fom 2 w 5 Au/sox (1.2 te 3.1 miles/see). At any single seismograph it is possible to determine the epicentral distance but not the dizection of the epicenter. This limited knowledge is eapressed graphically by plat ting a circle of radius equal to the epicentral distance. When the epicentral distance from a re moeraph -iave ausivals, aud v, and v, plotted asa i ts lowat 1, tue possible lucation uf tk epi- center 1s Narrowed to the two points of intersection of the circles, Obviously, a third seis- mograph is necossary to identify the most likely location of the epicenter as iMusueted in Figure 2.26. More refined estimates of the epicentral, or hypocentral, Iocation are made using multiple suismographs, a three-dimensional seismic velocity model of the earth, and mumerical optimization techniques. The accuracy of these techniques depends on the hie distribution of the seisuugapls aud uu the avcuacy ofthe seismic velocity model (Dewey, 1979). ps time shows that earthquake occurred at this distance trom station A Doicerter Figure 2.26 Preliminary locatinn of epicenter tom differential wave-arrival-time measurements at scismographs A.B, aad C, Most likety epicentral location is al the intersection of Ute three circles, (A ler Foster, R.., General Geology. Sfe. © (988. Acdoyted by peumissivn of Ptendce Hall, Upper Saddle xrver, 1New Jersey.)Sec. 2.9 Size of Earthquakes 5 2.9 SIZE OF EARTHQUAKES The “size” of an earthquake is obviously a very important parameter, aud it has been described in duttcrent ways, Prior to the development of modern instrumentation, methucls for characterizing the size of carthquakes were based on crude and Gualitative descriptions of the effects of the earthquakes. More recently, modern seismographs haye allowed the development of a number of quantitative measures of earthquake size. Since several of titese Measures are commoniy Used in both seismology and carthquake engineering, the d tinguishing features of each should be understood. 2.9.1 Earthquake Intensity ‘The oldest measure of earthquake size is the earthquake intensity. The intensi ty is a quali- lative description of the effects of the earthquake at a partien i , observed damage and human reactions at that location. Because qualitative descriptions of the etfects of earthquakes are available throughout recorded history, the concept of inteasity can be applica to historical accounts io estimate the locations and sizes of earthquakes that oceurred prior to the development of modern seismic instruments (preinsivumental earth quakes). This application has been very useful in characterizing the rates of recurrerice of earthquakes of different sizes in various locations, a critical step in ev: lihood of seismic hazards. Intensities can also be used to estimate strong ground motion ley- els (Section 3.3.1.1), for comparison of earthquake cffecrs in different gcoaraphie regions, and for earthquake loss estimation, ‘The Rossi-Forct (RF) scale of intensity. describing intensities wit values ranging from L to X, was developed in the [880s and used for many years. It has largely been replaced in English-speaking countries by the moilified Merealli sateneity (MMB + originally developed by the Ttalian seismologist Mercalli and modified in 1931 to beter rep- yesent conditions in California (Richter, 1958). The MMT sealo ig illustrated im Table 2-1. The qualitative nature of the MML scale is apparent from the descriptions of cach intensity level. The Japanese Metcorological Agency (MA) has its own intensity scale, and the Medvedev—Spoonheuer—Karnik (MSK} scale is used in central a nd parison of the RF, MMI, JMA, and MSK scales is shown in Figure 2.27. Earthquake intensities ure usually obrained from interviews of observers after the event, The interviews are often done by mail, but in some seismically active areas, perma- nent observers are organized and trained to produce rational and unemotional accounts of sround shaking. Since human observers and structures are scattered more widely than any seismological observatory could reasonably hope tn seater inete , ebacrva Hons provide information that helps characterize the distri>etion of ground shaking ina region. A plot of reported intensities at different lncations om a map allows contours of cquall intensity, or ixoseisms, to be plotted. Such a map is called an isoseismal map (Figure 2.28), The intensity is generally greatest in the vicinity of the opicenter of the earthquake, and the term epicentral intensity is often used as a crude description of earthquake size. Isoseismal maps show how the intensity dec! PARES, Ar ationnates, with increx distance, los46 Seismology and Earthquakes Chap. 2 Table 2-1 Madified IViercalli Intensity Scale of 1931 1 u ww vw vu 1X, XI Xu Not fott except by a very tow under especially fuvocable eireunrstances, LeU Uy auly a fow pursonts al sent, expecially unt upper (loors uf buildings. delicately suspended objects may swing: Halt quite noticoably indoors, especially un upper floors of buildings, bat many paople Uo not recognize it as an earthquake: standing motor eats may tock slighlly; vibration like passiag ‘of truck: duration estimated During the day telt indoors hy many. outdoors by Few: at night some awakened: dishes, windows, doors disturbed; walls make cracking sound: sensation like heavy truck striking buildmg; standing moter cars rocked noticably Felt by nearly everyone, many awakened: some dishes, windows, els. broken: a few instances of cracked plaster: Unstable objects Uvertummed: cisturbarces oF urces, piles, and oie tall objects sometimes noticed: pendulum clocks may slop, Folt by all, fallen plaster or damaged chimneys; damage slight Fverybaily rans eutinnes: damage nogligihl> in hunldings af gnoel drsign and conetanetion, slight to moderate in well-built ordinary structures, considerable in poorly built or badly designed structures: some chimmeys broken; noticed by persons driving motor cars Damage slight in spectally designed structures, consideratile ia ordinary suosiantial buildings, with partial collapse. ereat in poorly built structares: panel walls thcown out of Frame auuciures: full OF Chimneys, Factory stacks, CannMins, monamenks, Walls; heavy furniture over. Turned: sand sand mud ejected in simall amounts: changes in well water: persons driving mowr core disniebed Damage considerable in specially designed structures, well-designed frame structures thrown ‘ut of phuunby; great in substantial buildings, with partial collapse; buildings shifted off founda. tions; ground cracked conspicucesty; underground pipes broken Some well built wooden structures destroyed; most masomry and freme structures destroyed with foundations; ground badly cracked; rails bent; Jandslides ecmsidereble trom river banks and steep slopes; shifted sand and mud: water splashed over hanks Pew, if any (masonry) structures remam standing; bridges desiroyect: Meda fissteres an grown, underground pipclines completely cut of service; earth slumps and land slips in soft ground; rails bent greatly Damage total; practically all works of contruction are Ganmaped greatly or destroyed; waves seen ou ground surface; Lines of sight aud level are distorted: ubieeis thrown ite the air 2.9.2 Earthquake Magnitude The possibility of obtaining a more objective, quantitative measure of the size of an earth- quake came about with the development of modem instrumentation for measuring ground motion during earthquakes. In the past 60 ycars, the development of scistnic instruments, and our understanding of the quantities they measure, have increased dramatically. Seismic instruments allow an objective, quantitative mcasurement of earthquake size called carth- quake magnirude to be made. Most measurements of earthquake magnitude are instrumen- tal (ic., based on some measured characteristic of ground shaking).Sec. 2.9 Size of Earthquakes 47 T TTT T T T TT MMt | 1/u | yw] ov | va} vir | va | rm] x | xt RE | vo yey am piv) y | yr fal wi fox [ x uma. [ 1 a | my w v vt vir Msk | 1 o| m fw] vi] wf vafver | x | x ie Figure 2.27 Comparison of intensity values for modified Mercalli (MMIN, Rossi Forel (RF), Japancse Meteorological Agency (IMA), and Medvedev -Speotheuer— Kumik (MSK) scales, (After Richter (1058) and Murphy and O'Brien (1977),) New Zealand K ov IX Vn 2s We Le ig ‘3 Vv @ 200 am Kilometers (@) (>) Figure 2.28 Tsescisinal maps from (a) the 1968 Inangalua earthquake in New Zealand (After Eiby, 1980) and (b} the 1989 Loma Prieta carthguake in norihem California (Modified Mercali intensities). (After Houster, 1990.348 Seismology and Earthquakes Chap. 2 2.9.2.1 Richter Local Magnitude in 1935, Charles Riciwer used « Woud-Audersun scisumuncicr to scale for shallow, local (epicentral distances less than about 600 km (375 miles)) earth- quakes in suuthern California (Richter, 1935). Richter defined what is aow known as the local magnitude as the logarithm (base 10) of the maximum trace amplitude (in micro- meters) revorded on a Wood—Anderson seismometer Iocatcd 100 km (62 miles) from the epicenter of the earthquake. The Richter local magnitude (M4, is the best known magnitude scaie, but it is not always the most appropriaic scuie for dexuiipiiuu uf carthguake size. iagiittacle: 2.9.2.2 Surface Wave Magnitude The Richter local magnitude does not distinguish between different Lypes of waves, Other magnitude scales that base the magnitude on the amplitude of 2 particular wave have been developed. At large epicentral distances, body waves have usually been attenuated and. wave magnitude (Gutenberg and Richter, 1936) is a worldwide magnitude scale based on the amplitude of Reylvigh waves with a period of about 20 sce. The surface wave magnitude is obtained from M, = logA + 1.66 logA +2.0 23) wind displacament in micrometers and A is the epicentral dis- tance of the seismometer measured i in degrees (360° corresponding ta the circumference of the earth). Note that the surface wave magnitude is hased an the maximum ground displace- ment amplimude (rather than the maximum trace amplitude of a particular seismograph); therefore, itcan be determined from any type of seismograph. The surface wave magnitude is most commonly used to describe the size of shallow (less than about 70 km (44 miles) focal depth), distant (farther than about 110A Im (67? milesy) moderate ta large carthquakes, where 4 ie the 2.9.2.3 Body Wave Magnitude Pordeep-focus earthquakes, surface waves are often too small to permit reliable eval- uation of the surface wave magnitude. The body wave magnitude (Gutenberg, 1945) ix a worldwide magnitude scale based on the amplitude of the first few cycles of p-waves which are not strongly inflienced by the focal depth (Bolt. 1989}. The bedy wave magnitude can be expressed as my, = logA—logT + 0.01A+ 5.9 Ody where A is the p-wave amplitude in micrometers and T is the period of the p-wave (usually about one sec), Body wave magnitude can also be estimated from the amplitude of one-sec- ond-period, higher-mode Rayleigh waves (Nuttli, 1973); the resulling magnitude, m,,., is commonly used ta describe intraplate earchyuakes. 2.9.2.4 Other Instrumental Magnitude Scales Magnitude scales using different pasts of the instrumental record haye also been pro- posed. The coda of an earthquake motion are the backscattered waves (AKi, 1969) that fol- low passage of the primary (unrefiected) body and surface waves. Aki (1969), showing that certain characteristics of the coda are independent of the travel path, developed a coda magnitude, M,, thal could be obtaincd from those characteristics. The duration magnitude,Sec. 2.9” Size of Earthquakes M.» which is based on the to! raguitucie seaies are empirical ‘ound-shaking characteristics, gnitude (Kanamori, 1977; Hanks and Kanamori, 1979) since it is based on the seismic moment, which is a direct measure of the factors that produce rupture, along the fault. The moment magnitude is given by logs , = S20 a7 2.5) oe TS (3) where M, is the seismic moment (Section 2.5 3) in dyne-cm The refationship detween the various magnitude scak les can be seen in Figure 2.29, Saturation of the instcurnental scales is indicated by their flattening at higher mayuiwude Magnitude sd. seas: M4, (inoment magnitude), Af, magnitude), my, (shout-period body wave sanitade’, amd Mina Fapancre Meteuulughall Agency magnitude). (After Triss, 1083.)50 Seismology and Earthquakes Chap. 2 valucs. As an example of the effects of magnitude saturution, both the 1906 San Francisco 1960 Chile ce quakes produced giuuud shakiuy dat ied tu suuface wave meyuitudes of 8.3, even though the sizes of their rupture surfaces, illustrated by the shaded areas in Fig- ure 2.30, wore vantly different, The gical disparity iu cuergy release was, however, rflected in the moment magnitudes of the earthquakes: 7.9 for San Francisco and 9.5 for Chile San Francisco, 1906 - Me= 8.3 Chile, 1960 Mg= 8.8 My =9.5 Figure 2.30 Comparison of relative areas of Fault rupture (shaded) and magnitudes for 4906 Say Francisce and 1960 Chile eawthouakes, Alihough the shaking of both cattlijoakes procuced surface ware magnilades of 8.3, the amounls of energy released, weie veiy different, as retleeted in thew moment magmnudes. AMET Bore, 184J, Uhe Potion of the ground curing cartheuakes, Scientific American, Vol. 237, No. 6, used with permission } Bolt (19895 sugecsts thai My vt ing lac uscd fur shafluw carthquahes of maguiiude 3 iv 7, M, for magnitudes 5 to 7.5, and M,, for mugnitudes greater than 7.5. 