Canada I INFO-0604'1 I 1 I I I

INFO 060M

CA9600179

27fc 15

Atomic Energy Commission de contrôle Control Board de l'énergie atomique Canada I INFO-0604'1 I 1 I I I

I AECB Workshop on Seismic Hazard Assessment I in Southern Ontario — Program, List of Participants I and Abstracts I Ottawa, Ontario June 19-21, 1995

I I i 1 I t -) I I

NEXT PAGE(S) I loft BLANK I Atomic Energy Commission de contrôle I 1*1 Control Board de l'énergie atomique Canada I

I -111- AECB WORKSHOP ON SEISMIC HAZARD ASSESSMENT IN SOUTHERN ONTARIO - I PROGRAM. LIST OF PARTICIPANTS AND ABSTRACTS I PREFACE A workshop on seismic hazard assessment in southern Ontario was conducted on I June 19-21, 1995. The purpose of the workshop was to review available geological and seismological data which could affect earthquake occurrence in southern Ontario and to develop a consensus on approaches that should be adopted for characterization of seismic I hazard. The workshop was structured in technical sessions to focus presentations and discussions on four technical issues relevant to seismic hazard in southern Ontario, as I follows: (1) The importance of geological and geophysical observations for the determination of seismic sources. (2) Methods and approaches which may be adopted for determining seismic sources based on integrated interpretations of geological and seismological information. (3) Methods and data which should be used for characterizing the seismicity I parameters of seismic sources. (4) Methods for assessment of vibratory ground motion hazard.

I This document presents a copy of the workshop program, the list of participants and extended abstracts received from speakers. It was distributed to the participants prior to the workshop. The abstracts were intended to provide advance information and to afford some I basis for meaningful discussion and exchange of information. I PRÉFACE Un atelier sur l'évaluation des dangers sismiques en Ontario méridional a été tenu du 19 au 21 juin 1995. L'atelier avait comme but l'examen des données géologiques et sismologiques I disponibles concernant la possibilité de séismes en Ontario méridional et l'établissement d'un consensus concernant les méthodes à retenir pour la caractérisation de ces dangers. L'atelier était structuré en séances techniques visant à centrer les présentations et les discussions sur I les quatre thèmes techniques pertinents quant aux dangers de séisme en Ontario méridional : I 1) l'importance des observations géologiques et géophysiques pour la détermination des sources sismiques, 2) les méthodes et les approches qui pourraient permettre de déterminer les I sources sismiques d'après des interprétations intégrées de l'information géologique et sismologique, 3) les méthodes et les données qui pourraient servir à la caractérisation des I paramètres de l'activité sismique des sources sismiques, 4) les méthodes d'évaluation du danger de mouvement vibratoire du sol. I I I -iv- I Le présent document a été distribué aux participants avant la tenue de l'atelier. On y trouve une copie du programme, la liste des participants et les résumés de présentations • des conférenciers. Les résumés avaient pour but de fournir de l'information préalable et • de servir de point de départ à la discussion et aux échanges d'information lors de _ l'atelier. I I I I I I I I I I I I DISCLAIMER «

The Atomic Energy Control Board is not responsible for the accuracy of the statements made or the opinions expressed in this publication and neither the Board nor the authors assume liability with respect to any damage or loss incurred as a result of the use made of the information contained in this publication. I I I

| TABLE OF CONTENTS

| PREFACE iii

- PROGRAM 1

LIST OF PARTICIPANTS 5

I SESSION A: INTRODUCTION m Overview of the Workshop, J. Carl Stepp, Workshop Chairman Al

Purpose of the Workshop on Seismic Hazard Assessment in Southern Ontario, m J.G. Waddington, Director General, Directorate of Analysis and Assessment, AECB . A5

SESSION B: BACKGROUND m National Building Code Requirements, P.W. Basham, Geological Survey of Canada . Bl

Site and Regional Investigations According to CSA-N289.2, P.W. Basham, I Geological Survey of Canada B2 SESSION C: IMPORTANCE OF VARIOUS GEOLOGICAL AND • GEOPHYSICAL OBSERVATIONS FOR DETERMINING SEISMIC SOURCES IN SOUTHERN ONTARIO • Characteristics of the Niagara-Pickering and Geogian Bay linear zones and their implications for a large magnitude earthquake In the vicinity of the Darlington I and Pickering Nuclear Power Plants, near Toronto, J.L. Wallach, AECB Cl

Interpretation of aeromagnetic and gravity anomalies in the Precambrian shield of Southern Ontario, Walter R. Roest, Continental Geoscience Division, • Geological Survey of Canada Cll

I Extension of the St. Lawrence fault zone southwest of Cornwall into Lake Ontario, J.L. Wallach, AECB C23

I Deep Seismic Crustal Results in Southern Ontario, D. Forsyth (NOT AVAILABLE)

Implications for Seismic Risk in Southern Ontario Based on a Review of the • Region's Geology, R.M. Easton, D.K. Armstrong and P.C. Thurston, Ontario Geological Survey C29

Geophysical Signatures in the Bed of Lake Ontario, R.L. Thomas, Waterloo I Centre for Groundwater Research C37 I I I

-VI- B Neotectonic Features in the Quaternary Sediments of Lake Ontario: Interpretation of Geophysical Survey Data, C.F.M. Lewis, Geological Survey of Canada (Atlantic) . C39 I

Geophysical signatures beneath Lake Ontario in the vicinity of the Darlington _ Nuclear Generating Station; seismic hazard implications, J.L. Wallach, AECB . . . C49 •

Faults in the Metropolitan Toronto Area; Active vs Dormant Faults, A.A. Mohajer, _ University of Toronto / Seismican C61 |

On the interpretation of the Rouge River Faulting, Metropolitan Toronto — m What would an active fault with an unknown return period contribute to seismic | hazard assessment?, John Adams, National Earthquake Hazards Program, Geological Survey of Canada C75 •

Character and Reactivation History of the Clarendon-Linden Fault System: Evidence from Southwestern New York State, Robert D. Jacobi and John Fountain C77 •

SESSION D: METHODS AND APPROACHES FOR DETERMINING SEISMIC SOURCES BASED ON GEOLOGICAL AND SEISMILOGICAL INFORMATION •

Methods and Procedures for Determining Seismic Sources Based on Geological and Seismological Information, Kevin J. Coppersmith, Geomatrix Consultants . . . . Dl •

Seismicity of the Lake Ontario Region to 1994, Anne E. Stevens, Geological Survey of Canada D4 •

Seismogenesis and Structure in the Lake Erie-Lake Ontario Region of the U.S. from a Global Perspective, Leonardo Seeber and John Armbruster, Lamont-Doherty • Earth Observatory of Columbia University D5 *

The Southern Ontario Seismic Network - Application to Seismic Hazard and • Earthquake Engineering, R.F. Mereu and H.W. Asmis D21

Local Seismic Monitoring East of Toronto, A.A. Mohajer, University of I Toronto / Seismican D23

Seismic Source Zone Modelling in Eastern Canada for National Building Code g Applications, P.W. Basham, Geological Survey of Canada D30

Seismic Source Zone Interpretation in Southern Ontario, A.A. Mohajer, | University of Toronto / Seismican D31

The Tectonic Stress Field in Eastern Canada:- its use in seismic source interpretation, | John Adams, National Earthquake Hazards Program, Geological Survey of Canada D40 I I I

I -Vll-

• Pop-ups and offset boreholes as geological indicators of earhtquake-prone areas in intraplate eastern North America, J.L.Wallach, AECB D42

£ Hazards Associated with Seismically Active Linear Zones, Christine A. Powell, Department of Geology, University of North Carolina D53

| SESSION E: METHODS AND DATA FOR CHARACTERIZING THE SEISMICITY PARAMETERS OF SEISMIC SOURCES

| Stable continental earthquakes and seismic hazard in Eastern North America, Arch C. Johnston, Centre for Earthquake Research & Information (CERI), • University of Memphis El

Seismicity Parameters for Eastern Canadian Seismic Source Zones, P.W. Basham, • Geological Survey of Canada E7

SESSION F: METHODS TO ASSESS VIBRATORY GROUND MOTION • HAZARD AND APPLICATIONS

Uses of Probabilistic Seismic Hazard Analysis for Evaluating Nuclear Plant I Core-Damage Frequency, Robin K. McGuire, Risk Engineering, Inc Fl

Sources of Uncertainty and its Treatment in Seismic Hazard Analysis in Eastern Canada, John Adams, National Earthquake Hazards Program, • Geological Survey of Canada F9

• Ground Motion Relations in Eastern North America, Gail M. Atkinson Fll

— Deterministic Seismic Hazard Analysis, Kevin J. Coppersmith, Geomatrix ConsultantsF31

I I I I I I I I I I I I Program • AECB Workshop on Seismic Hazard Assessment in Southern Ontario

Ottawa Citadel Hotel | June 19-21, 1995 I I I I I I I I I I -2- AECB Workshop on Seismic Hazard Assessment I in Southern Ontario Ottawa Citadel Hotel June 19-21, 1995 I Monday, June 19, 1995 I Session A — Introduction 11:20 am Implications for seismic risk in Session Chair: J. C. Stepp Southern Ontario based on a review of the region's geology I 8:30 am Purpose of the workshop and AECB R.M. Easton and P.C. Thurston objectives J.G. Waddington 11:40 am Geophysical signatures in the bed of I Lake Ontario 8:40 am Overview of the workshop R.L. Thomas J.C. Stepp 12:00 pm Lunch I Session B — Background Session Chair: /. C. Stepp 1:30 pm Neotectonic features in the quaternary sediments of Lake 8:50 am National Building Code requirements Ontario: Interpretation of I P.W. Basham geophysical survey data in the vicinity at Darlington Nuclear 9:10 am Site and regional investigations Generating Station; seismic hazard I according to CSA-N289.2 implications P.W. Basham C.F.M. Lewis Session C — Importance of Various 1:50 pm Geophysical signatures under Lake I Geological and Geophysical Ontario Observations for Determining J.L. Wallach Seismic Sources in Southern I Ontario 2:10 pm Faults in the Metropolitan Toronto Session Chair: R. Price area; active vs dormant faults A. Mohajer 9:20 am Characteristics of the Niagara- I Pickering and Georgian Bay linear 2:30 pm Interpretation of the Rouge River zones and their implications for a Faulting, Metropolitan Toronto - large magnitude earthquake in the What would an active fault with a vicinity of the Darlington and known return period contribute to I Pickering Nuclear Power Plants near seismic hazara assessment? Toronto J. Adams J.L. Wallach I 2:50 pm Character and reactivation history of 10:00 am Coffee the Clarendon-Linden fault system: evidence from Southwestern New 10:20 am Interpretation of aeromagnetic and York State I gravity anomalies in the Precambrian R.D. Jacobi shield of Southern Ontario W. Roest 3:10 pm Coffee I 10:40 am Extension of the St-Lawrence Fault 3:30 pm Open for presentations by participants Zone southwest of Cornwall into Lake Ontario 3:50 pm General Discussion I J.L.Wallach 4:20 pm Summary and Recommendations for 11:00 am Deep seismic crustal results in research Southern Ontario I R. Price D. Forsyth I I -3-

Tuesday, June 20, 1995

Session D — Methods and 12:00 am Open presentations by participants Approaches for Determining Seismic Sources Based on 12:10 am General Discussion Geological and Seismological Information 12:30 pm Lunch Session Chair: K. Coppersmith 1:30 pm Summary and Recommendations for 8:30 am Methods and procedures for research determining seismic sources based on K. Coppersmith geological and seismological information K. Coppersmith Session E — Methods and Data for Characterizing the Seismicity 9:00 am Seismicity of Lake Ontario region Parameters of Seismic Sources A. Stevens Session Chair: G. Klimkiewicz 9:20 am Seismicity of Lake Ontario region 1:40 pm Stable continental earthquakes and L. Seeber seismic hazard in Eastern North America 9:40 am The Southern Ontario Seismic A. Johnston Network-Application to seismic hazard and earthquake engineering 2:10 pm Seismicity parameters for Eastern H. Asmis Canadian seismic source zones P. Basham 9:50 am Local seismic monitoring east of Toronto 2:30 pm Open for presentations by participants A. Mohajer 2:40 pm General Discussion 10:00 am Coffee 3:00 pm Coffee 10:20 am Seismic source zone modelling in Eastern Canada for National Building 3:20 pm Summary and Recommendations for Code Applications research P. Basham G. Klimkiewicz 10:40 am Seismic source zones interpretations Session F — Methods to Assess in Southern Ontario Vibratory Ground Motion Hazard A. Mohajer and Applications Session Chair: R. McGuire 11:00 am The tectonic stress field in Eastern Canada: - its use in seismic source 3:30 pm Uses of probabilistic seismic hazard interpretation assessment for evaluating nuclear J. Adams power plant core-damage frequency R. McGuire 11:20 am Pop-ups and offset boreholes as geological indicators of earthquake- 4:00 pm Sources of uncertainty and its prone'areas in intraplate Eastern treatment in seismic hazard analysis North America in Eastern Canada J.L. Wallach J. Adams 11:40 Hazards associated with seismically active linear zones C. Powell -4- I

Wednesday, June 21, 1995 §

Session F — Continued 8:30 am Ground motion relations for seismic I hazard analysis in Eastern Canada G. Atkinson I 8:50 am Deterministic Seismic Hazard Assessment K. Coppersmith I 9:10 am Potential vibratory ground motions — based on information presented at • this workshoDp ™ G. Klimkiewicz 9:30 am Open for presentations by participants B 9:40 am General Discussion 10:00 am Coffee I 10:20 am Summary and Recommendations for research R. McGuire I Session G — Workshop Summary and Recommendations 10:30 am Workshop Summary and I Recommendations J. C. Stepp I 11:30 am End of workshop I I I I I I I I -5-

List of Participants

1. Workshop Chairman J.C. Stepp Earthquake Hazard Solutions

2. Session Chairmen K. Coppersmith Geomatrix Consultants G. Klimkiewicz Weston Geophysical Corporation R.K. McGuire Risk Engineering Inc. R.A. Price Queen's University

3. Other Speakers J. Adams Geological Survey of Canada H. Asmis Ontario Hydro Nuclear G. Atkinson Consultant P.W. Basham Geological Survey of Canada R.M. Easton Ontario Geological Survey D. Forsyth Geological Survey of Canada R.D. Jacobi State University of New York at Buffalo A.C. Johnston Memphis State University C. F. Lewis Atlantic Geoscience Centre A. Mohajer Seismican Geophysical C. Powell University of North Carolina W. Roest Geological Survey of Canada L. Seeber Lamont-Doherty Geological Observatory A. Stevens Geological Survey of Canada R.L. Thomas University of Waterloo P.C. Thurston Ontario Geological Survey J.G. Waddington AECB J.L. Wallach AECB —6— I 4. Other Participants C. Alexander Ontario Hydro Nuclear I T.S. Aziz AECL CANDU I M.J. Berry Geological Survey of Canada J. Bowlby Neotectonic Associates I T. Brennand Geological Survey of Canada B.E. Broster University of New Brunswick I N.G. Brown Ontario Hydro Nuclear K. Burke University of New Brunswick I R. Charlwood ACRES M. Chénier Consulting Geologist I R. Clarke Ontario Hydro Nuclear J. Drysdale Geological Survey of Canada I R. Du Berger Université du Québec a Chicoutimi M.N. Garceau Hydro-Québec I A. Ghobarah McMaster University J. Gibbons Consulting Geologist I 5. Halchuk Geological Survey of Canada V. Hanemayer Siting Force Secretariat - Natural Resources Canada I Y.W. Isachsen New York State Geological Survey K. Jacob Lamont-Doherty Earth Observatory I S. Kumarapeli Concordia University I M. Lamontagne Geological Survey of Canada R. Mereu University of Western Ontario I D.J. Misener Paterson, Grant & Watson Ltd. K.-M. Nigge York University I J.S. Scott Siting Force Secretariat - Natural Resources Canada A. Seguin Siting Force Secretariat - Natural Resources Canada I D. Sharpe Geological Survey of Canada B. Todd Geological Survey of Canada I R. Wetmiller Geological Survey of Canada I I -7-

P. Wiebe - Ontario Hydro Nuclear D.J. Wilson - New Brunswick Power

5. AECB Participants G.J.K. Asmis A.J. Bishop D. Bottomley A.J.G. Faya P. Flavelle J.D. Harvie A.H. Ling C.B. Parsons J.K. Pereira C. Pilette R. Stenson H. Stocker J.G. Waddington (also on speakers list) J.L. Wallach (also on speakers list)

June 19/95 I i I I I I I I Session A I • Introduction I I I I I I I I I OVERVIEW OF THE WORKSHOP

J. Carl Stepp Workshop Chairman I I An assessment of the seismic hazard at a site relies on subjective evaluations of _ earthquake sources, earthquake sizes and recurrence rates (if probabilistic assessments are • desired) associated with each seismic source, and ground motion attenuation. Together, these three subjective interpretations constitute a basic requirement for every seismic hazard « assessment regardless of whether deterministic or probabilistic approaches are used to model g the hazard at any particular site. Thus the geological, geophysical and seismological data on which these evaluations depend are of fundamental importance. The more sparse the data • and the weaker our understanding of their significance as indicators of earthquake activity the P more uncertain our interpretations become. Thus, interpretations are generally less certain in continental interior regions, which are typically regions of low rates of earthquakes and m sparse and poorly developed geologic indicators of activity, than in intraplate regions, which • typically have relatively high rates of earthquakes and reasonably well developed geologic indicators of activity. Our general experience is that it is desirable and most productive to flj use the broadest available data (geological, geophysical and seismological) in an integrated B evaluation of seismic sources. Multiple approaches lend confidence to evaluations maximum earthquakes that can reasonably be expected associated with a seismic source and multiple fi ground motion attenuation relationships can result in more confident estimates or ground • motion and its variability. A good understanding of the importance of various data and how they can be used most productively to perform these essential evaluations, may indeed be the • most important step in any seismic hazard evaluation. ™

The technical sessions of this workshop have been structured to address each of the four I elements of a seismic hazard assessment mentioned above, focusing on the region of southern Ontario. Session C addresses the first and fundamental issue of the importance of various g. geological and geophysical observations for evaluating seismic sources in southern Ontario. 0 Speakers should discuss the importance of recently acquired data and interpretations for interpreting seismic sources and identify research that could improve confidence in seismic m source determinations. Session D continues discussion of basic data for evaluating seismic g sources and focuses on methods and approaches that integrate interpretations of broad geological, geophysical and seismological data to evaluate seismic sources. Session E is Q devoted to the evaluation of methods and data for determining rates of earthquakes and • estimating maximum earthquakes associated with seismic sources. For this and other sessions the emphasis is specifically on stable continental interiors and the appropriate use of B world-wide analog data to supplement regional data as a basis to improve confidence in these 9 two very important parameters. Session F addresses methods to assess ground motion hazard at a site using integrated source and seismicity parameter interpretations and their H applications. In Session G we want to achieve a workshop summary and recommendations. • The summary should identify areas where our knowledge is considered to be mature and areas where additional research is needed. Recommendations should identify specific fl research that has a high potential to reduce the uncertainty on seismic hazard assessment in ™ southern Ontario. I 1 I -A3-

In each technical session time has been made available for presentations by participants in the workshop who are not identified speakers. It is intended that these presentations address the topic of the session. They may be spontaneous or prepared. In any case, the presentations will be made part of the general discussions and may contribute to the recommendations for future research to reduce the uncertainty on seismic hazard assessment.

I want to welcome you to this very important workshop. I look forward to working with you over the next three days.

95-1122

NEXT PAQE(S) left BLANK -A5-

PURPOSE OF THE WORKSHOP

ON SEISMIC HAZARD ASSESSMENT

IN SOUTHERN ONTARIO

J.G. Waddington Director General Directorate of Analysis and Assessment Atomic Energy Control Board -A6- • To ensure that Canadian nuclear reactors do not pose an undue risk to the Canadian public, the AECB needs to be assured that the probability of an event which could cause significant damage M to the stations, and hence the potential of release of radioactive material to the environment, is very low indeed. Typically, we wish to satisfy ourselves that the probability of significant core B damage is less than 10 /per year at any station. I Knowledge of seismicity of Eastern Ontario, as elsewhere in the world, has developed substantially in the last 20 years. Many new seismological, geophysical and geological I observations have been made since the Pickering and Darlington nuclear power stations first received construction approval. The AECB needs to know whether or not this development of • knowledge and understanding could affect the licensing basis of these stations. It is clear that geological faults exist in many places and may exist near the stations. If they do, how likely are S they to give rise to seismic activity, and what sort of movement might be expected with what sort of likelihood? Would the seismic capability required of the station be affected, and if so, to what g extent? 1 It is also clear that over the past few years, differences of scientific opinion have developed in the world of the geologist and seismologist on how this new data should be interpreted. This is g[ healthy! Good science demands questioning minds, and open and vigorous debate. The purpose of this conference is to provide a forum for that open and vigorous debate. But, for the AECB and its licensees, that debate has to lead to real decisions: whether the current 9 capability of the stations to survive earthquakes is adequate or not; whether further efforts to — improve that capability are warranted or not, given the very low level of risk to the public that I we and the nuclear industry are seeking to achieve; or whether further research is needed before ^ such decisions can be made, and if so, of what nature. • I We believe we have assembled at this forum a highly talented and experienced body of respected • scientists in this field to help us make these decisions, and we can all be sure that the debate will M indeed be vigorous, and open, and represent the best that science can provide today. •• With your Chairman, Dr. Carl Stepp, these sessions have been designed to examine the issues, • and to encourage debate on them. ft

5156.jgw I Session B

Background 1 I t ,' AECB Workshop on Seismic Hazard Assessment in Southern Ontario

I Session B - Background

P.W. Basham I Geological Survey of Canada I ABSTRACT I National Building Code Requirements The procedures for earthquake-resistant design and construction of buildings in Canada is covered by the National Building Code of Canada (NBCC) which is prepared I by the Canadian Commission on Building and Fire Codes (CCBFC) of the National Research Council. The NBCC is a model national code providing minimum standards which can be adopted or modified, as appropriate, by the provinces who have jurisdiction I over building standards. The relative levels of seismic hazard across Canada are depicted in the NBCC by seismic zoning maps which are developed by the Canadian National Committee for Earthquake Engineering (CANCEE), which advises the CCBFC I on seismic issues. CANCEE bases its seismic zoning maps on seismic hazard estimates produced by the Geological Survey of Canada (GSC). There have been three generations of seismic zoning maps in the NBCC, 1953, 1970 and 1985, as the I understanding of seismic hazards and technologies in earthquake engineering developed over time. The GSC and CANCEE are currently working on the fourth generation of zoning maps, which will be issued for trial use by the engineering community later this I year, and modified as necessary in about 1997 for official adoption in the year 2000 NBCC. I The current (1985) seismic zoning maps of Canada depict ground shaking parameters with a 90% probability of not being exceeded in 50 years. This probability is common to many seismic zoning maps around the world. This probability, which can I also be viewed as a return period of 475 years, has come to be accepted as representing an appropriate trade off between acceptable risk and the economic penalty of providing greater protection. The NBCC philosophy in designing and constructing buildings i against earthquake loads is to prevent loss of life due to building collapse, although a building may be damaged beyond repair if it experiences earthquake loading at or above the design levels. The NBCC has stricter requirements for important buildings, such as I hospitals, and a building owner or jurisdiction can, of course, choose stricter requirements to prevent excessive damage. I Although NBCC seismic zoning maps are based on the best available information on earthquakes and their effects at the time of their adoption, having to depict seismic hazards over an area the size of Canada means that in the derivation of the maps no one I location receives a very thorough assessment of the hazard from the earthquakes in its region. It is considered adequate for building code purposes to depict seismic hazards as a simple zonation of ground motion parameters, the current maps having seven seismic i zones, from 0 to 6. I I -B2- I

Site and Regional Investigations According to CSA-N289.2 I Structures and facilities that are considered to be more hazardous to man or the g environment than common buildings, should they fail due to earthquake loading, can be designed and constructed to more stringent standards. In Canada, Canadian Standards » Association (CSA) standards have been developed for offshore oil and gas production S facilities (CSA-S471), liquefied natural gas storage facilities (CSA-Z276), and seismic qualification of CANDU nuclear power plants (CSA-N289). Canada also applies « internationally-developed standards to structures such as high dams, and agencies having fi jurisdiction will apply special standards to potentially hazardous facilities such as virus laboratories. Seismic hazard assessment for these types of special facilities will differ ^ from that done for the NBCC in three important ways: (1) the assessment will be B generally more conservative and produce hazard estimates with a lower probability of exceedence; (ii) the hazard assessment will focus on ground shaking information in M frequency ranges of direct relevance to the structure or facility and its components; and 8 (iii) the hazard assessment can be much more robust by bringing to bear all available * information on earthquake potential and earthquake effects as they may influence one _ geographical location, the site of the structure or facility. H CSA-N289 was developed by the CSA Technical Committee on Seismic Design which began its work in 1975. The standard is in five parts as follows, with the • publication date indicated: ™

N289.1 - General Requirements (1980) B N289.2 - Ground Motion Determination (1981) • N289.3 - Design Procedures (1981) N289.4 - Testing Procedures (1986) B N289.5 - Seismic Instrumentation Requirements (1991) •

CSA-N289.2 is the standard of most relevance to this workshop. It is the standard that fl describes the seismic hazard investigations, the results of which provide input to the • engineering design and testing described in N289.3 and N289.4.

N289.2 is not a "cookbook" because it does not proscribe exactly how the seismic B hazard investigation should be carried out. Rather, it provides guidelines for performing a state-of-the-art seismic hazard investigation that is deemed appropriate to a CANDU ft plant. The standard does not specify the probability of exceedence at which the hazard B estimate should be made. This is considered to be a matter to be decided as part of the licencing negotiations between the proponent and AECB. The body of the standard fl provides general guidelines on site and regional seismological and geological B investigations. A non-mandatory appendix discusses the manner in which the seismological and geological information may be developed into a description of the A seismotectonics of the region for purposes of selecting the "design basis seismic ground B motion" which the plant will be designed to withstand. 1 I I Session C

Importance of Various Geological and Geophysical Observations for Determining Seismic Sources in Southern Ontario Characteristics of the Niagara-Pickering and Georgian Bay linear zones and their implications for a large magnitude earthquake In the vicinity of the Darlington and Pickering Nuclear Power Plants, near Toronto

J.L. Wallach

To date, in eastern Canada, rigorous geological and geophysical investigations have played no role in assessing seismic hazard relevant to nuclear power plants. The Design Basis Seismic Ground Motions (DBSGM) for the nuclear generating stations at Darling- ton and Pickering were determined solely through the assessment of previous earthquakes with neither seismic monitoring nor any geological or geophysical input (Basham, 1975). Since then evidence has accumulated which suggests that conditions suitable for a moderate to large earthquake exist near both the Darlington and Pickering Nuclear Generating Stations which house 4x850 MWe nuclear reactors and 8x500 MWe nuclear reactors respectively. The impact of a major seismic event there could be serious since both are rather close to a population of 5-6 million. This paper describes some of the major features which suggest that the region encompassing the nuclear power plants may be subject to a greater seismic hazard than previously suspected.

Several linear aeromagnetic anomalies occur in the area of western Lake Ontario. The most prominent, named the Niagara-Pickering Magnetic Lineament (NPML) (Wallach, 1990), crosses the lake in a north-northc?sterly direction (015°-020°) and extends from at least the Niagara Peninsula, through Pickering, Ontario, then northward. A broad, north- northeast trending linear Bouguer gravity anomaly coincides with the NPML across most of the width of Lake Ontario and terminates against a similar, north-northwest trending Bouguer gravity anomaly just south of Pickering. Collectively the gravity and magnetic anomalies were referred to as the Niagara-Pickering Linear Zone (NPLZ) (Wallach and Mohajer, 1990).