2.9.3 Earthquake Energy The total seismic energy released during an earthquake is often estimated from the relation- ship (Gutenberg and Richter, 1956) logE = 118+ ; 2.6) where £ is expressed in orgs, This relationship was later shown (Kanamori, 1983) to be applicable to moment magnitude as well. It implies that a unit change in magnitude corre- sponds to a LO! or 32-fold increase ia seismic cnergy. A magnitude 5 carthquake therefore would release only about 0.001 times the energy of a magnitude 7 earthquake, thereby illus- Fectives aictgy tliat cauises very large earthquakes. Combining equations (2.5) and (2.6) (using M.,) shows thar the amount of cnergy ccleased during an van iquake iy proportional te the seismic montem, ‘The amount of energy released by earthquakes is often difficult to comprehend: although a single org is small (1 erg = 7.5 « 10 fi-lb), the energy released in an atomic bamb of the size used at Uiroshima (20,000-ton ‘TNT equivalent) would correspond to a tad 6.0 ou 5 of silat] Carthquabes in ville ving dhe buildup of straic magn thywake. On thai basis, the 1966 Chile cardyuahke (if, — 9.5} released ay auich energy as 78,000 such atomic bombs (Figure 2.31).Sec. 2.10° Summary 51 Energy (ergs) 10% Weterorite Impact (10 km diameter, 20 kmisee velocity) 10%. 10 Earth's daily receipt of solar enorgy. 40% Earth's annval heat flow U.S. Annual energy consumption Y zr 10°") Average hurricane (10 day ltetimel 108 Chile 1960 (9.5) ¥/-” | Alaska 1964 (M9.2) 75x10 Works largest git nuciearexpesion gf Average annul 8 New Madrid 1611-12 4° »,global selsmi gy Me i eneiay icase = E Hurricane Mt St. Retens B23 £1 ikinatic energy) Rens gy aes os §, _| Average annual stable gf 1 Magaton *. ws 31072) continental seismicity nuolear explosion = Alagka.1954 (Wf 9.2), — Arnual global seismicity: 2 to Hiroshima atomic bomb 5 tsi rere | Anne atabio 102 co oe ‘ Etectrical energy of \ te megaton 10” typical ununderstonn strstime \ racer Average tornado ea to!8 (kinetic energy} es 50 60 70 80 90 100 107 Mamant magnitude (4) 10°. / Lightning Bot 20 3.0 6.0 7.0 80 90 10.0 11.0 12.0130 Equivatent Moment Magnitude (44) Ruaiire cusizy of various matural shal numan-made phenomena, (After Johnston, 1990. Reprinted by permission of USGS.) 2.10 SUMMARY 1, The earth has a layered structure—the surficial crust is underlain in turn by the man- tle, the out > and the uuer core. The temperature ol cach iayer increases with depth. The temperature gradient ia the mantle causes the semimolten rock to move slowly by convection. 2. The erust is broken into a number of large plates and smaller platelets. Shear stresses on the bolioms of the plates, caused by lateral movement of the convecting mantle, and gravitational forces cause the plates to move with Tespect to each other. 4. Kelative movement of the plates causes stresses to build up on their boundaries, As movement occurs, strain chergy accumulates in the vicinity of the boundaries. This52 5 Pp b Seismology and Earthquakes Chap. 2 energy is eventually dissipated: either smoothly and continuously or ina stick-slip ct thai produces cantiquakes. The size of an earthquake depends on the amount of energy released. ‘There are three diflerent types ot plate boundaries and their characteristics influence the amount of strain energy that can build up near them. As a result, the different types. of boundaries have different earthquake capabilities: subduction zone boundaries can produce the largest earthquakes, foltowed by transform fault boundaries «nd then spreading ndge boundanes. ‘The surfaces on which relative movements occur are called faults. At a particular location, a fault is assumed to be plana with an orientation described by its strike and dip. Fault movement is divided into dip-slip components (normal and reverse fault- ing) and strike-slip components (left lateral and right lateral faulting) ‘The energy-releasing function of earthquakes suggests that 2 period of time fir strain energy accumulation should be expected between large earthquakes at the same loca- tion. Italso suggests that earthquakes should be most likely co occur along portions af a fault for which fittle seismic activity has been observed unless the plale movement has occurred aseismically. Earthquake intensity is a qualitative measure of the effects of an earthquake ata par ticular Incatian, 11 is re e 9 other factors. Isoscisinal maps cam be used to describe the spatial variation of intensity fora given earthquake. Because no instrumental measurements are required, histori cal accounts can be used to estimate intensity values for preinstrumental earthquakes, EHarthqueke magnitude is a quantilalive meanure of the size of an carthquake. Must magmtude scales are based on measured ground motica characteristics. The local magnitude trace aupliiude uf a particudan scisuuvuucier, die surface wave Magnitude on the amplitude of Rayleigh waves, and the body wave magnitude on the ampkitude of p-waves. Burause these aniplitudes tend ti teach limiting values, these magnitude scales may not accurately reflect the size of very large carthquakes, Tho moment magnitude, which is not obtaincd from ground motion characteristics, is able lo describe the size of any carthquake. Earthquake magnitude seaies are logarthnne, hcnec « uml change in magnitude cor- responds to a 10-fold change in the magnitude parameter (ground motion character- istic or seismic moment). the energy released by an earthquake 1s related to magnitude in such a way that a unit change in magnitude corresponds to a 32-fold change in energy. HOMEWORK PROBLEMS Convection caused by thermal gradients in the upper mantle is thought to bea primary cause of contincatal drift. Estimate the average thermal gradient in the upper mantle. The coeflicient of thermal expen sion of the upper mamile is about 2.5 x 10 °K. Estimate the ratio of the density at the top af the ripper mantle ta that ai the haifom avn the upper mantle,Chap, 2 Homework Problems 53 2.3 24 25 2.6 27 Using the data from Figure 2.21, determine whether the San Andreas and Calaveras fanits are undergoing right latera) or left lateral strike slip faulting. ‘Using the data from Figure 2.21, estimate the average rate of relative movementalong the San Andreas and Calaveras faults during the period from 1959 to 1970, Assuming p- and s-waves traveled through the crust at 6 km/sec and 3 kn/see, respectively, estimate the epicen:ral location (latitude and longitude) of the hypothetical earthquake whose chanacteristics are given below: eee Seismograph Latimde Longitude p-wave arrival time wave arrival time 3P2230" —2iPST30" 06:1 1:18.93 06:11:26.50 374500" 122°20'00" 06:1 1:14.84 6:11 18.71 305233" sa neaggge 6.4 117.26 MLL L3.393) Using a map of California, cietermine which fault the hy gotltctical eatthujuake of Problem 2.5 would most likely have occurred on? An earthquake causes sn average of 2.5 m strike olip displacement over an X0 kin long, 23 km deep portion of a transform fault. Assuining that the rock along the fault liad an average rupture strength of 175 KPa, estimate the seismic moment amd“ inn Strong Ground Motion 3.1 INTRODUCTION The earth is far from quiet—it vibrates almost continuously at periods ranging from milliseconds to days and amplitudes ranging from nanometers to meiers. The groal majority of these vibra- tions are so weak that they cannot be felt or even, deiected wil ithout specialized measurement equipment. Sach micros Earthquake engineers are interested primarily in strong ground motion (.c., motion of. salfickne strength to affect people and their environment). Evaluation of the effects of earthquakes al a particular site requires objective, quantitative ways of describing strong ground motion, ‘The ground motions produced by earthquakes can be quite complicated. At a given point, they can be completely described by thice components of translation and three com- ponents of rotatian orthogonal components of translational motion are most commonly measured. Typical ground marinn records, such as the acceleration time histories shown ia Figure 3.1, contain a tremendous amount of information. To express all of this information precisely (i.e., to reproduce each of the three time histories exactly), every twist and tura in each plot must be described. The motions shown in Figure 3.1, for example, were determined from 2000 accalaratinn vanes ponents are usually neglected: three ponents ore usually neglected; three 8 0 sec. This large amount of informa f 0.02 ec. This large informa tion makes precise description of a ground motion rather cumbersome.Sec. 3.1 Introduction 55 Gilroy No.2 ) 2 o Acceleration ( ° Acceleration (g) ° 2 am East West Wepre 05 05 2 £ L oe wa o 1% 20 a0 0 Time (ses) Time (sec) 06 8 - : 58 7 7 = North - South = Nosth - South 5 5 g 0 4+ Bo | Alasiennnn 3 y ; = bid nm ay 05 2 i Jt “os ; 0 10 20 30 40 0 10 20 30 Time (seo) Time (sec) 2 & Acceleration (3} ° = g. B 2 a Acceleration (g) ° 2 a } & Vertical ° 10 20 30 40 0 10 20 30 40 Time (sect Figure 3.1 Acceleration time histories recorded af two sites in Gilroy, California during the 1989 Loma Prieta enrthiyrshe, The Gilrny No. 1 instrumeut was locale an 4m oulerop o7 Franeiscan sandstone, while ihe Gilroy No. 7 instrament was underlain by 165 a1 (540 8) of stiff, alluvial soils. The Gilray No. 1 frock) and Gilny No, 2 (soil) Sites Were located at eprcentral distances of 21,8 kin 113.5 miles) and 22.8 km (14.2 miles), respectively Fortunately, it is not necessary to reproduce cach time history exactly to describe the ground motion adequately for engineering pnmpnses. It is necessary, however, to be able to describe the characteristics of the ground motion shat are of engineering significanes and to identify a aumber of ground mation parameters that reflect those characteristics, For enei- neering purposes. three characteristics of earthquake motion ate of primary significance: (1) the anzplituede, (2) frequency content, and (3) duration af the A nambor of ground motion parameters have been proposed, cach of which provides information about one or more of these characteristics. In practice, it is usnally necessary to use more than one of these parameters to characterize a particular ground motion adequately. This chapter describes the instruments and techniques used to measure strong ground motion, and the procedures by which measured motions are corrected. It then presents a vari- ety of parameters that can be used to characterize. the. amplitnde, tion of strong ground motions. Relationships that can be used to predict these Parameters are also presented. The chepter concludes with a brief description of the spatial variability of56 Strong Ground Motion — Chap. 3 ground motions. Before proceeding further, the reader should review the topics discussed in Appendices A and B—familiarity with the concepts presented in those appendices is assumed in this chapter and the remainder of the book. UD UNBNBIBRIBN Lecaboremener The ilentification and evaluation of are “requires aocese to meacursments of strong ground motions in actual cnguhen Accurate, quantitative measurement of sirong grim mation is eritical for both seismelogicul and earthquake engineering applications. As stated by the National Research Council Committee on Earthquake Engineering Research (Housner, 1982): “The recording of strong ground motion provides the basic data for earthquake engineering. Without a knowledge of the around haking generated by earthquakes, it is not pos- ite 101 assess bez opricte methode of celomie deciga” 3.2.1 Seismographs Although written descriptions of earthquakes date back as far as 780 B.c., the first accurate Theasurements of destructive ground motions were inade during the 1933 Long Beach, Cale ifornia earthquake (Hudson, 1984). Measurement of ground motion has advanced consid- Wiost rapidly iu ihe pasi 20 yoata ur 90. Varions instruments are available for grouad motion measurement. Seismographs are used to measure relatively weak wound motion, tie coords they produce are called seis mograms. Sitong ground motions are usually measured by acceleragranhs and expressed in the form of aecelereg rams. The simplest type of scismugraph can be iustrated by a mass spring damper single-degree-of freedom (SDOF) system, as shown in Figure 3.2, A rotet- jag drum i neciud ty the scismvgiaph housing with a siytuy atiavted io a vass. The mass is connected to the housing by a spring and dashpot arranged in parallel, and the hous- ing is connected to the ground, Since the spring aud Cashput are nut rigid, dhe motion of de mass will not be identical to the motion of the ground during an earthquake. The relative movement of the mass and the ground will be indicated by the race made by the stylus on Housing—, Figure 3.2. Simple mass-spring dashpot type of seisrnograph. The huusing is firey comectod to the ground. When the ground shakes, the stylus sneaks & trace ou the rorating drum that sinows ihe relative displacement between the mass and the ground, Most modern instraments are mote coiaalicated than the one shown hese.Sec. 3.2 Strang-Motion Measurement 57 Aty the rotating dra evaph siativn imay bave three seismographs oriented to record motion in the vertical and two Perpendicular horizontal directions. Sciemographs can be designed to ueasure various ground mottoa characteristics, To understand how this can be done itis lecessary Lo consider the dynamic response of a sim- ple seismograph such as the one shown in Figure 3,2. This seismograph is a SDOF oscillator whose response to shaking is given by the equation of motion (Appendix B) iii + cii-+ ka = —mnil, en) where w is the seismograph race displacement (the relative displacement betweeu thie seise mograph and the ground) and 1, is the ground displacement, If the ground displacement is simple harmonic at a circular frequency ,, the dis- piacement response ratio (the ratio of trace displacement amplitude to ground displacement amplitude) will be {el SF G2) (1 - By + (2epy where B (= ap/aig) is the tuning ratio, @, (= /k7m) is the undanped natural circular fre- quency, and & (= c/2.fiem) is the damping ratio. Figure 3.32 shows how the displacement response éatiu varies with Frequency and damping. For ground motion frequencies well above the natural frequency of the scismograph (ie., large values of | B), the trace amplitude is equal to the geound motion amplitude, Lhe lowest frequency for which this equality holds (within a given range af acenracy) depends on the damping ratio. Because the frequency response is flat and phase angles are preserved at damping ratios of 60%, SDOF displacement seismo. graphs are usually designed with damping ratios in that range (Richart et al, 1970), Shuilarly. the acceleration response ratio (the ratio of trave displacement araplitude to ground acceleration amplitude) is given by Jl 1 Mel bap? + ep) The variation of acceleration response ratio with frequency and damping is chown in Figure 5.3p. ‘he trace amplitude is Proportional to the ground acceleration amplitude for frequencies well below the natural frequency of the seismograph (i.e., low vaines of B.A Seismograph with 60% damping will accurately measure accelerations at fiequencies up to about 55% of its natural frequency. Most seismographs of this type have natural frequencies of about 25 Hz with demping ratios near 60%, with desirable flat response (constant accel- eration response ratio) at frequencies up lo about 13 Hz. ihe preceding paragraphs show how the same physical system can act as both a dis- placement seismograph and an accelerograph. It measures displacements at frequencies well abave and accelerations at frequencies well below its natal frequency. The Wood Anderson seismograph, used by Richler to develop the first carthanake magnimde scule, used a smali mass suspended eccentricaily on athin tungsten torsion wire. A mirror attached to the wire allowed optical recording with a ground motion magni Damping was provided electromagnetically at 80% of critical; the damped natural period was about 0.8 sec. . 3)58 Strong Ground Motion — Chap. 3 a2 04 0500 08 1 tS 2 8 # 8 678 910 BR } Figore 3.4 _ i) Displacement response ratio and (bi acceleration response ratio (t= 1 vadisce) for SDOF system sudjected te simple harmonic base motion. Tn mest modem seismagraphs, an cloctronic transducer ofien referred to as a seismo meter seases the motion and produces an analog (continuous) clectrical signal that is recorded for subsequent processing. Most aeceleragraphs currently in use are aecelerome- gers, electzonic transducers that produce an output yollage proportional to acceleration, A number of different typas af arrclernmetere are available. Sern lor force balance) accel erometers usc a suspended mass to which a displacement transducer is attached, When the housing is accelerated. the signal produced by the relative displacement between theSec. 3.2 Strong-Motion Measurement Ey honsing and mass is used fo generate a storing force thai pusiies tin mass back toward us equilibrium position. The restoring force ig proportional to Lhe acceleration and can be mea- snred electronicelly. Serve uecelerometers can provide very goud accuracy over the range of frequencies of greatest interest in earthquake engineering. Piezoelectric accelerometers use amass attached to a piezoelectric material (usually quarlz, tourinaline, or a fetroclectric ceramic) to sense accelerations. The piezoelectric materia! acts as the spring in che diagram of Fignre 3 9; damping is a Higible. Wheu avceteaica, the inertial force of the mass sirains the pievoelectric material, which develops an electrical charge on its surfaces. The resulting voltage is (if the dielectric constant docs iol val y with charge) proportional to acceleration. Because piezoelectric materials are quite stiff, their natural frequencies are very high, so they are particularly useful for high-frequency measurements, Their response atlow frequencies, however, can be strongly influenced by signal conditioning system char- acteristics, Trigxial se Wilogonal components of acceleration are measured witb a common time base, are commonly used. From the three components, the acceleration in any direction can be computed. Soine seisimouraphs use velocity ians- ducers, or geophones, in addition or as an alternative 10 accelerometers, Scismographs, accelerographs, and ancillary cyuipmient are protected by an instru- meat shelter (Figure 3.4). An important component of a seismograph or accelerograph is an acenrate cl i re than vue CULUpOuENt of motion is measured or wacn the ground motion at one location is being compared with that at another. Most moder instruments maintain time accuracy by synchiouiciug on a daily basis with radio time sig- nals transmitted by a standard time service or by recording such signals along with the ground-motion data. Universal Coordinated Time (the scientific equivalent of Greenwich Mean Time) is used as a common worldwide time basis. A selemaseope (ndcon, 1958} is a vclaii rely inexpensive ground motion instrument, Scismoscopes are conical pendulums (Figure 3,5a) in which a metal stylus attached to a sus- pended mass inseribes a record of ground sulion ow a suioked glass piate, producing a two- dimensional record of the type showa in Figure 3.3b. Scott (1972) found that small oscil- lations of the trace were roleted lo the instrument rather tran the earthquake and that they could be used to provide a time scale to the seismoscope truce. The time scale allows accel- eragrams to he larly whe larly when puted from the scisuuacupe imace. 3.2.2 Data Acquisition and Digitization Early ground motion instruments transformed the motion of the ground to the motion of a physical mechenism. The mectianism, perhaps in the form of a pen or stylus or reflective mirror, caused the motion to be recorded in analog form on paper or photographic film attached to a uutatiug dru. Later-gencration struments recorded motions electronically in analog form on magnetic tape. Rather than record continuously, instruments of these types lay dou ant until uiggcred by the exceedance of a smail threshold accelcration at the beginning of the earthquake motion. As a result, any vibrations that may have preceded trig- gering were not recorded, thereby introducing a baseline error into the acceleration record To use the recorded ground motions for engincering computations, the analog ground records auoi be digitized. Originanly, digitization w: performed manually with paper, pencil, and an engittcoring scale. Semiautomatic digitizers. with which a user moved alens with crosshairs aciuss an accelerograr mounted ona digitizing table, were commonly used in the lare60 Figure 34 (2) Modem digital strong motion iustrument (solar-powered, [6-bit dun, 250 saunpiessec. GPS tiuing. aud Ceiiuiar meden) amused in Gast-in-piave reinforced concrete vault, and (>) completed installation with insulated cover and salar panel (courtesy of Terra Technology Cerp, Reelmonil, Weshingien)Sec. 3.2 Strong-Motion Measurement 61 1 o. oO a“ - Smeked A glass plate | — - Suspension wire Damping magnets @ (b) Figure 35 (a) Typical Wilevot type seismoscope iu which a fixed stylus scribes record of relative motion on a stioked glass plate mounted on the suspended seismoseope: (b) typical seismoscope recon, (Atter Newamark and Rosenblueth. 1971.) 1970s. By pressing a foot-operated switch, the coordinates of the crosshairs were recorded. These fortis ul ig lavoived exacting and tiring work; operator accuracy and fatigue were important considerations (Tudson, 1979), Fully automatic computer-based digitization, typi- vally at sampling, rates uf 200 or tuore samples per second, is now commonplace. In recent years, digital seismographs have become much more commonly used. Although they usc analog transducers, digital instruments convert the analog signal to digital form in the field. They record ground motions continuously at rates of 200 to 1000 sam- plesisce witht 12- tv i6-bit resulution, saying the recorded data only if a triggering acceleration is exceeded. Their on-board memories can typically save 4 to 6 Mb worth of dala, from before an cartl quake Legins until after itends, thereby preserving the initial portion of the record that is lost with triggered analog systems. Because digital systems are more complex, more expen- sive, and more difficult to maintain in the field, they aave not yet replaced analog systems. 3.2.3 Strona-Motion Pracessing ‘The raw data obtained from a strong-motion instrament may include errors from several pos- sible sources, each 0! which must be carofully evaluated and comected to produce an accurate record of the actual ground motion. Raw data often include background noise from different sources. Microseisms trom ovean waves can be detected by seasilive instruments. Other noise may be caused by traffic, construction activity, wind (transmitted to the ground by vibration of Uces, Ounldigs, etc.), and even atmospheric pressure changes. Obviously, this tange of sources can produce nonscismic noise at both low and high frequencies. To isolate the motion actually produced hy the carthquake, background noise must he removed or at least suppressed.62 Strong Ground Motion Chap. 3 All accelerographs have their own dynamic response characteristics, or instrument response, thal can influence the motions taey measure. Consequently, instrument response must be corrected for in sirong-motion processing. Tnstriment response corrections are nen ally performed by modeling the instrument itself as a SDOF system and using the SDOF model to decouple the response of the instrument from the actual ground motion. Kor mast modern accelerographs with flat frequency response up to about 12 to 13 Hz, the instrument correction is oaly important for frequencies shove rhe nsval range of engineering interact However, some accelerogtaphs are located in buildings (usually on the ground floor or inthe basement) cr near the aburments of dams of bridges. The motions they record can he affected at frequencies of interest by the response of the structure in or near which they are located, ven the motions recorded by strong motion instruments located in the free Jiedd (away frown the influence of large structures) may be influenced by the response of their instrument shel- ter (Bycroft, 1978: Crouse et al, 1984), atthangh these effects are nenally impnrrant only ot selatively high frequencies (Crouse and Hushmand, 1989) for typical