The NPLZ appears to be the southern continuation of the the western limit of the Central Metasedimentary Belt (Forsyth, 1981; Forsyth and others, in press), which is known as the Central Metasedimentary Belt Boundary Zone (CMBBZ). Forsyth and others (in press), stated that the CMBBZ "appears to truncate (and offset ?) more easterly anomaly trends". Hanmer (1988) noted that the CMBBZ is unequivocally a reverse, ductile fault zone which formed approximately 1,060 ma ago. Physiographically the CMBBZ, and an area which extends at least 30 km to the east, are marked by a series of linear lake- and swamp-filled depressions, which trend about 000°-035°. This suggests that the CMBBZ, and the area to the east, constitute an unhealed zone of weakness caused by regional-scale, brittle failure. Mesoscopic-scale brittle structures were examined.

Mesoscopic-scale brittle faults, parallel and subparallel to the general trend of the CMBBZ, were found in Grenvillian rocks in the area from Norland to Minden, and from I

Norland to the Burleigh Falls-Buckhorn area, about 30 km to the east. They are expressed • as slickensided fractures, breccia zones and fractures across which there are recognizeable offsets. The slickensided surfaces indicate either dip-slip or strike-slip movement, though fi the former did not enable determining whether the movement was reverse or normal. The same applies to many horizontal to subhorizontal slickensided surfaces as well, but where the sense of displacement could be discerned, it was always dextral (Table 1). 9

A set of fractures in granite at Norland displays an average orientation of 030°/27°SE. _ Members of that set formed along pre-existing foliation planes, some of which are B polished and show dip-slip slickensides plunging in the directions 127°-135°. A parallel- striking, oppositely-dipping set of fractures (215°-35°) was recognized in a migmatite M north of Norland. Several members of this set also feature dip-slip slickensides with the B plunge in the direction 315°. Besides the gently dipping fractures, a few epidote-coated, nearly vertical, NNE-trending fractures were also identified, at least one of which displays flj sub-horizontal slickensides. Breccia zones are, in some instances, marked by Riedel shears, B suggesting a reverse sense of displacement, but east of Buckhorn, a normal fault cuts and displaces the Grenville-Shadow Lake (upper Middle Ordovician) contact (Table 1). That jK contact is subhorizontal and is, itself, a shear zone in which the Shadow Lake was • displaced to the southeast, relative to the underlying basement rocks, an apparent consequence of lateral compression. Thus, there is unequivocal evidence that the CMBBZ, fl and the at-least-30-km-wide area to the east, have been subjected to at least one period • each of brittle reverse, normal and strike-slip faulting. I Besides the mesosopic faults, there is also evidence of macroscopic-scale faulting along • major lineaments within the CMBBZ, and the 30-km-wide area to the east. Some are block faults, which displace both Precambrian and upper Middle Ordovician rocks B (Sanford, 1993). A right lateral separation of the Paleozoic bedrock surface, marked by Pontypool Ridge, is shown by Eyles et al (1993) to occur in the area of Lake Scugog, ^ along the NPLZ. Geophysical work carried out in Lake Ontario by McQuest Marine • (1995), in the vicinity of the NPLZ offshore of Pickering, revealed the presence of a fault which cuts the bedrock and displaces unconsolidated sediments estimated to be younger « than 11,400 years. 0

Except for the normal fault offset of the sheared Grenvillian-Shadow Lake unconformable m contact it has not yet been possible to establish the age relationship among the different B episodes of faulting. Nonetheless, it is important to note that the right lateral strike-slip indicators, on NNE-trending, steeply dipping fracture surfaces, and the displacement at M Pontypool Ridge, are kinematically compatible with the current ambient stress field in m eastern North America. I The characteristics of the CMBBZ, as originally defined, and the area to the east suggest • that the CMBBZ is actually a zone at least 30 km wide with the NPLZ serving as its western limit. Thus, the term NPLZ, is retained and, henceforth, refers to the western limit B of the CMBBZ. The newly defined CMBBZ is comprised of brittle mesoscopic and ~ macroscopic faults, indicative of repeated faulting. This implies that it is a brittle, unhealed I -C3- structural zone which could be capable of concentrating tectonically-derived, stored strain energy. Epicenters of several earthquakes on the Niagara Peninsula are spatially related to the NPLZ, including the 1853, MM intensity V earthquake and, possibly, the 1873 MM intensity VI event (Smith 1962). This suggests that the CMBBZ may be seismically active.

From the aeromagnetic map (Figure 1), it is seen that, in passing to the south, the NPLZ curves to the southwest, then southward, into Ohio. There, another prominent magnetic feature is located, the Akron Magnetic Boundary (AMB), with which the 1986, mb=4.9 Leroy earthquake was associated (Leblanc, 1989). The magnetic attributes of both the AMB and the NPLZ, combined with the curvature of one towards the other, suggests that they represent the same major structural boundary. The curvature seen on the magnetic map (Figure 1) is also very evident on the gravity map (Figure 2).

A prominent set of north northwest-trending (a=33O°) aeromagnetic lineaments defines a zone on the order of 50 km wide, which passes from just north of the CMBBZ toward the northeast corner of Georgian Bay, a distance of about 200 km. Associated with these lineaments is the straight line east coast of Georgian Bay, and a series of discontinuous linear Bouguer gravity anomalies. Near Attica, New York, one of these anomalies intersects a second gravity anomaly, which is oriented north-south and signifies the Clarendon-Linden Fault. The intersection marks the location of the 1929, mb=5.2 Attica earthquake. The differently expressed, parallel lineaments (aeromagnetic, gravimetric and physiographic) are practically superimposed on one another suggesting the presence of a major structural zone, named the Georgian Bay Linear Zone (GBLZ) by Wallach (1990).

Sanford, Thompson and McFall (1985) identified a physiographic lineament and interpreted it as a major block boundary tectonically reactivated at least during the earliest Paleozoic. This lineament coincides with the east coast of Georgian Bay and it may also continue into New York State. Both Precambrian and Paleozoic rocks in the GBLZ were examined, in a reconnaissance fashion, to determine whether or not there are brittle structures which parallel the GBLZ. Fracture trends were measured in a horizontally dipping gneiss, just west-northwest of Port Severn and in the Gull River Limestone at Fesserton, to the southeast; the principal fracture sets in both rock types are parallel to the GBLZ. At least one linear feature, several hundred meters long and present in glaciolacustrine sediments of Georgian Bay, was recognized by Blasco (personal communication). It is occurs within, and is parallel to, the GBLZ, and was identified as a pock mark by Blasco, who interpreted it as resulting from gaseous or liquid emanations. Thomas (personal communication) has suggested that active faults and fractures are apparent prerequisites for this phenomenon.

Mesoscopic normal faults occur in the vicinity of, and are subparallel to, the GBLZ. One, identified in a limestone quarry east of Lake Simcoe, strikes 308°, dips 75° to the northeast and shows a normal displacement of 5 cm. A second is located in the Georgian Bay Formation in Toronto, trends 145°, dips to the southwest at 75° and shows a minimum throw of 5.5 m (McFall, personal communication). The ages of the two normal faults are unknown. Normal faults, which pass upwards from the Whitby Shale into -C4- I interglacial sediments were also recognized in the Rouge River, in eastern Metropolitan I Toronto (Mohajer et al, 1992). Several orientations are represented there, among which g are faults which are parallel and sub-parallel to the GBLZ. Lastly, the Laurentian Channel, g a prominent NNW-oriented depression in the bedrock coincides with the geophysically expressed GBLZ and extends from south of Georgian Bay to Toronto (Eyles et al, 1993). m

Epicenters of at least eight small-magnitude earthquakes define a line which extends north- northwestward from western New York State to near the northeastern corner of Georgian • Bay. In all there are at least twenty earthquakes, including the 1857 Lockport (N.Y.) (est. 9 1 M=5.0) and the 1929 Attica (mb=5.2) earthquakes . The Lockport and Attica earthquakes also lie within the gravimetric zone of the GBLZ. Thus, rather than elucidating a simple fl line, the twenty earthquakes probably outline a zone, as do the gravity and magnetic ™ components of the GBLZ.

The zone of earthquakes trends north-northwest, passes through the sites at Darlington ™ and Pickering and continues along the east side of Georgian Bay. It is parallel to, and virtually coincident with, the spatially related, linear physiographic, aeromagnetic and • gravimetric features which are different expressions of the Georgian Bay Linear Zone. Pre-instrumentally recorded earthquakes may be in error by as much as 50 km, and even — the locations of the instrumental^' recorded events may also be imprecise. The definition I of the earthquake zone, including its width and the number of earthquakes contained therein, is also subject to interpretation. Nonetheless, the existing data show an g assemblage of earthquakes outlining a linear belt which parallels, and is proximal to, all of 9 the different expressions of the GBLZ. This could be mere coincidence, though it is difficult to ignore the possibility that it is not. n

Seismicity has been documented in western Lake Ontario and adjacent areas. Among the largest are the Attica (mb=5.2) and Lockport (M=5.0) earthquakes, which occurred within Êk 100 km of the Darlington and Pickering sites, and the M=6.25, Timiskaming earthquake, m located about 370 km northeast of the nuclear generating stations. Thus, from pre-existing data, it is evident that the area surrounding western Lake Ontario has already been B subjected to moderate to large earthquakes. •

Despite recent advances made in the study of seismotectonics (e.g. Mohajer, 1987; • Talwani, 1988), relatively little is still known about the moderate to large intraplate ™ earthquakes of eastern North America. This is principally due to their small number and rather disperse nature. One characteristic, however, that may provide a significant clue to • predicting the locations of future, large earthquakes is the recognition that many moderate to large seismic events have occurred at the intersection of major lineaments (Talwani, .— 1988). Several in eastern North America fall into this category and include the 1929 Attica • (mb=5.2), and the 1732 Montreal (est. M=5.6-6.0), the 1935 Timiskaming (M=6.25) and the 1944 Cornwall-Massena (M=5.9) earthquakes. This factor is particularly important in m assessing the seismic hazard in the region of the Darlington and Pickering nuclear power g|

'-Seven of them occur in a swarm in the vicinity of the 1929 Attica earthquake. V 1 -C5- plants because they are located, in effect, in the intersectiona] aiea of the CMBBZ and the GBLZ.

Relatively little is known about the neotectonics of the western Lake Ontario area although information is continuing to accumulate which suggests that the area features geologically young, tectonic structures. Moreover it is clear that there are brittle structures in both Precambrian and Paleozoic rocks near Georgian Bay, which parallel the trend of the GBLZ; similarly, there are brittle structures in the CMBBZ which are parallel or subparallel to the trend of this feature. Brittle structures within, and parallel to, the GBLZ and the CMBBZ respectively indicate that both are regional-scale, unhealed brittle structures.

Earthquake epicenters, distributed in a zone which is parallel and proximal to the various signatures of the GBLZ, imply that the GBLZ may be an active fault zone. The same implication, regarding a potentially active fault, may also be valid for the CMBBZ, at least on the Niagara Peninsula.

The western Lake Ontario region is technically active and may experience a moderate to large earthquake. If the CMBBZ and the GBLZ are seismically active faults then their intersection would be a likely location. The postulated extension of the St. Lawrence fault zone along Lake Ontario (Adams and Basham, 1989; Thomas, et al, 1993) adds to the probability of the aforementioned location being the site of a major earthquake, because this proposed structure appears to form a three-way intersection with the other two. If the St. Lawrence fault zone extends through Lake Ontario, as appears to be the case, an earthquake of M=7.0 must also be considered as a credible event. Thus, the seismotectonic symptoms of the region suggest that the potential for a large-magnitude earthquake exists near the nuclear power plants at Darlington and Pickering. -C6- I REFERENCES I Adams, J. and Basham, P., 1989. The seismicity annd seismotectonics of Canada east of the ft Cordillera. Geoscience Canada, 16, 3-16.

Basham, P.W., 1975. Design basis seismic ground motion for Darlington Nuclear fl Generating Station A. Seismological Service of Canada, Internal Report 75-16, • Division of Seismology and Geothermal Studies, Earth Physics Branch, Department of Energy, Mines and Resources, 34 p. plus tables and figures. M

Eyles, N., Boyce, J. and Mohajer, A.A., 1993. The bedrock surface of the western Lake Ontario region: Evidence of reactivated basement structures? In Wallach, J.L. and I Heginbottom, J.A., (eds.) Neotectonics Of The Great Lakes Area, Géographie physique et Quaternaire, 47, 269-283. —

Forsyth, D.A., 1981. Characteristics of the western Quebec seismic zone. Canadian Journal of Earth Sciences, 18, 103-119. m

Forsyth, D.A., Thomas, M.D., Real, D., Abinett, D., Broome, J., and Halpenny, J., in press. Geophysical investigations of the Central Metasedimentary Belt, Grenville • Province: Quebec to northern New York State. Proceedings of the 7th g| International Conference on Basement Tectonics.

Freeman, E.B., ed. 1978 Geological highway map, southern Ontario. Ontario Geological B Survey, Map 2418.

Hanmer, S., 1988. Ductile thrusting at mid-crustal level, southwestern Grenville Province. ™ Canadian Journal of Earth Sciences, 25, 1049-1059.

Leblanc, G., 1989. Minutes of the MAGNEC Meeting, Chicoutimi, Québec. 1m

McQuest Marine Sciences Ltd., 1995. The geophysical survey of Lake Ontario in 1993. B Atomic Energy Control Board INFO-0539, 17 pp plus figures.

Mohajer, A. A., 1987. Reappraisal of the seismotectonics of southern Ontario Task 1: • Relocation of earthquakes and seismicity pattern. Atomic Energy Control Board INFO-0293 (also MAGNEC Contribution 87-01), 58 pp plus Appendix. ta

Mohajer, A.A., Eyles, N. and Rogojina, C, 1992. Neotectonic faulting in Metropolitan Toronto: Implications for earthquake hazard assessment in the Lake Ontario M region. Geology, 20, 1003-1006. I I 1 -C7-

Sanford, B.V. 1993. Stratigraphie and structural framework of upper Middle Ordovician rocks in the Head Lake-Burleigh Falls area of south-central Ontario. In Wallach, J.L. and Heginbottom, J.A., (eds.) Neotectonics Of The Great Lakes Area, Géographie physique et Quaternaire, 47, 253-268.