instrument shelters, Another correction is required tn recnce the effects af errars in grannd mation mea. surement, such as those associated with the triggering of analog seismographs. If a seisrno- staph does not stari until some triggering level of motion is reached, the entire accelerogram, is in error by the level of motion at the Lime of triggering. Integration of an uncomected acceleration time history. for example, will produce: a linear error in velocity and # error in displacement. An acceleration error as small as 0.001 at the beginning of a 30-sec- long accelerogram wonld erroneously predi¢t # permanent displacement of 44.1 em at the end of the motion. Correction of such crrors, termed Aasefine correction, was criginally accomplished by subtracting a best-fit parabola from the accelerogram before imegrating to velocity and displacement but is now performed using high-pass filters and modern data processing techniques (Toyner and Ranea, 1988) The motinns shawn in Hignra 31, f example, were bandpass filtered to remove frequencies below 0.08 Hz and frequencies above 23 Hz. Campnter sftware for processing strang-gromndemotinn reenrds (Converse, 1992) is available from the U.S. Geological Survey (USGS). dratie 3.2.4 Strong-Motion Instrument Arrays Points oa the ground sueface. The spatial variation of ground motion, whether on worldwide, regional, or local scales, is important in both scismofogy and carthquake engineering. Arrays and networks of strong motion instruments have proven useful in determining the spatial. variation of strong ground motion, 3.2.4.1 Worldwide and Regional Arrays Understanding of earthquake and tectonic processes improved dramatically with the establishment of the Worldvide Standard Seismograph Network (WWSSN) in 1961. The WWSSN was originally developed, in large part, to monitor compliance with nuclear weap- ons testing bans. Before that ime. worldwide seismicity data were obtained from a wide. variety of very different lypes of instruments operated by many different organizations. Dif- ferences in instraments and operating procedures made comparison of results difficult WWSSN stations use standardized instruments; each station has at least two three-compo- nent analog seismographs to monitor both short- and long-period motions. The capabilities of the WWSSN instruments, however. are limited by modern standards (Aki and Richards,Sec. 3.2. Strong-Motion Measurement 63 1080), and they are being ri Uigiial insii wiuenis suci as tose of the Global Diguat Seismometer Network (GDSN) and the Global Seismographic Network (GSN). The Incor- porated Research Institutions for Scisuulogy (IRIS), « consortium of U.S, and foreign research instilutions, oversees operation of the GSN and a set of portable instruments that can be deployed to monitor aftershocks following laige earthquakes Regional arrays of seismographs are now operating ini most seismically active coun- tries, In the United States, for jc. dis USGS uperaicy regional arrays in different parts of the country. In the California Strong Motion Instrumentation Program (CSMIB}, the Cal- ifomia Division of Mines and Geolugy operates an extensive array of free-field seismo- graphs (Figure 3.6) as well as scismographs in buildings and bridges. New wadRI RESON % aie Fugue 3.6 Locations of strong motion instruments cperared by de U.S. Geological Survery in cooperation with othr agcacies as of April, 1990, Boxes in nenhem and ‘southern Califomia indicate urcas with high iuswuinein Ceusity, CAlicr Joyner and Boore, Geowchnical News, March, 1991, p. 24. Used by permission of BiTech Publishers, Led..) 3.2.4.2 Local and Dense Arrays While widely on: scFal for studying cailayuake mechanisms and the spatial distribution of many important earthquake parameters, geotech- nical earthquake enginecring often requives spatial distribution information un a suralle: az scale and below the ground surface. In recent years a number of local and dense arrays, some with downhole instrumentetion, have been installed at various locations around the world. Japan has been very active in the installation of local strong-motion instrument arrays. The threedimensi eleromet y at Chiba (Katayaina aud Saiz, 1982), fr example, includes 44 three-componem accelerometers, 15 of which are at the ground sur- face and the remainder at depths of up to 40m (130 1L). In Taiwan, the SMART-1 dense accelerometer atray near Lotung (Figure 3,7) consists of a central acceleromeier surrounded nal and worFigure 2.7 Oviginal eonfi SMART 1 anay in Loweg, Taiwan. The Fl Centra DiFferentizl Array 64 Strong Ground Motion Chap, 3 tiseaow - 7 33°00N Brawley o Ary & kn toi ® wes Noe Taipei ~ - ‘ a Ao mm Imperial = tke ElCentro'8§ ® / (200m Nota Noe | { 1 0 & 7 w NS | 3245 \ (omen \ == Surtace accelerometer cation of Figure 3.8 The El Centro is located’ near Station §, by three rings of [2 accelerometers each at radii of 200 m (630 f). 1 kin (0.6 mi), and 2 kin (1.2 oti), A few years after the SMART-1 array was installed, | and } -scale models of nuclear containment structures were constructed in tts midst (Vang, 1987). The structures were instrumented lo record their response during eaithquakes, und additional surface and downhole (1 depths of 47 m (154 10) ground motion instruments were installed adjacent to the 4 -scale model and in the free-tield. In the United States, one of the most important Jocal arrays has been the El Centro Array, a 45-kin-long (28 mi) array of 13 stations that crosses the Imperial and Brawley faults in southern Califoraia (Figure 3.8). It also contains the Fl Centro Differential Array, a dense array consisting of six three-component accelerometers along a 305-m (1000 fty ine. Shortly after installation, the arrays recorded the £979 Imperial Valley earthquake (M, = 6.9), which occurred only 5.6 km away and produced very useful information on near-field ground motions. Near Anza, California, an array of ten diree-component stations along a 30-km (19 nti) stretch of the San Jacinto fault was installed to study various earth- quake characteristics (Bergcr ct al., 1984). Data are telemetered by digital VHF radio to a nearby mountain peak station and then on to another station in La Jolla, California. These ate but 2 few of the many strong-motion arrays that have been installed in seis- ically active countrics around che world. The rapid proliferation of local, regional, and worldwide seismograph arrays in recent years has come hand in hand with technological advances in data acquisition, storage, and communication. The ability to acquire and store large quantities of digital seismic data at high speeds, and to retrieve the data from remote locations by telemetry, has end will continue to make such data more plentiful. 3.2.5 Strong-Motion Records Strong-motion records can now be casily obtained from a number of sources. ‘The U.S. Geclog- ical Survey. for example. published a compact disk (Seekins et al. 1992) that contained uncar- rected strong motion records from North American earthquakes between 1933 and 1986; more than 4000 records were incluiled. A number of strong-motion datahases can be accessédl over the, Internet, with individual records downloaded by anonymous ftp (file transfer protocol)Sec. 3.3 Ground Motion Parameters 6s Procedures. The Gilray records showm in Figure 3 mm the database maintaiued by the Lamont-Doherty Earth Observatory at Columbia University in conjunction with the National Center for Farthquake Engineering Rescarch at SUNY Buffalo. An cxample of the information provided with such records is shown in Figure 3.9. A useful World Wide Web site with links to many sources of ground motion and earthquake information is maintaincd at the University of Washington (http://mww. geophys. washington. edufseismosurfing. fitml). A variety af gentechnical carthemake angie a b site maintaiued at the ig University of Southern Califomia (iuip.// recg0T use. eduleqdata/home.himt), NCEER ASCII STRONG-HOTION DATA FORMAT EVENT PARAMETERS: parm: year=1989 month=10 day=18 TIME: hour|minute (24hr)=0004 second- 2.200 time code=UIC 27.93700 immyitudes -141.sus00 depth (cm) NAME: SANTA CRUZ MNS (LOMA PRIETA) RARTHQUAKE SITE PARAMETERS: = 18.0 LOCATION: latitudes 36.97300 longitude= -121,57200 elevation (m) o.0 SITE ID: 47379 - GILROY #1 - GAVILAN COLLEGE. WATER TANK CODE: 47379 RECORD/TRACE PARAMETERS: START DATE: yeer=1989 month=10 day=18 START TIME: hour|minute (24hr)=0004 second-23.900 time codesure EPICENTRAL DISTANCE: @ietance (kn)= 21.8 azimth (deg)=w2s9.1 SPECS: sampling rate (sec}=0.020 number of points= 2000 =CM/SEC/SEC data format=10F8.2 INTERPOLATED COMPONENT OF MOTTON: azinneh=no (oRrZoNTAL) COMMENT: POLARITY OF MOTION RELATIVE TO GROUND DATA: 715-68 -3.95 5.46 -3.90 6.749.490 2.38 0 9.810 -0.193,5 “3.67 -5.78 7.40 -4.94 -20.14 3,44 011.68 15.90 2.78 5.54 ~8.23 -17,18 -3.88 0 -4,98 -8.69 15.80 3.82 0-20.33 0-15.56 0-5. 71 70.330 5.18 -9.95 -10.07 -10.64 -3.87 29.66 -2.16 -13.34 -3.92 “22-97 G.31 0.25 1443 12.96 23,51 16.04 «15.48 0-12.18 9 -5.45 19.23 -37.29 -15.52 41.53 -~23.24 -14.09 37.55 -26.03 -43.07 12.52 26-66 -G.09 17.72 -20.30 39.13 15.81 -25.54 15.76 -46.07 -25.70 74.64 -31.37 -89.33 24.91 -4.54 9.83 -67.54 1.26 -32,89 42.30 Figure 3.9 Event, size, and recon infirmation pevecding the digitized acceleration data for the Gilroy Na. 1 troek) strong, motion record, WIE TERS Ground mation parsmeters are essential for describing the important characteristics of strong ground motion in compact, quantitative form. Many parameters have been proposed.66 Strong Ground Motion — Chap. 3 to characterize the amplitude, {frequency content, and duration of strong ground motions: some describe only one of these characteristics, while others may reflect two or three, Because of the complexity of carthquake ground motions, identification of a single param- eter that accurately describes all important ground motion characteristics is zegarded as impossible Jennings, 1985; Joyner and Boore, 1988) 3.3.1 Amplitude Parameters ‘The most common way of describing a ground motion is with a time history. The motion parameter may be acceleration, velocity. or displacement, or all rhrae may he displayed as shown in Figure 3.10. Typically, only anc of these quantities is measured directly with the others computed from it by integration and/or differentiation. Note the different predomi- nant Frequencies in the acceleration, velocity, and displacement time histories. The acceler- ation time history shows a significant proportion af relatively high freqmancies: Integration produces a smoothing or filtering effect |in the frequency domain, #(@) = a(c) 4a and Gilroy No.1 East - West Gilroy No.2 East - West BS O5r i BS O08 5 i 5 ooh 5 ll g “ape 5 2 at treme o a a [4 g tins go. ° - “05 0 10 20 Kt} 40 o 10 20 30 40 Time {seo} Time (gee) J 40, oa 1 = 20}. il E20} |i 2 4 lip . So het llalpatsoee teense pa ee “|S Op UN elteeee 8 20}. 8 20). | 3 3 s 3 40 -40 G 10 20 0 Time {sea} =z — 5 5 e 2 3 & & a S S 3 a Time {sec} Time (see) Figure 3.18 Acceleration, velocity, and dinglavemeny time histuries for the CW components of the Gilroy No. 1 (rock} and Gilroy No. 2 (soil) strong motion records, The velocities and displacements wore obtained Inv inteyrating the acceleration records shown in Figure 3.1 using the trapezoidal rule, Note that the Gilroy No. 1 (rock) site experienced higher accelerations. but the Gilroy No. 2 (soil) site experienced higher velocities and ehsplacerrentsSec. 3.3. Ground Morinn Parameters 67 (M) = 1M) fa, whe: ddispiavciuucut, veiuciiy, and ace. eration, respectively]. Therefore, ihe velocity time his ory shows substantially less high- qnency motion than the acceleration time history. The displacement Give history, obtained by another round of integration, is dominated by relatively low frequency motion. 3.3.1.1 Peak Acceleration The most commonly used measure of the amplitude of a particnlar gronnd motion is the peak horizontal acceleration (PHA), The PHA for a given component of motion is sim- ply the largest (absolute) value of horizontal acccleration obtained fram the. aceelerogram of that component, By taking the veetor sum of two orthogonal components, the maximum resultant PHA (the direction of which will usually not coincide with either of the measured components) can be obtained Horizontal accelerations have commonly heea used tn deserihe. gronnd motions because of their natural relationship to inertial forces; indeed, the largest dynamic forces induced in certain types of structures (i.e. very stiff sretnres) are closely related to the PHA. The PHA can also be cortelated to earthquake intensity (e.g. Trifunac and Brady, 1975a; Murphy and O'Brien, 1977: Krinitzsky and Chang, 1987) Although this correlation is far from precise, it can be very useful for estimation of PHA when only intensity infor- mation is available, as in the cases of earthquakes that occurred befiwe strang moti ments were available (preinstrumental earthquakes). A number of intensity—acceleration relationships have been proposed, several of which are shown in Fi igure 3.41, The use of ‘intensity—attenuation relationships also allows estimation of the spatial variability of peak acceleration from the isoscismal maps of historical earthquakes. Tritunac and Brady 100 (1978a) - horizontal —»,”, Richter (1968) | e ” 3 Trifunac and Srady LB g E (1975a) - vertical f- Medvedev and . Sponheuer (1963) g ie Savarensky and - S Kirnos (1955) g * Kawasumi (1951) = 1 ~ |shimoto (1932) © JMA (Okamoto, 1973) O41 vu ieusity x Xi Figure 3.11 Proposed relaionships hatween PHA and MMI. (Alier Triftinac and Bracly, 19752. Used by permission of the Seismological Sacicty of America.)68 Strong Ground Motion Chap, 3 Vertical accelerations have received less attention in earthquake engineering than hor- izonlal accelerations, primarily because the margins of salety against gravity-induced static vertical forces in constructed works usually provide adequate resistance Lo dynamic forces induced by vertical accelerations during earthquakes. For engineering purposes, the peak vertical acceleration (PVA) is often assumed to be wwo-thirds of the PHA (Newmark and Hall, 1982). The ratio of PVA to PHA, however, has more recently been observed 10 be quite variable but generally to be greater than two-thirds near the source of moderate to Large earthquakes and Jess than two-thirds at large distances (Campbell, 1985; Abrahamson and Litehiser, 1989}, Pcak vertical accelerations can be quite large; a PVA of 1.74g was mea- sured between the Imperial and Brawley {aults in the 1979 Imperial Valley earthquake, Ground motions with high peak accelerations are usually, but act always, more destruc- tive than motions with lower peak accelerations. Very high peak accelerations that last for only a very short period of time may cause little damage to many types of structures. A nuwber of carthquakes have produced peak accelerations in excess of 0.5¢ but caused no significant dam- age to structures because the peak accelerations occurred at very high frequencies and the dura- tion of the carthquake was not long, Although peak acceleration is a very uscful parameter, it provides no information on the frequency content or duration of the motion; consequently, it must be supplemented by additional information to characterize a ground motion accurately . 2.2.4.2 Pea Veiusiiy ‘The peak horizontal velocity (PHV) is another uscful parameter for characterization of givund imution amplitude. Since the velocity is less sensitive to the higher-trequency comiponenis of the ground mation, as illustrated in Figure 3.10, the PHY is more likely than the PIIA to characterize ground motion amplitude accurately at intermediate frequencies, For structures or facilities that are sensitive to lading iu this intermediate-frequency range (oz. ie buildings, bridges, vic.), the VH.V may provide a much more accurate indication of the potential for damage than the PHA. PHY has also been correlated to earth- quake imeusity (e.g,, Trifunac and Brady, £9/3a; Krimtzsky and Chang, 198/). 3.3.1.3 Peak Displacement Peak displacements are gencrally associated with the Jower-frequency components af an earthey F often difficul: to determ ately (Campbell, 1985; Joyner and Boore, 1988}, due to signal processing errors in the liltering and inegration of acceleregrams end due to long period noise. Ae a result, peak displucement is Jess commonly used as a measure of ground motion than is peak acceleration or peak velocity. Fxample 30 Determine the peak accelerations. velocities, and displacements for the E-W componenis of the Gilroy No. 1 (rark) and Gilroy No. ? (soil) grauna matinne Solution ‘The peak amplitude values can be estimated graphically from Figure 3.9. Tae aemmal peak values. basnd on the data from which Fignes 3.0 was plotted, arc: Parameter Gilroy No. 1 (Rock) Gilroy No. 2 (Seal) Peak acceleration O.442¢ 0.332g Feak velocity (cawsoe) a3. sue Peak displacement (crn) 85 133Sec. 2.2 Ground Motion Parameters 69 3.2.1.4 Other An; eeETS, Aithough the parameters discussed previously arv easily determined, they describe only the peak amplitudes of single cycles within the ground motion time history. In some cases, damage may be closely related to the peak amplitude, butin athers it may require sev- eral repeated cycles of high amplitude a develop. Newmark and Hall (1982} described the concept of an effective acceleration as “that acceleration which is most closely related to. ctural we puiential Of an earthquake. It ditters trom and is less than ihe peak free-field ground acceleration. [Lis a function of the size of the loaded area, the fre- queney content of the excitation, which in tum depends on the closeness to the source of the earthquake, and to the weight, cinbedment, damping characteristic, dnd stiffness of the structure and its foundation.” : Some time histories are characterized by single-cycle peak amplitudes that are!much greeter than the itudcs uf vilivt cycies. An example of such a case 1s the Stone Canyon. record shown in Figure 3.12a, These single cycles often occur at high fiequencies and eonse- quently have little effect ou structures with lower natural frequenetes, In ather time histories, such as the Koyna record of Figure 3.126, a number of peaks of similar amplitude are observed. blir " @ ) Figure 3.12 Acczlerograms from (a) the NOW Melendy Ranch revocd of tse 1972 ‘Stone Canyon (Mf = 4.6) earthquake and (h) the longitudinal record trom the 1967 Koyna (M4 — 6.5) carthquakes, The timc and accclaaivn scales are identical tor hoth records. Peaks accelerations afe very clase, illustrating the limitations of using peak amplitude as a sole Incasure of strong ground motion, (After Hudson. 1979: sed hy permission of EER) Sustained Maximum Acceleration and Veloci ty. Nuttli (1979) used lower peaks of the aecclerogram to characterize strong motion by defining the sustained imaxinum acceleration for three (or five) cycles as the third (or fifth) highest (absolute) value of acceleration in the time history. The sustained maximum velocity was defined sim- ilarly. Although the PHA values for the 1972 Stone Canyon earthquake and 1967 Koyna earthquake records (Figure 3.12) were nearly the same. a quick visual inspeptinn indicates that their sustained maximum accelerations (shree- or five-cyele} were very different, Fora structure that required several repeated cycles of strong motion to develop damage, the70 Swong Ground Motion — Chap. 3 Koyna motion would be much more damaging than the Stone Canyon motion, even though they had nearly the same PHA. For these motions, the sustained maximum acceleration would be a better indicator of damage potential than the PHA. Example 3.2 Determine the three- and five-cycle sustained maximum acceterations and velocities for the E-W components of the Gilroy No. 1 (rock) and Gilray No. 2 (soil) ground motions, Solution ‘The sustained maximum acceleration und velocity values can be obtained graphically trom Figure 3.10. The actual values, based on the data from which Figure 3.10 was ploucd, ae: Parameter Gilroy No.1 (Rock) Gilroy No. 2 (Soil Sustained maxirium acceleration Three-cycle Oa3dy recycle OAl8¢ Sustained maxintuin velocity ern/sec) Three cycle 316 384 Five-eycle 29.9 38.2 Effective Design Acceleration. The notion of an effective design accelera- vin, with ditlerent detunitions, 42s been proposed by at least two researchers. Since pulses of high acceleration at high frequencies induce little response in most structures, Benjamin and Associates (| Y¥8} proposed that an effective design acceleration be taken as the peak acceleration that remains after filtering out eecelerations above 8 to 9 Hz. Kennedy (1980) proposed that the etfective design acceleration be 25% greater than the third highest (abso- lute) peak acceleration obtained from a filtered time history. 3.3.2 Frequency Content Parameters Only the simplesi of analyses (see Section B.5.3 of Appendix B} are required to show that the dynamic response of compliant objects, be they buildings, bridges, slopes, of soil depos- its, is very sensitive to the frequency al which they are loaded. Farthquakes produce com- plicated loading with components of motion that span a broad range of frequencics, The frequency conient describes how the amplitude of a ground motion is distributed among dit- ferent frequencies. Since the frequency content of an earthquake motion will strongly influ- ence the effects of that motion, characterization of the motion cannot be complete without consideration of its frequency content, 3.3.2.1 Ground Motion Spectra Auy periodic function {ie., any function that repeats itsetf exactiy at a constant inter- val) can be expressed using Fourier analysis as the sum of a series of simple harmonic terms uf dilfetcnt Srequeney, amplitude, and phase. Using the Fourier series (see Section A.3 ot Appendix A), a periodic function, x(4), can he written as ¢, sin (@,¢ + 6) GALSec. 3.3 © Ground Motion Parameters ” Tn this form, c, and ¢, are the amplitude ed phase angle, recpectivaly, of the mth harm: of the Fourier series [see equations (A.10) and (4.11) for their definitions]. The Fourier series provides a complete description of the grammd mation since the motion ean be com pletely recovered by the inverse Fourier transform. Fourier Spectra. A plot of Fourier umplitude versus frequency [c, versus @y from equation G.4)) is known as Fourier amplitude spectrum: a plot of Fourier phase angle (9, versus @,) gives the Fourier phase spectrum. The Fourior amplitude spectrin of a strong grotind motion shows how the amplitude of the motion is distri buted with respect to frequency (or period). It expresses the frequency content of a motion very clearly. The tourier amplitude spectrum may be narrow or broad. A narrow spectrum implies that the motion has a dominant frequency (or period), wiiich can produce a smooth, almost sinusoidal time history, A broad Spectrum corresponds to a motion that contains a variety of frequencies that produce a more jagged, j istory. The i tide spou- tra tor the E-W components of the Gilroy No. | (rack) and Gilroy No. 2 (soil) motions shown in Figure 3.13. The jagged shanes of the spectra are typical of those obsci ved for % 90 s a | | 2 zo $ 2 | & 10} 3 Wiawe 5 oe & im wl a 4 0 1 2 3 4 5 6 7 8 Pernod — sec 0 er : ; \ 20 ( g | 2 & 10 & 5 Lo 3a che ~— —<} 2 la 0 1 2 3 4 5 8 7 8 Pejiod — sec Figure 3.13 Fourie: ampliride spectra for the EY components of the Gilroy No. | {rock) and Gilroy No. 2 vil) strong auotiva records. Four spectra were cbtined by discrete Fourier transform (Scotion 4.3.3 of Appendix A) and consequently have units of vale sds aciva van ais Ne plotted as Tunchoas of trequency (see igure E3.3).Fourer amplitude (lag scale) 72 strong Ground Motion — Chap. 3 individual ground motions. The shapes of the spectra are quite different: the Gilroy No. 1 (Gud) opecisuum is sttunyest at iow periods (or high frequencies) wile uw reverse is observed for the Gilroy No. 2 {soil} record. A difference in frequency content can be detevicd by closely examining the motions in ihe Ume domain (Figure 3.1), but the uiffer- ence is explicitly illustrated by the Fourier araplitude specu. Whea the Fourier amplitude spectra of actual earthquake motions are smootied and plotied on logarithmic scales, their characicristic shapes can be seen more easily. As illus- uaicd iu Figuic 3.14, Fourier acceicration umpiiuudes tend .o be iargest over an jmermediate range of [requencies bounded by the corner frequency fon the low side and the cwloff fre- YueNCY figs OU the high side. The comer frequency can be shown theoretically (Brune, 1970, 1971) to be inversely proportional to the cube root of the seismic moment. This resolt indi- cates that large earthquakes produce greater low-frequency motions than do smaller earih- quakes, The cutoff frequency is not well understood: it has been characterized both as a f 2) and as a source effect Wapageorgion and Aki, ) usually assumed to be constant for a given geographic region, Figure 3.14 Téeuliced shape of smoothed Fovrier amplitude spectrum illustracing the & Trac comer frequency. /°, and cutoff frequency, Frequency (log scale) Saas Since phase angles control the times at which the peaks of harmonic motions occur (Appendix A), the Fourier phase