Sanford, B.V., Thompson, F.J., and McFall, G.H., 1985. Plate tectonics-a possible controlling mechanism in the development of hydrocarbon traps in southwestern Ontario. Bulletin of Canadian Petroleum Geology, 33, 52-71.

Smith, W.E.T., 1962. Earthquakes of eastern Canada and adjacent areas 1534-1927. XXVI, 269-301.

Talwani, P., 1988. The intersection model for intraplate earthquakes. Seismological Re- search Letters, 59, 305-310.

Thomas, R.L., Wallach, J.L., McMillan, R.K., Bowlby, J.R. Frape, S., Keyes, D and Mohajer, A. A. 1993. Recent deformation in the bottom sediments of western and southeastern Lake Ontario and its association with major structures and seismicity. In Wallach, J.L. and Heginbottom, J.A., (eds.) Neotectonics Of The Great Lakes Area, Géographie physique et Quaternaire, 47, 325-335.

Wallach, J.L. Newly discovered geological features and their potential impact on Darlington and Pickering. AECB INFO-0342, 20 pp. plus figures.

Wallach, J.L. and Mohajer, A.A., 1990. Integrated geoscientific data relevant to assessing seismic hazard in the vicinity of the Darlington and Pickering Nuclear Power Plants, in Prediction and Performance in Géotechnique, Canadian Geotechnical Conference, Tome II, October 10, 11 and 12, 1990. Québec, Québec, pp. 679- 686. -C8- I

TABLE 1 1 Examples of Mesoscopic-Scale Faults Within the CMBBZ I Location Lithology Orientations Kinematics of Fractures I Minden Interlayered hornblende-biotite 215°/55°NW Right-lateral gneiss &garnet-biotite gneiss 210°/50°NW Right-lateral 202760°NW Right-lateral i Norland. Marble 035744°SE Reverse I

Miners Bay Migmatite 170790°W Right-lateral I 000790° Right-lateral

N of Buckhorn Sheared granitic to granodior- 037750°SE Right-lateral I itic gneiss

E of Buckhorn Shadow Lake carbonate 030770°SE Normal I 1 1 I I I i I I I I -C9-

75 km

Figure 1. Total-field aeromagnetic map showing the Niagara-Pickering Linear Zone (NPLZ), and the zone to the west, curving to meet the Akron Magnetic Boundary (AMB). The lower arrow points to the boundary in Lake Erie, just to the NNE of the Ohio-Pennsylvania state line; the upper arrow points to the NPLZ, northeast of Toronto, which is indicated by the black cirrated rectangle. (Map from the Geological Survey of Canada) -C10- I I 1 I I I I I I I I I I 75 km I I Figure 2 Bouguer gravity map showing the same curvature of the NPLZ, and the zone to the west, as seen in Figure 1. The upper arrow points to the NPLZ I in Lake Ontario; the lower arrow notes the curvature of the NPLZ/AMB, in Lake Erie. The curvature is most evident in the medium to light gray interval immediately to the west of the tip of the arrow. (Map from the I Geological Survey of Canada) I I -Cll-

Interpretation of aeromagnetic and gravity anomalies in the Precambrian shield of Southern Ontario by Walter R. Roest Continental Geoscience Division Geological Survey of Canada, Ottawa Ontario

Introduction

Magnetic and gravity anomalies over Southern Ontario provide a means to extrapolate what is known from the exposed shield in the north, to the region covered with Paleozoic sediments and Lake Ontario further south (Figure 1). As a geophysicist, I will restrict myself here to a preliminary analysis of the potential field anomalies, with emphasis on the orientation of potential field anomalies, and the spatial distribution of magnetic and gravity anomaly sources.

The regional magnetic map of Southern Ontario and the northern U.S., based on a 2 kilometre regional magnetic grid, outlines the general NNE-SSW trending structures of the study area. Magnetic anomalies are the result of variations in the magnetic properties of rocks, and are generally most sensitive to the near surface geology. Magnetic contrasts can result from the juxtaposition of different rock types, or for example from changes in metamorphic grade, reflecting the thermal and tectonic history of the underlying rocks, and from variations in induced versus rémanent magnetization components. Given the many different sources of magnetic anomalies, it is hard to directly relate magnetic lineations to structural elements. Combination with other data can solve some of the ambiguities.

Gravity anomalies, on the other hand, reflect density variations between different rock units, as well as, to a certain extent, the general state of crustal stability. This is because the Earth is generally in isostatic equilibrium, with excess mass near the surface compensated by a deficiency deeper down and vice-versa, and/or mass anomalies that are supported by the strength of the lithosphère. In that sense, gravity anomalies provide information on the thickness and behaviour of the elastic lithosphère and are important to study the processes that play a role in crustal dynamics. The regional gravity anomalies over southern Ontario and the northern US show trends similar to the magnetic map: a generally NNE-SSW oriented series of lineations with a curvilinear appearance.

Data

Aeromagnetic Data Aeromagnetic data over Ontario have been acquired over a large number of years, for a large part under agreements between the federal and provincial governments. In the early 90's, all available magnetic data in the province were combined and levelled to a common datum to produce a high resolution digital data set, with a grid spacing of 200 m. This new grid contains a wealth of -C12- I

information that was not visible in the older digital data sets. The aeromagnetic flight lines M coverage over part of Southern Ontario is shown in Figure 2. The different survey blocks are easily recognized by their flight line orientation and spacing. This information is essential to g ensure that later interpretation is not biased by flight line artifacts and/or survey boundary effects. £ Some of the data sets shown in Figure 2 were acquired with analogue chart recorders and later digitized. The most recent surveys are entirely digital and result in very high quality data. All A observations were interpolated onto a regular grid with a 200 m grid interval. A shaded relief I representation of this grid is shown in Figure 3, displaying a large dynamic range and sharper detail than previous digital data sets. The illumination is from the NW, emphasizing the general tt structural trends which are roughly NE. Important features in the magnetic anomaly data include •» from west to east (see Figure 1 for locations): the generally NW-SE trending lineations of the Central Gneiss Belt; the western boundary zone of the Central Meta-sedimentary Belt; the I generally NE-SW trending anomalies of the Elzivir and Frontenac Terranes; and the Elzivir- * Frontenac boundary zone. The Niagara-Pickering Lineament (Wallach, 1990) appears to consist — of a series of more or less en-echelon anomalies, in a zone that widens to the north and south of gj the Lake Ontario shoreline near Pickering. I A complication in the interpretation of magnetic anomalies is that they are not necessarily located above their sources. This is a result of the fact that the magnetic field is a vector, and unlike B gravity, the direction of the magnetic vector can deviate significantly from the vertical. One m technique we developed (Roest et al, 1992) to address this problem is the 3-dimensional analytic signal: a method that places magnetic anomalies directly above their sources, without assumptions % on the direction of source body magnetization and the local direction of the Earth's magnetic • field. Another interpretation method (Blakely aridSimpson, 1986) is the pseudo gravity transformation. In this case, one calculates the gravity anomaly that would result from the sources • of the magnetic anomalies, if they were assigned densities proportional to their actual magnetization. This method produces very stable results but one needs to know the direction of _ magnetization (whether it is induced magnetization or some form of rémanent magnetization) and g also the local direction of the Earth's magnetic field. The latter can be easily obtained from reference field models, but the contribution of rémanent magnetization remains uncertain.

As part of the pseudo-gravity technique, one calculates the horizontal pseudo-gravity gradient. The whole procedure is very stable and produces continuous features that can be further analysed in terms of structure and depth-to-magnetic-source. Maxima in the pseudo gravity gradient indicate the most probable locations of magnetic contacts. Figure 4 shows the location of linear maxima in the pseudo gravity gradients as adjoining circles with variable sizes: The deeper the magnetic sources, the larger the circles. The depth to magnetic source is remarkably stable along continuous features in this region (for example along the Niagara-Pickering Lineament a gradual deepening is seen towards the SW, consistent with a thickening of the Paleozoic cover). The stability of the solutions partly reflects close flight line spacing and a relatively good data quality. The calculated depth values can be used to produce a "depth to magnetic source" map. In this case, the magnetic sources are close to, or at, the surface in the area of the exposed shield, and gradually deepen when moving toward the southwest, under the Paleozoic cover. Comparison with the wide angle reflection lines in Lake Ontario (Forsyth et al, 1994; Forsyth, this meeting), -CIS- shows that the magnetic sources are generally located within the top two kilometers of the crust beneath the Paleozoic cover. Given the shallow easterly dipping reflectors observed in the reflection profiles, the relevance of magnetic lineations mapped at the surface to the location of earthquakes depends on the depth at which these earthquakes occur. Following the dipping boundaries in the reflection data, features that are seen at a depth of 10-20 km are located several tens of kilometres to the east of the mapped anomalies.

Gravity Data Figure 5 shows the distribution of gravity stations over part of Southern Ontario and the northern US. The data density is highly variable, which will affect the ability to produce a comprehensive interpretation. The Bouguer gravity anomaly map in general terms has relatively positive anomalies reflecting mafic and metamorphic rocks and relatively negative anomalies over granites and deep basins. The general trend of the anomalies mimics those seen in the magnetic data, but gravity generally provides a more integrated signature that averages over a larger part of the crust, both vertically and horizontally, consequently showing less detail than magnetics (Figure 6). By calculating the horizontal gravity gradient, the location of density contrasts can be better defined (Figure 7), and the recognition of important features is less dependent on the color scheme chosen, as is the case for a standard Bouguer map. Generally, the gradient map shows that features are striking NNE-SSW in the Central Meta-sedimentary Belt and more NW-SE in the Central Gneiss Belt. A preliminary lineation and depth-to-source analysis (not shown) indicates that gravity sources are located deeper than magnetic sources, which is not surprising since the gravity anomalies reflect an average density contrast in the upper portion of the crust.

Georgian Bay Linear Zone, a discussion

After this general analysis of the regional magnetic and gravity anomalies in Southern Ontario, some discussion is needed on the detection of linear zones in potential field data. In particular very long "ruler-straight" (Wallach, 1990) linear zones are suspect since geology generally produces curvi-linear zones rather than straight lines. Of course, the magnetic lineament between Pickering and Niagara Falls can be recognized unambiguously. However, it is not the highest amplitude magnetic anomaly over Lake Ontario, and the associated gravity signature is very subtle as compared to gravity anomalies further east. Furthermore, the magnetic data seem to indicate that it is not an isolated linear feature, but forms part of a curvi-linear family of lineations.

The existence of the Georgian Bay Linear Zone {Wallach, 1990), i.e. the portion south of Georgian Bay proper, and extending further south towards the Clarendon-Linden Fault Zone in New York State is, however, very questionable. The horizontal gravity gradient (Figure 7) is generally low between Georgian Bay and Pickering, and the anomalies further south do not suggest any linear feature extending to Clarendon-Linden, but rather display general NNW-SSE trends. It should be noted that the Clarendon-Linden Fault Zone itself is accompanied by relatively large gravity gradients anomaly (> 1.5 mGal/km), as are more active fault zones, for example in California. In the magnetic anomalies, a very subtle expression along the Georgian Bay Linear Zone may seem to reach Pickering, but this could very well be an artifact of the NW-SE -C14- I orientation of the flight lines in this area (Figure 2), and a NNW-SSE trending boundary between I surveys near 44° 15' N, 79°30' W. After application of a flight line decorrugation method, ™ evidence for this part of the Georgian Bay Linear Zone largely disappears. Moreover, the pseudo- gravity method applied to the data before this correction (Figure 4) yields only very few trends • between Georgian Bay and Pickering, one of which lineations is clearly a flight line artifact (44°12' N, 79°45' W), and another ( 44°15' N, 79°30' W) caused by the above mentioned ^ survey boundary. •

Conclusions 0

A more detailed analysis of the magnetic data over Southern Ontario, including forward fl modelling, will have to be carried out in conjunction with gravity and seismic reflection data and is • beyond the scope of this presentation. However, what I have presented here forms the basis of a quantitative analysis of potential field data in the study area. Based on this preliminary analysis the S following conclusions can be formulated: •

- Aeromagnetic and gravity anomalies generally reflect NNW-SSE trends in the Central • Meta-sedimentary Belt - Going from north to south, the Niagara-Pickering Lineament shows a gradually increasing « depth to magnetic sources along its strike, between 2 and 4 km below the surface. ]£ - High amplitude magnetic anomalies in the Elzivir Terrane and along the Elzivir-Frontenac Boundary Zone parallel the Niagara-Pickering Lineament, and exceed it in amplitude. - The horizontal gravity gradient in western Lake Ontario is generally small. This in contrast to the active Clarendon-Linden Fault Zone which shows high gravity gradients - The relatively shallow magnetic sources indicate that the observed lineaments may not be relevant to seismic hazard, in view of shallow dipping boundaries seen in seismic reflection profiles {Forsyth et al, 1994). - The Georgian Bay Linear Zone to the south of Georgian Bay proper is not present in horizontal gravity gradient maps, and is questionable in aeromagnetic data, due to flight line orientation and survey block edge artifacts. - A quantitative analysis of the potential field data in conjunction with existing seismic reflection profiles is essential to understand structural implications, and tectonic history of this area in general, and the Niagara-Pickering Lineament in particular. -CIS-

References

Blakely, R.J., and Simpson, R.W., 1986, Approximating edges of source bodies from magnetic or gravity anomalies: Geophysics, 51, 1494-1498.

Forsyth, D.A., Milkereit, B., Davidson, A., Hanmer, S., Hutchinson, D.R., Hinze, W.J., and Mereu, R.F., 1994, Seismic images of a tectonic subdivision of the Grenville Orogen beneath lakes Ontario and Erie, Canadian Journal of Earth Sciences, 31, 229-242.

Roest, W.R., Verhoef, J., and Pilkington, M., 1992, Magnetic interpretation using the 3-D analytic signal: Geophysics, 57, 116-125.

Wallach, J.L., 1990, Newly discovered geological features and their potential impact on Darlington and Pickering, Rapport, Atomic Energy Control Board INFO-0342, 20 pp.

Figure Captions

Figure 1: Map of southern Ontario, showing the highly simplified locations of geological and geophysical features referred to in the text (P=Pickering; D=Darlington).

Figure 2: Aeromagnetic flight line distribution over part of southern Ontario and Lake Ontario. Differences in flight line separation and orientation reflect the different survey blocks.