spectrum influences the variation of ground motion with time. In contrast to Fourier amplitude spectra, Fourier phase spectra from actual earthyuake recoris do not display characteristic shapes. Power Spectra. The frequency content of a ground motion can also be described by apower spectrum ox power spectral density finction. The power specual density function can also be used to estimate the statistical properties of a ground motion and to compute sta- chastic response using random vibration techniques (Clough and Penzien, 1975; Vanmarcke, 1976, Yang, 1986). ‘The total intensity of a ground motion of duration 7, is given in the time domain by the area under the time history of squared acceleration: 1, iy = face a GS) 0Sec. 3.3. © Ground Motion Parameters 72 Using Parseval’s theorem, the total intensity can alsa he expresced in os he Effet des (3.6) z ° iAtis the Nyquist frequency (the highest frequency in the Fourier serics), The average intensity, J, can be obtained by dividing cywativus (3.5) and (3.6) by the duration. Ts ly at age Lf do = pf ten dr = ar faa GD 0 ° Notice that the average insensity is equal to the mean-squared acceleration. The power spec- defined suen that intel density, Gu), oy Ay few deo G.8) a from which we can easily see, by comparing equations (3 7) and (3.8), 6 = Le a9 aty The close zelationship between the power speciral density Fanction and the Fourier ampli- ucle spectrum is apparent trom equation (3.9). The power spectral density is often normal- ized by dividing its values by the area beneath it Grin) = Lew) G.10) dy where Ap, us before, is the mean-squared acceleration, The power spectral density function is useful in characterizing the earthquake as a random process. The power spectral density function by itself can describe a stationary ran- 3 (ice, ON whuse slatistical parameters da not vary with time}. Actual strong motion accelerograms, however, frequently show thal Lhe intensity builds up to a maxima value in the carly part of Uc unuiion, then remains approxunately constant for a periad of time, and finaily decreases near the end of the motion. Such nonstationary random process behavior is often modeled by multiplying a Stationary time history by a deterministic inten- sity function (e.g., Hou, 1968; Shinozuka, 1973; Saragoni and Hart, 1983). Changes in fre- suring the motion nave been described using an evolutionary power spectrum approach (Priestley, 1965, 1967; Liu, 1970). Response Spectra. 4 tinrd type of spectrum is used extensively in earthquake engineering practice, The response spectrum describes the maximum Tesponse of a single- degree-of-freedom (SBOF) system to a particular input motion as a function of the natural frequency (or natural period) and damping ratio of the SDOF system (Section B.7 of Appeudia B). Compuced response spectra lor the Gilroy No. I (rock) and Gilray No. 2 (soil) fecords are iliustated in Figure 3.15, quency conte74 Strong Ground Motion Chap. 3 Gilroy No.1 frock) Gilroy No.2 (soit s, 1 © os 1 2 3 190 100 Ss, BQ. S50 ‘omsee) temicacy av i L 0 1 2 3 £9 1 2 3 20 I S310 - fom ~ 9/4 0 1 2 3 Poriod, Tisec) Period, Tisec} Figure 3.15 Response spectra (5% damping) for Gilrey No. t (rock) and Gilroy No. 2 Coil) & the response spectra, The Gilroy 1 jrock) motion, for example, produced higher spectral aceelorations at low periads than did the Cilkoy 2 (S213) motion, and Jewer spectral aceetorations at highcr periods. The higher long-period content af che Gilroy 2 (soil) motion produced spectral velocities and displacements atuck higher than those of the Gilroy 1 (cock) motion plotted individually t 2 plotted lyf bined, by virtue of the relationships of equation (3.11), in Wipartive plots (Section A.2.2). The tripartite plot displays spectral velocity on the verlical axis, natural frequency (or petiod) on the horizontal axis, and acceleration and displacement on inclined axes, The acceleration and displacement axes are reversed when the spectral values are plotted against natural period rather than natural frequency. The shapes of typical response spectra indicate that peak sp crithmetic scales, or may be ¢. ‘0 associated with tacoment * placoment value: ferent frequencies (or periods). AL low frequencies the average spectral displacement i is nearly constant; at high frequencies the average spectral acceleration is fairly constant, In between lies a range of nearly constant spectral velocity. Because of this behavior, respons spectra are often divided into acceleration-controtied (high-frequency), velocity-controlled (intermediate-frequency), and displacement-controlied (low- Frequency) portions, Fi velocity, and elocity, and aceeleratio: response apectra assume Hncar struc: ceponse epectra assume Lincar struc! many real structures, however, inelastic behavior may be induced by earthquake ground i fores diss nt bohavis al foree displaccment bchaviSec. 3.3 Ground Motion Parameters 15 motions, An inelastic response spectrum (i.¢., one that corresponds to a nemiinear farce displacement relationship, can be used to account for the effects of inel: lastic behavior. Kig- ure 3.16 shows inelastic response spectra for acceleration and yield displacement for vari ous values of the ductility factor Wl = tiga/ity, WhEIC Uggs 18 the maximum allowable displacement and x, is the yield displacement. A separate inelastic spectrum must be plotted to show total (clastic plus plastic) displacement, Spectral accelerations decrease with increasing ductility, put total displacements increase. Sy HS Uygdtt, | PHY H ps ! ! 9.08 ii, Ye ax 0.03 LS! Displacement 0030.05 01 62 05 1 2 §& 410 2020 Undamped natusal frequency (Hz) Figure 3.16 [uelastic rospaase spocina for the El Centra N-S component of the 1940 Imperial Valley earchevake. Oaly the elastic component of the displacement is plot Spectral accelerations are correct, but spectral velocities are not, (After Newonark and sei by permission OF RERL) Response spectra reflect strong grommd motion characteristics indirectly, since they are “filtered” by the response of a SDOF siructure, The amplitude, trequeney content, and toa lesser extent, duration of the input motion all influence spectral values, The different frequency contents af the Gilroy No, 1 (rock) amd Gilroy No. 2 (soil) ground motione are clearly illustrated by the different shapes of their Tespectiv igure 3.15}. Tt is important to remember that Tesponse spectra represent only the maximum responses of a number of different structures: However, the zesponse of structures is of grcat importance in earthquake engineering, and the response spectrum has proven to be an important and useful too! for characterization af strong ground motion. pon: it 3.3.2.2 Spectral Parameters Section 3.3.2.1 described throe types of spectra that can he used to characterize sirong ground motion. The Fourier amplitude Spectrum and the closely related power spectral den- sity, combined with the phuse spectrum, can descrihe a #round motion completely. The response spectrum: does not describe the actu! ground motion, but it does provide valuable additional information on its potential effects on structures. Rach of these spectra is a com- Plicated function and, as with time histories, a great many data are required to describe them compieiciy. A number ot spectra! parameters have been proposed to extract important Pieces of information from each spectrum.76 Strong Ground Motion — Chap. Predominant Period. A single parameter that provides a uselul, althoug somewhat crude representation ol the Lrequency content of a ground motion is the predon inant period, T,. The predominant period is defined as the period of vilyatign pgergsspondi to the maximum value ot the Founer amphtude spectrum. ToS¥dihtnidae mnftuetice of ind vidual spikes of the Fourier amplitude spectrum, the predominant period is often obtaine from a sMOetHéd spectrum. While the predominant period provides some informatic regarding the frequency content, it is easy to see (Figure 3.17) that motions with radical different trequency contents can have the same predominant period, Period Figure 3.17 Two hypothetical Fourier amplitude spectra with the same predominant period bul very diffcrent frequency contents. The upper curve describes a wideband: Taotion and the lower a narrowband motion. Example 3.3 Determine the predominant periods for the E-W components of the Gilroy No. 1 (rock) and G toy No. 2 (soil) ground mations. Solution ‘The Fourier amplitude spectra of most ground motions are quite jagged in the vici ity of their peaks, sq same smoothing is required to identify the predominant period, T smouthing and predominunt period identification is most easily accomplished by plotting t Fourier amplitude spectrum as a function of frequency. By numerically smoothing their Four amplitude spectra (Figure 13.3), the predominant periods are Gitroy No.1 (rock): 1, = 0.39 sec Gilroy No. 2 (soil) T, = 0.53 see ote thar the predomment period of the Giiroy ‘Vo. Z (soli) motion is greater du that of the C roy No. 1 {tock} motion, thereby illustrating the relative strength of the longer period (low trequcucy) ummpnenis of the Gilroy No, 1 (scil} motion, Bandwidth. — The predominant period can be used to Jocate the peak of the Fo tier amplitude spectrum: however, it provides no information on the dispersion of spect amplitudes about the predominant petiod, The bandwidth of the Fourier amplitude spe frum is the range of Irequency over which some level of Fourier amplitude is exceed Bandwidth is usually measured at the level where the power of the spectrum is half its me imum value; this corresponds Lo a level of 1/./2 times the maximum Fourier amplituc ‘The irregular shape of individual Fourier amplitude spectra often renders bandwidth dif cull to evaluate, It is determined more easily for smoothed spectra. Central Frequency. The power spectral density function can be used to es mate stalistical properties of the ground motion. Defining the nth spectral moment o ground motion hy ~Sec. 3.3- Ground Motion Parameters qT e s 2 8 (a) 3 ° ° = Fourier amplitude (g — sec) Khan os LITWata lar nlite ieeetge gs | § 10 15 20 25 Frequency - Hz 30 20 b) Fourieramplitude (g - sac) She EE ro endantane 0 5 40 15 20 25 Prequency - Iz ° Figure 63.3 Raw and smoothed ['A§ for Gilroy No.1 (rocky and Gilroy No.2 (oil) motivas. Oy hn = force) da BAL 9 the ceniral frequency Q(Vanmareke. 1976) is given by 3.12) ‘The central frequency is a measure of the frequency where the power spectial density is von centrated, It can also be used, along with the average intensity and duration, 10 calculate the theoretical median peak acceleration G13 ab Fan ~ Shape Factor. Tie shape foctor (Wanmarcke, 1976) indicates the dispersiua of the power spectral density function about the centtal frequency: —_— d= ft yo The shape factor always lies between 0 and 1, with higher values corresponding to larger bandwidths. G14)78 Strong Ground Motion — Chap. 3 Kanai-Tajimi Parameters. Although individual power spectral density fune- tions may have highly irregular shapes averaging a number af normalized power spectral density functions for similar stcong ground motions reveals a smooth characteristic shape, Kanai (1957) and Tajimi (1980) ised a Hmited number af strang motion records to propose the following three-parameter model for power spectral density: 1+ 128, (0/0,)1* TC yeP + DE, (m/e? Gla) = Gy (3.15) where the parameters G,, £, , and a, determine the shape of the function (Figure 3.18). Gem a te Figure 3.18 Shape of the Kanai -Tajimi By power spectral density function, The displacement response ofa SDOF system with natural frequency @, and damping Tatlo S» to white noise base motion would be described by a Kanai-Tajimi power spectral density function, As such, high-frequency components of the input motion will be attenu- ated, and frequency components in the vicinity of ¢o, will be amplified. Typical values of Kenai—Tajimi parameters for various site conditions are shown in Table 3-1 Clough and Penzien (1975) proposed a correction to the Kanai—Tajimi power spectral density function to prevent excessive velocities and displacements at very low frequencies, The corrected Kanai—Fajimi power spectral density function requires two additional param- eters to describe the power spectral density 3.3.2.3 Vinae/@max Because peak velocities and peak accclerations are usually associared with motions of different frequency, the ratio vpaedng, should be related to the frequency content of the UK. 1973, Secdcial., 1976, MoGuite, 1978). For a sinipie ter monic raotion Table 3-1 Ground Intensity, Cround Frequenuy, ond Ground Denapiny fur Yatious Site Conditions Ground Ground Ground ‘Ground Site Number of Tntensily, Frequency, Damping, Mutu Condiiinns Keeores, Gy ®, Ss Horizontal Alluvium 161 0.102, [84 0. Alluvinm on rock 60 O07, 229 0.30 Rock 26 0.070 270) 034 Vertical afluvium 8 .080 262 946 Aliuvium «m rock oo 0.072 29.1 nae Ron a 0.083 38 146 Source: Elghadarusi et al. (1988). .Sec. 3.3 ‘Ground Motion Parameters 79 of period 7, for example, Ymad@yax = 1/27. For earthquake motions that include many dicquencies, the quantity 20 Vmug/dman) can De interpreted as the period of vibration of an equivalent harmonic wave, thus Providing an indication of which periods of the gropnd tution are most signtticant. Seed and idriss (1982) suggested the following representative. average values for different site conditions less than 50 km from the source: eee Site Condition vmaxemar ee Roek SScnvsecly = 0.056 sec Soff veils (<200 4; Lideniseeig = 0.112 see Deep Sti soits (6200 1) 135 eunfsee/g = 0.138 see ee ‘The corresponding periods of equivalent harmonic waves for the rock, stiff soil, and deep SAE soil site: conditions ste (1.25 cee, 0.78 sve, aiid 0.87 cv, respectively, which indicates a shift toward longer-period (lower-freynency) motion on softer soil deposits, Example 3.4 Determine the ratio Va /égx forthe N-S components of the Gilroy No. 1 (rock) and Gilroy No. 2 Goi!) ground motions. Comysare the quantities: 2(/yauétygx) with the predominant peri= ods of the mations, Solution Based on the via. mi 4, ay 33.7 emisec Gilroy No.1 (rock): tom 2 __ 33 7emsee _ yo7e , BRENDA (ook: > OEE GST ome ee Gilroy No.2 (soil): teas _ ___39.-2em/seo__ oy 104 see @max 0.322 (981 emia?) The quantity 27(,4,/2r94) 8 equall 6 the predominant period of a simple harmanir mavian, To See flow well it corresponds fo the predominant period of the Gilroy No. | (rock) and Gilroy No. 2 (soil) ground motions, Gilroy No.1 (rock): InP = 0.49 see Ty = 0.39sec Gro Gilroy No.2 (soil) 2ntm% = 078 see rp = 0.5380 Though the ratio Yng/amax emtainly Tadicates that Lhe Gilroy No. | (rock) motion has « higher frequency content tian the Gilroy No. 2 (oil) motion, it everesti:ualys the predominant period of both the Gilroy No. I (rock) motion and Gilroy No.2 (soil) motions. Due to the approximate nature of the predominant petiod and the stochastic name of beth py, ard dqup close agroc~ ment between Yoa/dma, and predominant period should not be expected. The duration of strong ground motion can have a strong influence on earthquake damage ‘Many physical processes, such as the degrecation of stiffness and strength of certain types of structures and the buildup of porewater pressures in loose, saturated sands, are sensitive to the number of load or stress reversals that occur during an earthquake. A motion of short duration may not produce enough load reversals for damuging response te build ast ture, even if the amplitude of the motion is high. On the other hand, a motion with moderate amplitude bur long duration can produce enough load reversals to cause substantial damage,80 Strong Ground Motion Chap. The duration of a strong ground motion is related to the time required for release 0 accumulated gtr: ture along the faull. A pture along the faut. / increases, the time required for rupture increases. As a result, the duration of strong motio increases with increasing earthquake magnitude. While this relationship has been supporte by empirical evidence for many years, advances in source mechanism modeling (Hanks an McGuire, 1981) have provided theoretical support, indicating that the duration should b Proportional to the cube root of the seismic moment. When bilateral rupiure |Le., ruptur (a ofthe 1989 Loita Prici earthquake] occurs, the strong motion duration may be considerably lower. An earthquake aceelerogram generally contains all accelerations from the tine in earthquake begins until the time the motion has returned to the level of background noise For engineering purposes, only the strong motion portion of the accelerogram is of inicres| Diffetent approaches have heen taken to the problem of evaluating the duration of stra notio fon (Bolt, 1960) is defined as the i between the first and last exceedances of a threshold acceleration (usually 0.05¢). Anothe detinition of duration (Urifunac and Brady, 1975b} is based on the time interval betweon Ln points at which 5% and 95% of the total energy has been recorded. Boore (1983) has taker the duration to be equal to the corner period (i.¢., the inverse of the comer frequency). Thi tate of chenge of cumulative root-meane square (rms) acceleration has also been used as th revaluation of at propagutes in apposite directions from the fo that opposi ns from the fo an aceslerogram. Tho bn and Shah. 1979), Power syccira density concepts can also be used to define a strong-motion duration (Vanmarcke and Lai 1977), Other definitions of strong-motion duration have been proposed (Perez, 1974; Tri funac and Westermo, 1977). Because it implicitly reflects the strength of shaking, the bracketed duration is most commonly used for earthquake cngineezing purposcs. The d duration of strong motion has been investigated by interpretation of accelero MeCa o livid avccleiation Chaag and i Krinitercy (1977) estimated the bracketed durations for soil and rock sites a short (lest than 16 km) cpiceniral distances shown in Table 3-2. Duration has also been expressed in terms of equivalent cycles of ground motion. On: such approach was developed in conjunction with aa carly procedure fer evaluation of lig ucfiction potential (Seed et al., 1975). To represent an irregular time history of shear stres Table 3-2. Typical Earthquake Durations at Epicentral Distances Less Than 10 km Duration (sec) Magnitude: Rock Sis Soil Sites 3, 4 $ 35 6 2 60 8 lo 53 n B 70 16 2 75 2 43 40 31 62 Ro a 84 Source: Chang and Krinitesky (977Sec. 3.3. Ground Motion Parameters 8 by a uniform series of harmonic stracs cyclen, the tof an equivalent mt nificant stress cycles was developed, The equivalent number of uniform stress cyctes shown in Table 3-3. was selected to cause pore pressure buildup cquivalent to that of a actual shear stress-time history at a harmonic stress amplitude of 65% of the maximun actual shear stress, Tabla 33° Equivalent Number of Uniform Stroco Cycles ee ee Earthquake Number of Signiticant Maguitude Suess Cyelos 24 5 10 is 26 Example 3.5 Determine the bracketed durations of the E-W components of the Gilroy No. 1 (rock) and Gil roy No. 2 (soil) ground motions. Solution Based on z threshold acceleration of 0 Og, the bracketed cur: graphically irom the accelerograms shown in Figure E35. 0.5 — = | First 3 excesdance = f & | ‘i Lit a) Bou i DRL Arenson ttn yyemrnc ewer nf 5° SAE Mee : § i | Last 0.059 exceedance 0.5 0 3 10 15 29 25 30 35 40 Time (sec) Os Last exceedance (b) Accalera ion (g) ° exceedance 2 aw a 5 40 18 20 25 30 35 20 Time (seo) Figure R3.582 Strong Ground Moiion Chap, 3 Giltoy No.1 (rook): fy = 98 see Gilroy No.2 (soul): Ty = IAT sec 3.3.4 Other Ground Mo! mn” Parameters The preceding parameters related primarily 10 the anaplitude, frequency content, or duration of'a ground motion, Since all of these characteristics are important, gcound motion param- élers that reflect more than one are very useful. The following paragraphs prcscui a umber of parameters that refleet two or three important ground motion characteristics, Assingle parameter thal includes the effects of the amplitude and frequency content of a strong motion record is the rns acceleration, defined as [he dine = {a fi ated = {Ty id) jas 5 a where Ty is the duration of the strong motion and hs is the average intensity (or mean- squared acceleration). Because the integral in equation (3.16) is not strongly influenced by farge, high-ficquency accelerations (which occur only over a very short period of time) and because it is influenced by the duration of the motion, the rms acceleration can be very use- ful for ongincoring purposes. its value, however, can be sensitive to the method used io define strong motion duration, A parancter closely related to the rms accelcration is the Arias intensity (Arias, 1970), defined as unk fiw yar G17) The Anas intensity has units of velocity and is usually expressed in meters per second, Since it is obtained by integration over the entire duration rather than over the duration of Strong motion, its yaluc is independent of the method used to define the duration of sitong motion, example 3.6 Determine the rms accelerations and Afias intcusities of the F-W components of the Gilroy No. 1 (tock) and Gilroy No. 2 (soil) ground motions. Solution By integrating the aecelerograms of the Gilroy No, 1 (rock) and ground motions numerically, the rms accelerations and Arias intensitics are roy No. 2 (soil) Gilroy No.1 (rack) fous — OLD, Gilroy No.2 (soil): Hing = 0,072, : 8 8 R a The characteristic intensity, defined as de = aT is related linearly to an index of structural damage absorbed hysteretic energy (Ang, 1990).Sec. 3.3. Ground Motion Parameters 83 The cumuintiys velocity is simply the area under the absolute acvelerogram: CAV = Jie (olat ‘ B.19) ° The cumulative absolute velocity has been found w correlate well with strictival damage Potcatial. Tur caampie, a CAV ot U.20g-sec (obiained after fi tering out frequencies above 10 Hz) corresponds to the lower limit for MMI VII shaking (Benjamin and Acsociates, 1988), Since many structures have fundamental periods between 0.1 and 2.5 see, the response spectrum ordinates ia this period Tange should provide an indication of the poten- tial response of these structures. The Fesponse spectram intensity (Honsner, 1959} y therefore Ucfined as 25 SUS) = fosv (& D4? 43.20) a Gee, the area under the pseudovelocily responce spa 22 sec. The response spectrum intensity, as indicated in equation (3.20), can be computed for any structural damping ratio. It captures important aspects of the amplitude and fre- Suency content (in the range of primary importance for structures) in a single parameter. Von Tawa et al, (1988) referred to the response spectrum inten sity for 5% daniping 2s the velocity spectrum intensity. The velocity spectrum inten: sity was suggested as being use- ful for evalnation of the response of earth and raciefill dams, whieh typically lave func mental periods between 0.6 and 2.0 see (Makdisi and Seed, 1978). To characterize strong, ground motion for anaiysis of concrete dams, which generally have fundan rental periods of Jess than 0.5 sec, Von Thun et al. (1988) introduced the acceleration spectrum intensity, defined as periods uf G1 see and os ASI = fous 05,7) at G21) Mh (Ge. the area under the acceleration response spectrum between periods ef 0.1 see and 0.5 sec). The Applied Technology Councit (1978) defined two factors by which standard Fesponse spectra cond he ny The effective peak acceleration (EPA) was defined as the average spectral accelcration over ihe period range 0.1 t0 0.5 soc divided by 25 (the Stindard amplification tactor or a 5% danwping spectrum). the effective peak velocity (BPV) vas defined as the averoge socctral velocity ata peviod af | see divided by 2. Determination of EPA and EPV is shown schematically in Figare 3.19, the process of ayer ‘ging the spectral accelerations und velocities over a range of peciods minimizes the tata nec of local spikes in the responce84 Strang Ground Motion Chap. 3 | » vo a oo : : 8 - 7 e 8. & EPA = 38 ! - | 04 ve i 5, | EPV = Bs | O41 Os 1 5 10 50 Period (sec) Figure 319 Determination ol eftective peuk acceleration and etfective peak velocity trom response spacira, (After Applied Technology Couneil, 1978.) 3.3.5 Discussion A wide variety of strong motion parameters have been presented, Some desenbe only the amplitude of the motion, others only the frequency content or duration. Some of these parameters are fluenced by (wo or three of these important ground motion characteristics. Table 3-4 indicates which ground motion characteristics strongly influence the various ground motion parameters. Seismic hazard analyses (Chapter 4) and the devclopment of design ground motions (Chapter ¥) rely heavily on the characterization of strong ground motion by ground motion parameters, Characterization by a single peramete: is only rarely appropriate; the use of several parameters is usually required to describe adequately the important characteristics of a particular ground motion. Since different engineering problems are influenced by dif- lereal ground motion characteristics. the significance of ditteren parameters depends on the types of problems for which they are used. 3.4 ESTIMATION OF GROUND MOTION PARAMETERS Proper design of carthquake-resistant structures and fi 5 requires cstimation of the level of ground shaking to which they will be subjected. Since the level of shaking is mostSec. 