Figure 3: Shaded relief map of the magnetic anomalies. Note that lighter greys generally reflect more positive anomalies. An artificial illumination from the NW is superimposed to highlight some of the magnetic trends.

Figure 4: Circle representation of pseudo-gravity depth solutions, derived from the magnetic data. The positions of the circles generally reflect the likely locations of magnetic contacts. The size of the circles is a measure of the depth to the magnetic source, with larger circles reflecting deeper sources.

Figure 5: Gravity station distribution over southern Ontario and northern U.S.

Figure 6: Bouguer gravity anomaly map based on stations in Figure 5.

Figure 7: Shaded relief representation of the horizontal gravity gradient, calculated from Figure 6. High gravity gradients are dark, low gradients light. Superimposed is an artificial illumination from the NW, to highlight some of the more subtle variations. Central Gneiss Belt

45 N Central Meîa-Sedimentary Beit Frontenac I 44 N

Lake Ontario n Elzivir y—• I

Niagara '* 43 N

Lake Erie Clarendoh-Linden f 42 N n 81 W 80 W 79 W 78 W 77 W 76 W -C17-

o _J O LU Z o o DC LLJ

Figure T latitude increment 1.00 longilude increment 1.00

I o CO I

Scale 1 : 2157152. -81 -80 -79 -78 -77 Tronsverse Mercator Projection f n -C19-

Fi sure -C20- I I 1 I I

0) c o • ••• ai

CD o -C21-

IT)

I I I I I I I I I I I

V- .3!

\ ! I a. I I I I § s I te.8 ?! SI oo I Figure I -C22- I I 1 CM «- I I I I I I 1 I I 1 I I I t I Figure I I -C23-

I Extension of the St. Lawrence fault zone southwest of Cornwall 1 into Lake Ontario J.L. Wallach

I Adams and Basham (e.g. 1989) stated that most large earthquakes in eastern Canada have occurred near Paleozoic or younger rift zones. They suggested that the branch of the St. Lawrence fault zone (SLFZ), to which they referred as the St. Lawrence rift system I (Kumarapeli and Saull, 1966), and which extends southwestward along most of the length of the St. Lawrence River, continues through lakes Ontario and Erie, toward New Madrid, Missouri. Seismic activity associated with this zone includes at least two M«7 earthquakes I in the Charlevoix region of Quebec, though Adams and Basham argued that similar earthquakes could occur anywhere along it. That being the case may add significantly to the seismic hazard at Darlington and Pickering because the zone apparently passes within I about 30 km of both. I Following the papers published by Adams and Basham, on the extension of the St. Lawrence fault zone, Adams subsequently announced (MAGNEC meeting, 1991) that the zone does not continue upstream beyond Cornwall, Ontario, a position that has been I reinforced in a recent publication (Adams et al, 1995). There they stated that a robust approach to seismic hazard assessment assumes that anywhere along a major tectonic rift, a large magnitude earthquake could be expected, even where no major event has been I heretofore documented. They showed a seismic hazard map of the SLFZ with elevated levels of ground motion, parallel to the trend of the zone, as far southwest as Cornwall, I but no further. This implies that they continue to believe that the SLFZ stops there. Prior to the most recent paper by Adams et al (1995), preliminary reconnaissance geological work was undertaken, southwest of Cornwall, in the area extending from ft Prescott, Ontario and Ogdensburg, New York, southwest, to Prince Edward County in Ontario, about 240 km from Cornwall. The purpose was to determine whether or not the SLFZ does extend beyond Cornwall. Because rocks deform at all scales, emphasis was i placed on seeking outcrop-scale evidence of faults, expressed as breccia zones, slickensided surfaces or displacements. Numerous examples of mesoscopic-scale brittle 1 faulting occur in that area. Macroscopic-scale differential vertical displacements, on structures which are parallel or sub-parallel to the SLFZ, are also present (see below), and a fault extends from western Prince Edward County, southwestward beneath Lake Ontario i towards Hamilton Harbor (Ontario Geological Survey, 1991). Near Ameliasburgh, in Prince Edward County, fractures oriented about 065°, with i obliquely pitching slickensides, cut a faulted syenite. Just east of Kingston, an east- northeast-oriented purple-black dike, possibly diabase, cuts through intensely fractured Grenvillian quartzite. The dike is so completely weathered that it can be grooved with a knife blade, and it is characterized by a series of very closely spaced fractures, inferred to be Riedel shears. An outcrop along Highway 401, near Gananoque, reveals at least 2 l normal faults, both of which are oriented about 220°. Slickensides on 045° to 060°- I -CM- I 1 oriented fractures, in the highly fractured Rockport granite exposed on Hill Island, indicate ** left-lateral shear, with a relatively minor component of dip-slip movement. Left lateral shear was also recognized on a nearly east-west striking fracture surface in the upper M Middle Ordovician Verulam Formation, near Napanee (McFall, 1993). McFall (1993) reported strike-slip faults in the limestone quarry at Picton, in Prince Edward County. One ^ is oriented 080°/73°, and deforms an ultramafic dike of Jurassic age. The second is • oriented 058778°W; the nature of the strike-slip movement (dextral or sinistral) in either case could not be discerned. Closely spaced and polished fractures, trending about 040°, g cut an intensely deformed granite adjacent to Black Lake, south of Edwardsville, New I York. Both dip-slip and strike-slip slickensides were recognized, but it was not possible to determine which was younger. Second-order shears and dip-slip slickensides, associated * with at least one prominent outcrop-scale fault, indicate normal fault movement. g The larger scale vertical block movements occur in the area of Brockville, Ontario and fl Black Lake, New York (Figure 1). In the former, lower Paleozoic sedimentary rock units I abut against, and dip away from, a ridge of Grenvillian quartzite and granite which rises prominently above them. On the southeast side of the 030°-trending ridge, the March M Formation carbonates have been uplifted and, near the contact with the topographically • higher Grenvillian granite, are actually bent. The individual beds within the March retain a uniform thickness across the bend, the axial orientation of which is 020°, indicating that I the warping is not the result of onlap, but uplift. On the northwest side of this arch, at Lyn, • bedding in the sandstone of the Nepean Formation is oriented 213705°, 210703° and 218704°. The fault at Black Lake was initially postulated by Guzowski (1978), who • recognized that the prominent Black Lake lineament separates Precambrian rocks on the ™ southeast from Lower Paleozoic rocks on the northwest. That relationship was verified in — this study. M

From direct geological observations it is clear that a major fault zone extends upstream ^ along the St. Lawrence Valley, southwest of Cornwall, to at least Prince Edward County. » The work of the Ontario Geological Survey (1991) and McFall (1993) points out that a fault, named the Hamilton-Presqu'ile, extends further to the southwest, from Prince û Edward County almost into Hamilton Harbor at the western end of Lake Ontario. £

In addition to bedrock faulting, geologically young faulting has been recorded in • unconsolidated sediments beneath Lake Ontario along the extension of the SLFZ. P Thomas, et al (1993) and McQuest Marine (1995) identified a series of east-northeast trending faults cutting glaciolacustrine clays and modern muds in the SOSZ of the • elongate Rochester Basin, in southeastern Lake Ontario. Those faults, with throws in the • lake-bottom sediments of up to 22 m, lie in a 15 km-wide zone along the the southern margin of the extension of the SLFZ, as originally postulated in the series of papers I written by Adams and Basham (e.g. 1989). McQuest Marine (1995) also carried out ™ seismic surveys in the vicinity of the Niagara-Pickering linear zone (NPLZ) in western Lake Ontario, along the lakeward extension of the Dundas Valley, which parallels and lies • along the SLFZ extension. They reported that "four clearly expressed faults are also seen transecting the lower channel fill deposits, reflectors C and D, as well as the bedrock". — f I -C25-

I Lewis et al (1995 and personal communication) produced a set of seismic reflection profiles, on which they imaged a number of monoclinal warps in unconsolidated sediments extending up to the bed of Lake Ontario (Figure 2), in the same general area of western I Lake Ontario surveyed by McQuest Marine. Those warps occur at the intersection of the SLFZ and the NPLZ, and are believed by Lewis et al (1995) to be the result of deposition on an irregular surface. The cross sectional configurations, however, show either that I individual units retain a constant thickness along the higher block, across the warp, and along the lower block, or that a given unit on the upper block is thinner than its counterpart on the lower block. Furthermore, in at least two cases where an unconformity I is well expressed, the same sense of displacement occurs, both above and below the unconformity. Those characteristics argue strongly that the monoclinal warps are the product of recent faulting. Similar features were recognized in the New Madrid seismic 1 zone, by Williams et al (1995), who also interpreted them as faults.

One might argue that the probability of a M=7 earthquake occurring near the two nuclear I power plant sites is very low, and perhaps even so low as to not be of concern. However, eastern Canadian earthquakes have already occurred in unexpected locations, as in the cases of the 1982 Miramichi (mb=5.7) and 1988 Saguenay (mbLg=6.5) earthquakes. I Evidence of outcrop-, and macroscopic-scale faulting along the SLFZ, southwest of Cornwall, proves that the zone extends at least as far as Prince Edward County. Work by the Ontario Geological Survey shows that it extends further to the west Observations of I recent displacements in the lake, along the extension of the SLFZ, as projected by Adams et al (e.g. 1989), suggest that the zone is tectonically active. The foregoing combined with I the position that large magnitude earthquakes can occur anywhere along a seismically active structure, even where none has been recorded previously (Adams et al, 1995), argues that a large magnitude earthquake in the vicinity of the Darlington and Pickering I nuclear power plants should be regarded as a credible event.

1 I I I I I -C26- 1 REFERENCES I Adams, J. and Basham, P., 1989. The seismicity and seismotectonics of Canada east of the Cordillera. Geoscience Canada, 16, 3-16. I Adams, J, Basham, P.W and Halchuk, S., 1995. Northeastern North American earthquake • potential - new challenges for seismic hazard mapping. In Current Research, 1995- fi D. Geological Survey of Canada, 91-99. I Basham, P.W. and Adams, John, 1989. Problems of seismic hazard estimation in regions 9 with few large earthquakes: Examples from eastern Canada, in Berry, M.J. (éd.), Earthquake Hazard Assessment and Prediction. Tectonophysics, 167,. 187-199. fl Guzowski, R.V., 1978. Stratigraphy, structure and petrology of the Precambrian rocks in the Black Lake region, northwest Adirondacks, New York. Unpublished Ph.D. • thesis, Syracuse University, Syracuse, New York, 184 pp. plus maps. '

Kumarapeli, P.S. and Saull, V.A., 1966. The St. Lawrence valley system: a North I American equivalent of the east African rift valley system. Canadian Journal of Earth Sciences, 3, 639-658. ~

Lewis, C.F.M., Cameron, G.D.M., King, E.L., Todd, B.J. and Blasco, S.M., 1995. Structural contour, isopach and feature maps of Quaternary sediments in western Lake Ontario. Atomic Energy Control Board, in press. 72 pp. plus figures fg

McFall, G.H. 1993. Structural elements and neotectonics of Prince Edward County, M southern Ontario, in Wallach, J.L. and Heginbottom, J.A. (eds.) Neotectonics of £ the Great Lakes Area, Géographie physique et Quaternaire, 47, 303-312.

McQuest Marine Sciences Ltd., 1995. The geophysical survey of Lake Ontario in 1993. p Atomic Energy Control Board INFO-0539, 17 pp plus figures.

Ontario Geological Survey, 1991. Bedrock geology of Ontario, southern sheet. Ontario ™ Geological Survey, Map 2544, scale 1:1 000 000.

Thomas, R.L., Wallach, J.L., McMillan, R.K., Bowlby, J.R., Frape, S., Keyes, D. and ™ Mohajer, A. A., 1993. Recent deformation in the bottom sediments of western and ft southeastern Lake Ontario and its association with major structures and seismicity. • in Wallach, J.L. and Heginbottom, J.A. (eds.) Neotectonics of the Great Lakes Area, Géographie physique et Quaternaire, 47, 325-336. g

Williams, R.A., Luzietti, E.A., and Carter, D.L., 1995. High-resolution seismic imaging of Quaternary faulting on the Crittenden County fault zone, New Madrid seismic zone, northeastern Arkansas. Seismological Research Letters, 66, 42-57. I 1 I I -C27- I I I I

I Undifferentiated, flat-lying I Nepean and March Formations I

I _5000 meters

I Precambrian Basement I Fault I I I Figure 1. Generalized geological map of the Brockville-Black Lake area. I I I 1 I -C28- i N METERS I 300 380 460

140- I M90 1 I I I I E I I I 180-L: VE: 4.2x I I Figure 2. Seismic profile showing a monoclinal warp, in unconsolidated sediments beneath western Lake Ontario, in the vicinity of the NPLZ and the SLFZ. O and Y refer to I orange and yellow reflectors (Reproduced from Lewis et al, 1995). I 1 1 1 i -C29- rAVfx1^ Implications for Seismic Risk in Southern Ontario • Based on a Review of the Region's Geology

• KM. Easton, D.K. Armstrong, and P.C. Thurston, Ontario Geological Survey,

933 Ramsey Lake Road, Sudbury, Ontario P3E 6B5 I • Historically, southern Ontario has long been regarded as having a geologic structure different from more

seismically active areas of eastern North America. Over the past decade, however, a variety of mapping,

P compilation and multidisciplinary research efforts all suggest that this view is incorrect. In this paper, we review I the geology of the area, discuss the implications with rebpect to seismic risk, and identify research projects that will allow us to better evaluate the seismic risk of southern Ontario. B

• Observations Based on Precambrian Bedrock Mapping

As recently as 1982, interpretations of the extension of major Precambrian basement structures such as the

I Central Gneiss Belt/Central Metasedimentary Belt boundary zone (CMBBZ) into southern Ontario (Figure 1) I suggested that these structures trend from Lake Simcoe toward Cambridge and Point Pelee, and had no strong associated magnetic or gravity expression. Subsequently, detailed mapping of the exposed CMBBZ in Ontario by

• the OGS-GSC, the acquisition of modern aeromagnetic data over the Great Lakes, and mapping of the basement

• beneath the Paleozoic cover using drillhole data and geophysics by OGS-MNR, has indicated that many northeast-

striking structures in the exposed shield became north-trending beneath the Paleozoic cover (Figure 1), and that

• the major Precambrian structures have associated with strong aeromagnetic, and locally gravimetric anomalies.