3.4 Estimation of Ground Motion Parameters 85 Table 3-4 Ground Motion Characteristics That Are Stronaly Reflected in Various Ground Motion Parameters Growad Motion Claaactesist Ground Motion Parameter Amplitude Frequency Content Duration Poak acceleration, PHA smd PHY % Pak velonity, PHV « Sustained maxistium seccleration, SMA x Effective desipn acceleration, EDA x Predomiinsnt period, 7, % Bandwidth x Central frequency, 2 x Shape factor, 8 x Power spectrum intensity. Gp « v * Ground frequency. ay x Ground damping, &, x eran ans x Daradion, £5 x Tins aceeleratiON, 45 x x Characteristic intensity, J, x x x Acias intensity, J, x x x Cumulative absolute velocity, CAV x x x Response spectrum intensity, SI x x Velocity specu uu intusity. VSI * x Acceleragion spectrum intensity, AST x x Liftective peak acckecation, FPA, x x Eficetive peak velocity, EPV x “ conveniently described in terms of ground motion parameters, methods for estimating ground motion parameters are requited. Predictive relutionships, which express a particular ground motion parameter in terms of the quantities that affect it most strongly, are used t estimate ground motion parameters. Predictive relationships play an important role in scis- mie hazard analyses (Chapler 4) used for seismic design. Much of the energy released by rupture along a fault takes the form at stress waves. Since the amount ol cnergy released in an earthquake is strongly related to its magnitude, the characteristics of the stress waves will also be strongly related to magnitude. Figure 3.20 illustrates the influence of earthquake magnitude on sctual ground motion characteristics in the Gime domain. Each earthquake came from essentially ihe same sourre, amd aac! rogram Was measured at about the same dislance from the source. The variations in ampli- tude, frequency content, and duration with magnitude are: appareni As stress Waves travel away [rom the source of an carthquake, they spread out and are partially absorbed by the materials they travel through. As a result, the specific energy (cnergy per unil volume) decreases with increasing distance {rom the source. Since the characteristics of stress waves are strongly related 10 specific energy, they strongly related to distance. The distance between the source of an cz thquake and a86 Strong Ground Motion Chap. 3 . Lereleration wear scale ~ 4 . too [Mf ess ’ Figure 3.20 Accelerograms from six, ! earthquakes off the Pacitic coust of Mexico. i Lach acceierogram was measured at. nearly the same epicentral distance. The reeord uT ay oo fauna tie Af 8.4 (19K Michoacan Aecoloration (emisasicee) = i nt & earthquake coninues for anorher 25 sce. (After Andersan, 1991. Geotechnical News, o 10 20 30 4 50 Vol. 9. No. I, p. 35. Used by permission of Time (sec) BiTeeh Publishers, Led.) particular site can he interpreted in different ways. Figure 3.2 illustrates some of the most commonly used measnrec nf distanca RI and Rare the hypocentral and epicontval dic. tances, which are the easiest distances to determine alter an earthquake, If the length of faulk rupture is a Significant fraction of the distance hotween the fault and the site, however, cnergy may be released closer to the site, and RI and R2 may not accurarefy represent the “effective distance." R3 is the distance to the zone of highes1 energy release. Since ruprite of this zone is likely to produce the peak ground motion amplitudes, it represents the best distance measure. tor peak amplitude predictive relationships Tnfneminately, its locatinn is difficult to determine alter an carthquake and nearly impossible to predict before an earth- quake. R4 is the clasest distance to the zone of rupture (not including sediments averlying basement rock) and RS is the closest distance to the surface projcction of the fault rupture. Ré and R5 have horh haen used a in predictive relationships, Surface projection Site a Epicenter fal ase NX : High-stress 2 zone -———~ Figure 3.21 Various measures of distance used in siong-motion predictive relationships, Me (Alter Shakal and Becpseuter, 1981.) Hypocenter f 3.4.2 Development of Predictive Relationships Fault rupture surface Predictive relationships usually express ground motion parameters as functions of magni- tude. distance. and im some cases, other variables, for example, Y= f(M,R,P) (3.22)Sec. 3.4 Estimation of Ground Motion Parameters a7 where Y is the orannd mati: zi uf ihe cariiquake, & a measure of the distance from the source to the site being considered, and the P, are other parameters (which may he used to characterize the earthquake souce, wave propapauion path, and/or local site conditions). Prediétive relationships are developed by regression analyses of recorded strong motion databases. As such, they change with time as additional strong motion data become available. Mast predictive relationships are updated in the lit- erature cvery 310 5 years or shortly after the occurrence of large castlyuahes in wellinsiru- mented regions. The functional form of the predictive relationship is usually sclected to reflect the mechanics of the ground motion process as closely as possible. This minimizes the number of empirical coefficients and allows greater confidence in application of the pre- dictive relationship t© conditions (magnitudes and distances} that are poorly represented inthe datahase Cammon farm, loth observations: tclationships ac bascd vu tie following 1. Pouk values of strong mnotion parameters are approximately lognormaliy distributed Ge., the logarithms of the parameters are approximately normally distributed}. As a result, the regression is usually performed on the logarithm of ¥ rather than on Y itself, 2. Karthquake magnitude is typically defined as the logarithm of same peak motion parameter (Scetion 2.9.2). Consequently, In ¥ should be approximately proportional tom, |, page . . 3. The spreliding of stress waves as they travel away trom the source of an carthquake causes body wave [p- andl s-wave (Section 5.2.2.5)] amplitudes to decrease according to VR and surface wave [primarily Rayleigh wave (Section 5.3.1)] amplitudes to decrease avcording to 1/./8 4. The arca over which fault rupture occurs increascs with increasing earthquake mag- nitude (Section 4.2.1.2), As a result, some of the waves Uhl produce strong motion at a site arrive from a distance, X, and some attive from grcater distances, The effective distance, therefore, is grealer than 2 by an amount that increases with increasing magnitude. a5 itigy Lanivd vy suress waves is absorned by the materials they travel through [material damping (Section 5.5.1)]. This material damping causes ground motion amplitudes to Uevivase exponentially with K. 6. Ground motion parameters muy be influenced by source characteristics (e.g. strike- Ha slip, normal, or reverse faulting) or site characteristics (e.g. hard rock, sofi roc! vium, ete.}. Combining these observations. « typical predictive relationship may have the form In¥ = G+ OM+ CM + Cyln[R + C, exp(GMyl Sse Se @ @ @ a # f{souee} + fisite) Oy, = Cy C at @88 Strong Ground Motion Chap, 3 where the circled numbers indicate the observations associated with cach term. Some pre- dictive relationships utilize all these terms (and some have even more) and others do not. ‘The Gjyyterm describes uncertainty in the value of the ground motion parameter given by the predictive relationship. Statistically, it represents an estimate of the stmdard devia- tion of la Yatthe magnitude and distance of interest. Historically, most 6j,y values have been constants, but several recent predictive relationships indicate dj,-values thal vary with mag~ nitude. Ata given magnitude, therefore, the probability that the ground motion parameter will exceed a value ¥* would be 1 - F,,(z*) where F(z") is the value of the standard cumu- lative distribution function (see Section C7.2 of Appendix C) at (In ¥* — In ¥}/a,, When using any predictive relationship, it is very important to know how parameters such as Mf and R are defined and lo use them ina consistent manner. It is also important to recognize that different predictive relationships are usually obtained from different data sets. To make reasonable predictions of ground motion parameters, a predictive relationship based on data that are consistent with the conditions relevant to the prediction is required, 3.4.3 Estimation of Amplitude Parameters Predictive relationships for parameters that decrease with increasing distance (such as peak acceleration and peak velocity) are often referred to as attenuation relationships, A few of 2 large number of useful attenuation relationships for different geographic and tec- iuuie caves arc described in the Toiiowing sections. 3.4.3.1 Peak Acceleration Since peak acceleration is the most commonly used ground motion parameter, many peak acceleration attenuation relationships have heen developed. All are best suited to con ditions similar (o those in the databases from which they were developed. As additional strang motinn dats have become ailuble, attenuation relationships have become mora retined. Consider, for example, lwo aticnuation relationships developed some 13 years apart, In 1981, Carnpbell (LO81) used worldwide data to develop an attenuation relationship (or the mean PHA for sites within 50 km of the fault rupture in magnitude 5.0 to 7.7 earthquakes: InPHA(g) = —4.141 + 0.868.M — 1.09 In| R + 0.0606 exp(0.747}] Sinus = 037 (3.24) where M is the local magnitude or surface wave magnitude for magnitudes less than or grearer than 6, respectively, and X is the closest distance to the fault rapture in kilometers, In this relatively simple attenuation relationship, which represented the state of the art in 1981. the peak seccleration was taken as a function of M and Ronly and Gy,py Was con stant. In 1994, Campbell and Bozorgnia (1994) used worldwide accelerograms from earth- qnakesaf moment magn dn PHA(gals) = — 3.512 + 0.904M§,,— 1.328 In ,/R2 +(0.149exp (0.6470, +(1.125 — 0.112 nk —0,0957 MJ # + (0.440 — 0.171 INR} SsR + (0.405 — 0.222 in R)Siie (25) fogeo oneoLM = M<74 Sn = REA jo M>TASee. 3.4 Estimation of Ground Motion Parameters 89 where R is the closest distance (< 60 kin) to seismic rupture in kilomezere (arith . values of 7.3, 5.8, 3.5, and 3.0km for magniludes of 5.0, 5.5, 6.0, and 6.5, respectively); the source term, F, takes on values of 0 for strike-slip and normal fanlting, and | for reverse, reverse-oblique, and thrust faulting: Sg = 1 forsoft-rock siles (sedimentary deposits of Ter- Uary age), Sux = 1 for hard-tock sites (primarily older sedimentary deposits, metamorphic rock, and crystalline rock), and Ssx ~ Syyy = 0 for alluvium sites. The 1994 relationship. which is based on more data, is clearly more specific (and more compliratod) than the 1087 relationship. The incorporation of additional terms reflecting source and site characteristics are typical of the refinement of predictive relarionshins that has taken place in recent years. Boore ct al. (1993) used data from western North America earthquakes of magnitude 5.010 7.7 at distances within 100 km (62 mi) of the surface projection of the fault wo develop the predictive relationship log PHACs) — by + bai, 6) + dsm, — 0)? + OR 4 Bs lOBR + bgGp + brGe 3.26) whereR = Jd? +12, dis the closest distance to the surface projection of the fault in hilu- meters, and 0 for site class A le for site class A Gr=}) for site class B 10 favsite class R [0 for sito elass C 1 for site class C Note that the Buore et al. (1993) attenuation relationsh:p is expressed in terms of the com- mon (base 10) logarithm rather than the natural logarithm. The site classes are defined on the basis of the average shear wave velocity in the upper 30 m (100 ft) (Table 3-5). Coef- ficients for the Boore ct al. (1993) attenuation relationship were developed for two mea- of peak avceieiatian: the randomiy onented component and the lar ger horizontal component {the former considers two orthogonal horizontal tecords at a Particular site as scpuate cvcrts and the Jutter considers only the larger of the two). Thc coefficients are given in Table 3-6. Table 3-5 Definitions of Site Classes for Boore et al. (1993) Attenuation Relationship Sie Classi Upper 30.mi (100 fy A > 750 avsee (2500 ft/sec) B 360--750 mfsee (1200-2500 fidsec) C 180-360 m/sce (600-1200 fuser) Table 3-6 Coefficients for Boore et al. (1993) Attenuation Relationship Component a ty Fy by bs % by A Ghegriia Random = nis 77a nga ates Laer 003% 0216 90 00 -0; 0.158 0.254 5.48 0.205