— A prime example is the coincidence of the CMBBZ in the Toronto area with the Niagara-Pickering aeromagnetic

lineament (NPLZ). Seismic reflection studies have shown that both the CMBBZ and the Elzevir-Frontenac

• boundary zone (EFBZ: roughly coincident with the Clarendon-Linden faults, Figure 1) extend deep into the crust

(cf. Forysth et al. 1994). I I -C30-

8V 7ff IT 76* I

<Ù 100 Km Sudboty I Quebec I 46-' I I PROVINCE Central Metasedimentary I 1

PALEOZOIC STRATA I I I I U.S.A. P = Pickering D = Darlington I = areo of post—Ordovician calcite—flourite-barite vein systems + isotopic resetting I = area of barite-calcite-veins I I

Figure 1: Geologic sketch map of south-central Ontario showing major Precambrian and Paleozoic fault zones, I regional lineaments, and distribution of main Paleozoic vein systems and zones of isotopic resetting. 1 1 I ~c31~ * Most of the Grenville Province structures show dominantly ductile fabrics, suggesting that they originated by

I Proterozoic-age faulting. In contrast, the Roberston Lake mylonhe zone (RLMZ), which is the on-shield extension

of the Clarendon-Linden Fault Zone (CLFZ, Figure 1) is dominantly a brittle structure, of presumed

™ Neoproterozoic age, although there is some indication that the RLMZ might locally offset west-trending Ottawa-

I Bonnechere graben faults of Paleozoic age.

Detailed mapping of the Precambrian basement in eastern Ontario, away from the Ottawa-Bonnechere and St.

H Lawrence grabens indicates a complex, post-900 Ma, fracture framework, consisting of northwest- to west-

• trending faults, with apparent sinistral and dextral offsets of 100-500 m, and with vertical offsets sufficient to

disturb metamorphic isograd patterns within the basement. The occurrence of Grenville swarm diabase dikes along I some of these faults, as well as U-Pb dating of hydrothermal zircons adjacent these faults suggests a maximum age I of 600 Ma (Corfu and Easton 1995; Kamo etal 1995).

I Observations Based on Paleoùc Bedrock Mapping I Faults which cut Paleozoic strata in southern Ontario are best known from the subsurface of southwestern Ontario and in eastern Ontario. Faults with small-scale displacements (i.e., not

mappable), mineralized fractures, and stress-release features (i.e., pop-ups) are known from surface mapping of

Paleozoic bedrock in south-central Ontario. The faults in southwestern Ontario have been identified by subsurface

mapping using the extensive database of petroleum exploration wells and by industry-conducted seismic reflection

• surveys (Carter et al. 1993). Most of these faults are known to cut only the Precambrian basement and lower

« Paleozoic section, however thick drift cover obscures evidence of more recent fault displacements. Gas seeps do

indicate that conduits extend from the lower Paleozoic to surface, however it is not known whether these conduits

H are faults or fractures and whether these formed due to neotectonic or glaciotectonic forces.

a An extensive and well documented fault system cuts the Paleozoic strata of eastern Ontario and is related to

the development of the Ottawa-Bonnechere graben. Although mostly confined to the eastern side of the Frontenac I I I -C32- 4 I axis, possibly related faults are known from as far west as Rice Lake. Latest tectonic-driven movement on these I faults probably occurred during the Mesozoic, as indicated by the presence of slickenslides on Mesozoic dikes in

Prince Edward County. Seismic reflection data in Lake Ontario (Forysth et al. 1994) shows offset of the Paleozoic ' section along the EFBZ, which is coincident with regionally thinning of the Paleozoic section over the EFBZ and JÊ restriction of preserved Cambrian strata to east of the EFBZ (Liberty 1969). Timing of these offsets is unknown.

No comprehensive structural study of the Paleozoic bedrock in south-central Ontario has yet been conducted. •

Sanford (1993), using detailed surveying of specific contacts, reported that in the Burleigh Falls-Head Lake area, 1 some lineaments are related to faults with significant vertical displacements. Rutty and Cruden (1993) found that lineaments in the Balsam Lake area coincided with fractures, joints and pop-ups. Armstrong and Rheaume (1993) ™ report a variety of structural features in the Lake Simcoe area, including: mineralized fractures (some with a small I strike-slip component), a large dolomitized zone, numerous normal faults with small-scale vertical displacements and a few faults with metre-scale throw. • 1

Observations Based on Quaternary Mapping Studies

Mohajer et al (1992) identified normal faults in the Paleozoic bedrock and the overlying Quaternary sediments ™ in the Rouge River valley. The origin of these faults is unclear, neotectonic and glaciotectonic origins have been • suggested. It is not known if these faults are localized elsewhere over other Precambrian structures (EFBZ, other 1 parts of the CMBBZ), or whether normal faults occur in greater abundance in the Rouge River area relative to ^ the Quaternary section elsewhere in south-central Ontario. I I

Other Observations I

Hyodo et al. (1993), on the basis of paleomagnetic and Ar/Ar studies suggest the significant differential uplift occurred across the Ottawa-Bonnechere graben. This is consistent with the aforementioned extensive fracture • I I [ -C33- 5

framework in eastern Ontario. This technique also provides a means of estimating relative displacement across

I faults in the shield.

• Calcite+flourite+/-celestite+/-barite veins in eastern Ontario are associated with many of the larger regional

structures, such as the Clarendon-Linden Fault Zone (CLFZ-EFBZ) and its branches (Figure 1), but also occur

I along northwest- and west-trending faults. These veins are associated with regional isotopic resetting of Ar-Ar n and Rb-Sr systems that give apparent ages of ca. 420 Ma, and fluid inclusion studies that suggest minimum filing

temperatures of 122-132°C (McCartney 1964). It can be argued that the faults served as pathways for fluid

• migration, which is reflected in the distribution of these veins and the pattern of isotopic resetting. Similar vein

M systems occur within the Paleozoic bedrock. For example, Gross et al (1992) studied the Silurian Lockport

Formation in New York State and suggested that the presence of a calcite-filled vein system on the east side of

M the CLFZ indicated that the fault zone was a major conduit during the Acadian Orogeny, releasing fluid and high i0 fluid pressures on the east side of the fault as the result of Acadian compression. Recently completed isotopic studies (Bowins et al. 1992) and lake bottom geological and geophysical work

• (Blasco et al. 1995) clearly indicate further work is needed in these fields with respect to Southern Ontario I seismicity. Discussion and Synthesis 1 Until recently, no regional framework existed to account for the seismicity or the reactivation of older structures in eastern North America. Recently, both Wheeler (1995) and Adams et al. (1995) have suggested that 1 • seismicity in eastern North America is related to the reactivation of faults developed during Iapetan rifting (c. 620-

1 590 Ma), either in orogen parallel fault systems (e.g., the CLFZ) or in orogen-normal failed rift systems (e.g.,

Timiskaming-Ottawa-Bonnechere graben). This framework also explains the aforementioned distribution and

? timing of calcite-vein and isotopic resetting systems in south-central Ontario.

H If this framework is a useful predictor of seismic hazard, than the area east of the CLFZ should be more

seismically active (and hazardous?) than the area to the west. As shown by Wheeler (1995), the Niagara-Pickering I I 1 -C34- 1 6 linear zone (NPLZ-CMBBZ) lies in a less seismically active area, although given the regional scale of Wheeler's i study, this difference may not be geologically (or seismologjcally significant), especially as the NPLZ-CMBBZ and 8 the CLFZ-EFBZ converge near the Niagara Peninsula (Figure 1). In addition, a zone of slightly above average _. seismicity in the vicinity of Lake Erie and western Lake Ontario (cf. Adams et al. 1995) remains unaccounted for ™ by the reactivated Iapetan fault model. fl

Recent studies in the New Madrid area and along the New York-Alabama lineament, both along the Iapetan- M related seismic zone of Wheeler (1995), indicate that even though these areas are currently seismically active, this activity is relatively recent (> 10,000 ka), older strata do not show evidence for a long history of faulting. If a I similar pattern were present along either the CLFZ or the NPLZ, then the present paucity of evidence for older ~ seismic activity along these structures does not necessarily indicate low seismic risk in these areas. ™ I Summary

Over the past decade, we have been able to more precisely map the distribution of Precambrian and post- I

Precambrian structural (fault) zones in southern Ontario, to link these structures with aeromagnetic and H gravimetric anomalies, and to use seismic reflection imaging to better understand the crustal extent of these _ structures. In doing so, we have found that there are both ductile and brittle faults within the Precambrian. Some of the Precambrian faults can be traced into the lower Paleozoic section. To date, our best estimates indicate that • the fracturejramework seen in the Precambrian shield in southern Ontario developed mainly during Iapetan rifling m 1

(ca 620-590 Ma), and was locally reactivated in the Ordovician (ca 420 Ma) and in the Mesozoic (ca 175 Ma).

We are now at the point of being able to group these fault systems into different sets, of possibly different seismic I risk. _

Areas of uncertainty remain in establishing if any of these fault systems affect Quaternary strata in the area, the extent and timing of any Tertiary movement along any of these faults, and in understanding the distribution of I I I -C35- I small-scale structural features (pop-ups, joints, outcrop-scale faults) relative to the regional-scale structures shown i in Figure 1. In summary, as far as the question of seismic risk is concerned, it is our view that we have moved from a

I geologic framework indicative of low seismic risk to a framework suggestive of higher seismic risk, especially I given the current gaps in our knowledge of the geologic history of southern Ontario. A much better understanding of the seismic risk of the region can be gained by undertaking the following research:

I • shallow seismic studies in the Rouge Valley area should be able to resolve the question of the continuity (or I lack thereof) of Paleozoic and Quaternary faults, and possibly their origin. • a variety of dating methods, including electron-spin resonance, should be applied to the youngest faults we can

I identify, to determine age of last movement. i • a combined ground and remote-sensing structural study of the Paleozoic from the Lake Simcoe to Ottawa River region would serve to characterize regional structural trends and identify any changes in density of structures

I across major Precambrian structures such as the CMBBZ and the EFBZ. This would help establish if these I smaller features are actually bellweathers of accumulating stress.

References i Adams, J., Basham, P.W., and Halchuk, S. 1995. Northeastern North American earthquake potential - new challenges for seismic hazard mapping; Geological Survey of Canada, Paper 95-ID, p. 91-99. Armstrong, D.K. and Rheaume, P. 1993. Paleozoic mapping and alkali-reactive aggregate studies in the Lake Simcoe area; Ontario Geological Survey, Miscellaneous Paper 162, p. 149-153. I Bowins, R.J., McNutt, R.H., Drimmie, R.J., and Frape, S.K., 1992. elemental and Sr itotopic Analyses of Lake Ontario bottom sediment Porewater - possible groundwater contributions from other sources. EOS supplement 73:139 Carter, T., Trevail, R., and Easton, R.M., 1993. Oil and Gas accumulations and basement structures in southern Ontario, Canada, Ont. Petroleum Inst Proc. 32nd Ann. conf. p. 1-27. I Corfu, F. and Easton, R.M. 1995. U-Pb geochronology of theMazinaw Terrane, an imbricate segment of the Central Metascdimentary Belt, Grenville Province, Ontario; Canadian Journal of Earth Sciences, v. 32. Forsyth, D.A., White, D. Zelt, C.A. and Easton, R.M. 1994. Seismic images of Grenville basement and Appalachian cover I beneath southeastern Ontario and adjacent New York State; Geological Society of America, Abstracts with Program, v. 26, #7, p. A-470. Gross, M.R., Engelder, T. and Poulson, S.R. 1992. Veins in the Lockport Dolostone: Evidence for an Acadian fluid circulation system; Geology, v.30, p. 971-974. I Hyodo, R, York, D. and Dunlop, D.J. 1993. Tectonlhermal history in the Mattawa area, Ontario, Canada, deduced from paleomagnetism and 40Ar/39Ar dating of a Grenville dike; Journal of Geophysical Research, v. 98, p. 18001-18010. Kamo, S.L., Krogh, T.E., and Kumarapeli, P.S. 1995. Age of the Grcnville dyke swarm, Ontario-Quebec: implications for the timing of Iapetan rifling; Canadian Journal of Earth Sciences, v. 32, p. 273-280. I Liberty, B.A. 1969. Paleozoic geology, Lake Simcoe area; Geological Survey of Canada, Memoir 355,201p. McCartney, W.D. 1964. Barite-Fluorite study; Geological Survey of Canada, Paper 64-2, p. 75-78. I I I -C36- I I Mohajer, A., Eylcs, N. and Rogojina, C. 1992. Neotcctonic faulting in metropolitan Toronto: implications for earthquake hazard assessment in the Lake Ontario region; Geology, v. 20, p. 1003-1006. Rutty, AX. and Cruden, A.R. 1993. Pop-up structures and the fracture patterns in the Balsam Lake area, southern Ontario; I Géographie Physique et Quaternaire, v. 47, p. 379-388. • Sanford, B.V. 1993. Stratigraphie and structural framework of upper middle Ordovician rocks in the Head Lake-Burleigh Falls area of south-central Ontario; Géographie Physique et Quaternaire, v. 47, p. 253-268. a Wheeler, RX 1995. Earthquakes and the cratonward limit of Iapetan faulting in eastern North America; Geology, v. 23, p. H 105-108. " I I I I I I I I I I I I I I I I -C37— I I ABSTRACT Geophysical Signatures in the Bed of Lake Ontario

I R.L. Thomas Waterloo Centre for Groundwater Research

1 A number of geophysical cruises of short duration were run on Lake Ontario directed towards obtaining data on a number of features that were initially observed in 1987 (Thomas et al., 1993). Three kinds of features were observed, the origins of which appear to be a matter I of some controversy. These geophysical features are discussed briefly as follows:

Plumose Structures: Observed only by side scan; a long feather-like feature etched in bottom I silty-clay sediment south of Bronte in the northwestern part of the lake. Thomas et al. (19^3) ascribed this feature to a neotectonic origin since it was close to the epicentre of the 1987 Burlington earthquake of magnitude 3.5. Lewis (AECB report) identified bed features in the I anchorage of the Welland Canal, some of which were quasi-feather-Hke in appearance. By analogy, he thus concluded that all plumose structures are anchor drag. Firstly» it should be I noted that anchor makings are generally quite distinct and it is difficult to consider how a ship whilst dragging its anchor could create a feature two kilometres long showing almost perfect bilateral symmetry. Secondly, Pecore and Fader (1990) stated that plumose structures observed I in Passamaquoddy Bay can be interpreted to be die result of recent activity on faults. Clearly, in Lake Ontario, further evaluation of the Bronte plumose structure must be undertaken in order to understand the mode of origin. Such evaluation must be accompanied by high resolution I seismic to determine the structure of the underlying material. Side Scan Backscattering Features: Thomas et al. (1993) observed linear dark features on side I scan sonar records of the lake bed off Bronte. In subsequent surveys, these features with many variants of the backscattering appearance were observed at all areas in the lake which are believed to overly major structures. These areas include the Burlington-Toronto, Niagara- I Pickering, Wilson-Port Hope, and the SOSZ inferred fault zones. Lewis (AÉCB report) has called these features LABAs (Linear Acoustic Backscattering Anomalies), some of which he observed radiating from the Welland Canal. Bed sampling revealed the presence of cinders I derived from old steamers. Lewis thus concluded that all LABAs originated by stern chute ejection of cinders from old steamers. It is well known that samples taken on the ship tracks on the Great Lakes produce abundant cinders. As a consequence, it may be assumed that the cinders i change the acoustic properties of tiîe surface sediment and hence will result in LABAs in a aide scan record, however, it is misleading to assume (as many have done) that all LABAs are cinders. Many examples of intense backscattering occur off the main shipping tracks and at a I frequency and orientation that make the cinder hypotheses untenable. This is particularly true of the deep waters of the SOSZ. Alternate hypothesis would imply the unlikely effects of water content (ie., upwelling groundwater) or a more plausible degassing through recent sedimentary I layers. Indeed, piston cores taken in the SOSZ self extruded due to natural gas expansion particularly in the deeper glacial clays. This gas expansion is often seen in Lake Eric. With the array of backscattering features that are observed, it does not appear to be necessary to evoke I a single origin but that all mechanisms are in effect at different locations in the lake. High I -C38- I precision sampling and selective analyses will be necessary to satisfactorily resolve the | controversy related to LABA origin. However, the feet that they are easily found relative to the structural fabric of the lake indicates the strong possibility that certain LABAs may tie a • powerful means for reconstructing structure and for the exploration of natural gas and | hydrocarbons.

South, Ontario Structural Zone: Following the original echo-sounding work reported by Thomas I et al. (1993), which covered the entire width of the SOSZ, more recent work has focused on evaluating the apparent fault structures on the southern margin of the zone. Hutchlnson et al. • (1993), in their work in the same area, postulated that the configuration of this zone was due I to the occurrence of drumlins. Cross-sectional seismic data in the central core of the area indicate the occurrence of till which provides strong support for their hypothesis. However, their B data does not account for bedrock displacements which transmit throughout the post-glacial fl sequences, suggesting recent faulting along the prime orientation of the SOSZ (065°-070°). Further, sedimentation in the SOSZ shows relatively constant thicknesses on both sides of the B displacements and also undulates over underlying material without any indication of overstepping • or thinning which implies that movement occurred during and following sedimentation. This is totally different to the deep northern basin of Lake Huron where modern sediment is filling in B hollows and overstepping the peaks to varying degrees. These were characterized as Type B and • C basins (Thomas et al., 1973). Again, as for the previously described features, different origins can be ascribed to the I features in the SOSZ. Both structural and glacial origins and modifications can be used. • However, to what extent the glacial features (trending parallel to the structural features) are attributable to structural control remains to be determined. fl Further work is clearly needed and it is now imperative that the full extension of the * SOSZ from Rochester to Mexico Bay be studied in some detail. Such an evaluation is needed to confirm or reject the extension theory of the St. Lawrence rift fault system through Lake I Ontario and onwards to Lake Erie and beyond.

REFERENCES |

Hutchinson, D.R., Lewis, C.F.M., and Hund, G.E., 1993. Regional stratigraphie framework m of surficial sediments and bedrock beneath Lake Ontario. Géophysique, B Geographiephysiquc et Quaternaire, 47:337-352. Lewis, C.F.M., 1994. Report to the Atomic Energy Control Board. • Pecore, S.S. and Fader, E.B.J., 1990. Surficial geology, pock marks and associated neotectonic I features of Passamoquoddy Bay, New Brunswick, Canada. Geological Survey of Canada, Open File Report #2213, 46 pp. • Thomas, R.L., Wallach, J.L., McMillan, R.K., Bowlby, J.R., Frape, S.K., Keyes, D., and fl Mohajer, A.A., 1993. Recent deformation in the bottom sediments of western and southeastern Lake Ontario and its association with major structures and seismicity. • Géophysique, Geographiephysique et Quaternaire, 47:325-335. B Thomas, R.L., Kemp, A.L.W., and Lewis, C.F.M., 1973. The surficial sediments of Lake Huron. Can. I. Earth Sci.. 10:226-271. • I I -C39-

I NEOTECTONIC FEATURES IN THE QUATERNARY SEDIMENTS OF LAKE ONTARIO: INTERPRETATION OF GEOPHYSICAL SURVEY DATA

I C.F.M. Lewis, Geological Survey of Canada (Atlantic) Box 1006, Dartmouth N.S. B2Y 4A2 I INTRODUCTION Neotectonism in Lake Ontario was first interpreted by Thomas et al. (1989a,b) on the basis of I previously unrecognized surface sedimentary features imaged by sidescan sonar. These included ridges interpreted as stress-release pop-up structures in the underlying bedrock, delicate plumose structures believed to be a surface expression of stress release below modern I silty clay sediment, and dark linear features thought to be sonar responses to intrusion (from below) of coarse sediment into surface silty clays or dewatering of sediment. It was further noted that areas containing the plumose structures and dark linear features (western Lake Ontario, I north side) fell into an alignment conforming to those revealed by aeromagnetic surveys. Aeromagnetic and other geophysical lineaments were recognized and postulated as signatures of Pre-Paleozoic structures which might be rejuvenated with consequences for the seismic I hazard at Darlington and Pickering nuclear power stations (Wallach and Mohajer, 1990). Two lineaments, termed the Niagara-Pickering Linear Zone (NPLZ) and the Georgian Bay L inear Zone (GBLZ), were observed to intersect at the northern shore of Lake Ontario in the vicinity of I the power stations. A consideration of the implications of both the newly discovered lake sedimentary features and the geophysical lineaments as evidence of neotectonism suggested I an increased seismic hazard for Darlington and Pickering (Wallach, 1990). In 1991, fault-like structures trending 070° in southeastern Lake Ontario were reported in echosounding and shallow seismic records (Thomas et al., 1991,1993). These structures were I designated as the South Ontario Structural Zone (SOSZ), and together with the western Lake Ontario features and other alignments of geophysical attributes were proposed as evidence of I the St. Lawrence Rift extending through Lakes Ontario and Erie. The foregoing observations and deductions suggest that an active fracture framework exists beneath Lake Ontario. Regional geophysical surveys using high-resolution seismic reflection I profiling, sidescan sonar and swath bathymetry mapping were conducted in 1992-1994 to image-key areas of the lakebed and the underlying Quaternary sediments in Lake Ontario to determine the presence of faults or effects of earthquakes, and whether these had been active I since the last glacial ice cover about 13,000 radiocarbon years ago (13 ka). Such evidence could have implications for the seismic hazard of the north shore of Lake Ontario where the I Pickering and Darlington power stations are located. SURVEYS AND SAMPLING I In 1992, high-resolution seismic reflection profiles and sidescan sonar records were obtained with CCGS Griffon on a rectangular grid of tracklines with 5 km separation north-south and 10 km east-west in the western part of the lake (Fig. 1)(Lewis et al., 1992). This grid was designed to sample unaffected areas and those potentially affected by the NPLZ and GBLZ geophysical I lineaments which passed near the Darlington and Pickering sites. The boomer seismic profiling I system (SEISTEC) operated at frequencies between 2 and 8 kHZ and was capable of vertically I -C40-

resolving bed thickness changes down to about 0.25 m. Sleeve gun (airgun) seismic profiles I were also obtained to track the bedrock surface and, in shale lithology, some subsurface reflectors. A 100 kHz Klein sidescan sonar was operated at 150 m range to either side of its — towfish. In 1993, swath bathymétrie map coverage was obtained in two test areas in eastern I and western Lake Ontario using the Simrad EM1000 technology aboard CSS Frederick G. ™ Creed (Fig. 1). This 90 kHz system also mapped acoustic reflectivity. Additional high-resolution seismic profiles and sidescan sonar were obtained from Canadian Coast Guard vessels in 1993 S and 1994 to add data from eastern Lake Ontario and to augment the 1992 coverage of the • western part of the lake. Sleeve gun (airgun) seismic reflection profiles were also obtained on these cruises, but were only occasionally consulted for the present study. Sediment cores up to I 15 m in length were recovered to groundtruth selected reflections and features identified in the • seismic profiles. Surface sediment samples (Van Veen grab permitting 10 cm cores) were obtained in 1993 and 1994 to determine the physical nature of some acoustic features in the • sidescan sonar records. Submersible diving in 1994 using facilities provided by Harbor Branch I Océanographie Institution and the Canada Department of National Defence supported visual observations and sampling of some features. •

RESULTS

These studies show that the Quaternary sediments in western Lake Ontario range up to about B 75 m in thickness in offshore areas where a complete section since the last ice cover is generally found (Lewis et al., in press). Most of this thickness was deposited during a 2500-year • interval since the last glacier about 13 ka to about 10.5 ka when deposition from glacial g| meltwater ceased. Since that time, lacustrine fine-grained sediments have slowly accumulated to about 5-10 m thickness in some basinal areas. Thus the Late Quaternary sediments offer an « excellent opportunity to resolve deformation events within the resolution of the survey tools I within the past 13,000 radiocarbon years. The upper parts of this section are missing on the basin margins, especially on the north side, due to erosion by waves and currents in previous — low level lake phases (more than 100 m lower than at present in the western area). The thickest I sediments infill the eastward extension of the bedrock Dundas Valley. This valley trends east- northeast with a slight veering to the north and shallowing south of Toronto over the NPLZ; _ farther eastward the valley broadens into the general bedrock basin beneath Lake Ontario. The • bedrock surface beneath the sediment section in most of the survey is clearly defined within the B resolving power of the seismic reflection profilers which is estimated to be better than 1 m for the SEISTEC profiler and within 1-5 m vertically for the sleeve gun system. I

Bedrock Features fl

In the area off Toronto in 50-100 m water depth an orthogonal pattern of small bedrock ridges (up to 2-3 m high) trending northeast and northwest approximately was imaged that is • consistent with an origin of bedrock buckling or pop-ups as previously interpreted (Thomas et • al., 1989a,b; 1991,1993; Wallach 1990). The pop-ups were confirmed by visual observations from submersible diving in 1994. It is unlikely the pop-ups would have survived the last • glaciation and thus, these features signify neotectonic failure of strata on the Paleozoic I sedimentary bedrock surface under the influence of regional compressive stress within the North American plate (Wallach et al., 1993). Elsewhere in western Lake Ontario, small-scale • I I -C41- I bedrock rises in the high-resolution reflection profiles were interpreted as bedrock pop-ups similar to those verified south of Toronto. These features are widely distributed on the flanks of I the basin, and are spatially unrelated to the geophysical lineaments (Fig. 2). However, deeper occurrences are clustered south of Toronto on the north side of the Dundas Valley in the vicinity of the NPLZ and the Hamilton-Presqu'ile Fault Zone (HPFZ). This association of bedrock ridges, I valley morphology and geophysical lineaments may have tectonic significance and warrants further investigation.

1 Apart from the interpreted pop-up ridges, a few depressions and many positive relief features were imaged on the bedrock surface ranging from 1-15 m in height. Near-surface fold structures were detected in places, mostly in a shale lithology which could be penetrated by the available I seismic energy. In general there is insufficient information to determine the trends of these features due to the relatively wide survey line spacing. The distribution of bedrock relief features and structures tends to follow the basin margins or the side slopes of the Dundas Valley rather I than having a specific relation to the cross-lake geophysical lineaments. Lithological control may be influencing the distribution of bedrock scarps which appear to be more numerous in carbonate rather than shale terranes (Fig. 3). Penetration of the limestone bedrock by the I available airgun seismic reflection system was quite limited; a more powerful system is desireable to detect folding and faulting of subsurface reflectors where the lake is underlain by I carbonate rocks. Sedimentary Features

I The interpretation of the sedimentary section is complicated by the variety of processes which may account for anomalous features, in addition to disturbances caused by neotectonism. In addition to processes associated with glacial loading and slumping of glacial sediment, and of I lacustrine deposition involving consolidation, settling, fluid migration (dewatering) and adjustment to slopes, short-lived but catastrophic outbursts of subglacial meltwater down the axis of the Ontario basin have been recently postulated (Shaw and Gilbert, 1990). Also, man's I activities may affect the lakebed and surface sediments, for example, the effects of shipping which include anchoring, cargo sweeping by modern bulk carriers, and ash disposal by former steamship traffic. Interpretation is also complicated in areas of slopes because off-axis reflections are likely, and these are recorded as though they originated directly below the survey I vessel.

The two lakebed features found on sidescan records which are thought to indicate neotectonic I activity, areas of dark return (here referred to as linear acoustic backscatter anomalies or LABAs) and plumose structures, were also found in the present surveys but their character and geographic distribution were more consistent with shipping than with neotectonic activity. Two of I the LABAs were found to contain gravel-sized pieces of crushed limestone and unburned coal respectively (Cameron and Lewis, 1994). The coarse fragments are the most likely origin of the acoustic backscatter received by the sidescan sonar, and thus, these LABAs probably I originated as debris dumped from bulk cargo carriers while underway. The general trends of LABAs when plotted on a map of the lake radiate from the entrance/exit to the Welland Canal at Port Weller (Fig. 4) or otherwise conform to possible shipping routes. This finding does not rule I out the formation of LABAs by neotectonic activity, but it suggests that any LABA which is I thought to be neotectonic should be associated with and supported by other evidence of recent I I -C42- I faulting or earthquake shaking. The acoustic reflectivity image mapped by CSS Frederick G. 1 Creed in western Lake Ontario in 1993 (Fig. 1) is dominated by linear tracks of high backscatter, generally of north-south orientation; these features are also suggestive of ship debris, possibly steamship ash debris. I

Some plumose structures are recognized as ship anchor or ship anchor-chain markings (Thomas et al., 1993) in addition to those of possible neotectonic origin. It is likely that the I plumose structures seen in the present study are probably related to shipping as they were • found in the designated anchorage area near Port Weller where ships wait their turn to transit the Welland Canal (Fig. 5). As for the LABAs, neotectonic plumose structures should also be H supported by other evidence of recent faulting or earthquakes. •

Enigmatic features of possible but not exclusively neotectonic origin were detected within the • sediment section on seismic profiles and mapped. Widespread discontinuities or notches in a I reflector of 6-8 ka age (Fig. 6) could be buried pockmarks and represent evidence of an earthquake shaking and dewatering event at about that time. Similarly, discontinuities in • packages of acoustic reflections formed in older sediments might represent one or more • seismic shaking events about 12-13 ka (Fig. 6). These reflection eradications (REDs) are also likely erosion effects by catastrophic outbursts of subglacial meltwater. Closely spaced surveys • of these features are required to delineate their structure (e.g. in planview) in order to resolve | their origin and significance. A few normal faults of small (<0.5 m) offset are limited to a zone of acoustically stratified sediment of about 11 ka age on the trend of the NPLZ (Fig. 6). These m could arise as adjustments to settling of underlying sediment; their significance for seismic g hazard is uncertain.

Sediment depressions, which have been draped with packages of acoustically stratified j| sediment, are another widespread feature. These appear to have formed by a hydraulic erosive process such as by subglacial meltwater discharge as they are widespread in the basin (Fig. 5). _ A few depressions which are found on the NPLZ also resemble monoclinal flexures (J.L. I Wallach, personal communication), and these should be more intensively surveyed to deeper levels to confirm or refute the possible presence of an underlying fault.

The swath bathymetry map for a part of the deep water area of southeastern Lake Ontario ™ confirmed the presence of a field of ridges trending about 235+/-50 (Mayer et al., 1994). However, recent seismic profiles (1994) across some of these features indicate that much of I their 10-20 m relief is caused by glacial deposits resting on bedrock. These ridges most closely * resemble drumlins of the Peterborough field which are interpreted to have resulted from selective erosion of an existing till sheet by separated turbulent flows in a subglacial fl sheetflooding event (Shaw and Sharpe, 1987). It is possible the Lake Ontario drumlins were • formed in a similar way by a later meltwater flood event down the basin axis postulated by Shaw and Gilbert (1990). However, secondary features on the swath image are more suggestive of I neotectonism. Depressions of 100-500 m diameter are common on the muddy lakefloor; these I features resemble pockmarks which indicate venting or seepage sites of fluids which originate within or beneath the sediment pile. Most are randomly distributed but alignments of pockmarks • occur in the southeastern sector of the swath image (Fig. 7). These pockmark alignments H suggest the presence of subsurface linear conduits such as fractures or faults; one alignment favours the trend of the SOSZ postulated by Thomas et al. (1991,1993). A faint bathymétrie I I I -C43- I lineament trends northwest across the northern part of the image. This swath area overlies the I Eastern Lake Ontario shear zone as recognized in the Grenville section of a deep seismic reflection profile (Forsyth et al., 1994). More extensive mapping of the lakebed by swath bathymetry is desireable to detect potential alignments of features which could be targeted for I further study as evidence of rejuvenation of deeper structures. CONCLUSIONS

I 1. Surficial bedrock pop-up ridges are common and indicate the pervasive presence of regional compressive stress in the rock mass beneath the lake. However, surveys with more closely spaced tracklines are required to determine the trends of sediment-buried ridges and whether I the sress orientations in these areas are similar throughout and consistent with those onshore.

2. The presence of anomalous bedrock ridges and relief, and possible monoclinal flexures at the I intersection of the NPLZ and the Dundas Valley raises the likelihood of tectonic features in this area; further investigation of potential neotectonic activity is desireable.

I 3. Tectonic features (folds and deformation) at the bedrock surface can often be detected in shale lithologies. More powerful systems are needed to obtain structural information in carbonate terranes. More closely-spaced lines are needed, as for the bedrock ridges, to verify I the trends of tectonic features where imaged.

4. Some acoustic backscatter images on the lakebed (LABAs), which are imaged as areas of I dark return on sidescan sonar records, are known to be caused by ship debris. Where similar features are interpreted as neotectonic, they should be supported by additional evidence.

I 5. Further study of plumose structures is desireable to increase confidence in criteria for distinguishing betwenn those of neotectonic and ship origin. I 6. Enigmatic features of several possible origins, including neotectonism, are found within the Quaternary sediment section. Several seismic shaking events about 12-13 ka and one such event about 6-8 ka could be indicated. Further detailed survey work is needed to better I determine the likelihood of paleoseismicity and faulting. 7. Alignments of pockmarks are visible in a limited swath bathymétrie image of the SOSZ. Similar rapid mapping over wide areas of the lakebed, including areas over suspected I basement structures, is desireable to reveal the presence of anomalous surface features, and to better focus and optimize time-consuming followup surveys.

1 ACKNOWLEDGEMENTS

This paper is drawn from joint work with several colleagues including S.M. Blasco, G.D.M. I Cameron, A. Godin, E.L. King, L.A. Mayer, K.M. Moran, R. Pippert, A.G. Sherin and B.J. Todd. I I I I -C44- I REFERENCES I

Cameron, G.D.M. and Lewis, CF.M Lewis. Linear acoustic backscattering anomalies in Lake — Ontario: Their anthropogenic origins. Abstract. Geological Association of Canada I meeting, Waterloo, May 18,1994. Forsyth, D.A., Milkereit, B. Zelt, C.A., White, D.J., Easton, R.M. and Hutchinson, D.R. 1994. Deep structure beneath Lake Ontario: crustal-scale Grenville subdivisions. Canadian I Journal of Earth Sciences 31: 255-270. • Lewis, C.F.M., Sherin, A.G., Atkinson, A.S., Harmes, R.H., Jodrey, F.D., Nielsen, J.A., Parrott, D.R., and Todd, B.J. 1992. Report of Cruise 92-800, CCGS Griffon, Lake Ontario. I Geological Survey of Canada, Open File Report 2678, 69 p. • Lewis, C.F.M., Cameron, G.D.M., King, E.L., Todd, B.J. and Blasco, S.M. in press. Structural contour, isopach and feature maps of Quaternary sediments in western Lake Ontario. B Atomic Energy Control Board Research Report, 70 p., 46 figs., Appendix of core logs, 7 • p., 12 maps, 1:250.000 scale. Mayer, L.A., Lewis, C.F.M., Godin, A. and Sherin, A.G. 1994. Swath bathymétrie mapping in two • Lake Ontario areas: Evidence for lakebed drumlins and ship traffic debris. Abstract. I Program, Geological Association of Canada, Waterloo meeting, May 18,1994. Sanford, B.V., and Baer, A.J. 1981. Geology - southern Ontario, Sheet 30S. Geological • Survey of Canada, Map 1335A, Scale 1:1 000 000. | Shaw, J. and Gilbert, R. 1990. Evidence for large-scale subglacial meltwater floods in southern Ontario and northern New York State. Geology, 18:1169-1172. • Shaw, J. and Sharpe, D.R. 1987. Drumlin formation by subglacial meltwater erosion. Canadian | Journal of Earth Sciences 24: 2316-2322. Thomas, R.L., McMillan, R.K., Keyes, D.L., and Mohajer, A.A. 1989a. Surface sedimentary m signatures of neotectonism as determined by sidescan sonar in western Lake Ontario. | In Program with Abstracts, Geological Association of Canada, Annual Meeting, Montréal, Québec, 14: A128. _ Thomas, R.L., McMillan, R.K. and Keyes, D. 1989b. Acoustic surveys: implications to the • geoscience discipline. Lighthouse, Journal of the Canadian Hydrographie Association, 40:37-42. - Thomas, R.L., McMillan, R.K., and Keyes, D. 1991. Evidence of neotectonic activity in Lake I Ontario. In Program with Abstracts, Geological Association of Canada, Annual Meeting, - Toronto, Ontario, 16: A124. _ Thomas, R.L., Wallach, J.L., McMillan, R.K., Bowlby, J.R. Frape, Keyes, D. and Mohajer, A.A. I 1993. Recent deformation in the bottom sediments of western and southeastern Lake ™ Ontario and its association with major structures and seismicity. Géographie, physique et Quaternaire, 47: 325-335. I Wallach, J.L. 1990. Newly discovered geological features and their potential impact on ™ Darlington and Pickering. Atomic Energy Control Board Report, INFO 0342, 20 pp. Wallach, J.L., and Mohajer, A.A. 1990. Integrated geoscientific data relevant to assessing I seismic hazard in the vicinity of the Darlington and Pickering nuclear power plants. In • Proceedings Canadian Geotechnical Conference, October 1990, Québec City, pp. 679- 686. 9 Wallach, J.L., Mohajer, A.A., McFall, G.H., Bowlby, J.R., Pearce, M. and McKay, D.A. 1993. • Pop-ups as geological indicators of earthquake-prone areas in intraplate eastern North America. Quaternary Proceedings No. 3, pp. 67-83. • I Lake Ontario

CCGS Griffon 1992 survey lines

NSC F.G. Creed Simrad EM 1000 Mosaic Locations (19 9 3)

'•-'•' J

Merco tor 550000 o t 43.67 N

Figure 1. Map of Lake Ontario showing trackiines of the 1992 grid survey using seismic reflection profiling and sidescan sonar, and two areas mapped by swath bathymetry in 1993. P = Pickering, D = Darlington. For scale refer to the 10 km east-west trackline spacing or the 5 km north-south trackline spacing. -C46- I -s' » ••' \ » Ricgss (from s=;Ernic profile data) / Ridges (from sidescan data) I -' Area of numerous bedrock ridges -60- Bedrock structural contour (depth in millisec) I I I I I * . ' . "BO- " / I

Figure 2. Map of western Lake Ontario showing the distribution of bedrock ridges interpreted as pop-ups in the 1992 grid sur/ey and the geophysical lineaments. Multiply millisecond values by 0.75 for an estimate of depth in metres below lake I level. From Lewis et al., in press. I I I I I I I I I Figure 3 Map of western Lake Ontario showing the distribution of bedrock scarps, relief and structure, and the geophysical lineaments. Bedrock age and lithology from Sanford and Baer, 1981. From Lewis et al., in press. I I -C47-

LABAs (not to scale) I .-- DISCONTINUOUS ^ CONTINUOUS I EHJ NUMEROUS I I I I I

I Figure 4. Map of western Lake Ontario showing the distribution and orientation of linear acoustic backscatter anomalies (LABAs) detected in the 1992 grid survey. From Lewis et al., in press. I i 78" 30' W .

Depression I Plumose structure I I I r/4fy I I I Figure 5. Map of western Lake Ontario showing the distribution of sediment depressions and plumose structures detected in the 1992 grid survey. From Lewis et al., in press. I I Normal fault Furrows I Extensive reflection eradicalior C-3 Notches I I I I I I

Figure 6. Map of western Lake Ontario showing the distribution of notches (6-8 ka), reflection eradications (REDs)(12-13 ka), small normal faults and sedimentary furrows detected in the 1992 grid survey. From Lewis et al., in press. I I I I I I I I Alignment of pockmarks? Alignment of I pockmarks?

Figcrç 7 Map of swath bathymetry in eastern Lake Ontario clearly showing a prominent field of sub-parallel narrow ridges in a pan of the SOSZ. See Figure 1 for location. Dimensions of the parallelogram swath area are 10.7 x 21.5 km. I Pockmarks, interpreted as sites of fluid venting at the lakebed, are outlined in white. Note the cross-cutting alignments of pockmarks in the southeastern section, suggesting fluid flow along fractures. A faint topographic lineament trending north- northwest is traced in white in the northwestern section of the swath area. I I -C49-

Geophysical signatures beneath Lake Ontario in the vicinity of the Darlington Nuclear Generating Station; seismic hazard I implications I J.L. Wallach A study by Ontario Hydro (1989) revealed a linear break in slope in both bedrock and the overlying unconsolidated sediments beneath Lake Ontario (Figures 1 & 2), about 2.5 km I from the Darlington Nuclear Generating Station. This break is oriented approximately 050°. Another(?) linear zone, trending about 035°, is seen on a side-scan sonar mosaic and separates distinctly different patterns (Figure 3). High-resolution, digital aeromagnetic 1 data show a linear aeromagnetic high in the vicinity of Darlington (Figure 4). The northwest boundary of that linear positive anomaly, which is parallel to the break in slope I observed on the lake bottom, and oriented about 050°, was interpreted by Paterson, Grant and Watson as a fault (Figure 4, Interpreted Fault #1). Inasmuch as that linear feature is interpreted from magnetic data, and the limestones, shales and overlying unconsolidated I sediments are essentially ferromaguesian-poor, it implies that the fault is present in the Precambrian basement.

I The break in slope (linear ridge) in both the bedrock and the overlying sediments, as well as the distinct difference in side-scan pattern on opposite sides of a linear band, suggest the possible presence of faulting on the lake bed about 2.5 km from the Darlington NGS. I The fault, inferred from the magnetic data and parallel to the break in slope, is not coincident with the lineament imaged on the bedrock or bathymétrie contour maps. That implies that there are either two distinct parallel lineaments, which probably form part of a I zone, or that the break in slope may be controlled by one lineament which dips toward the northwest, and extends downward into the Precambrian basement rocks. The linear ridge, defined by the structural contours on the bedrock surface (Figure 1), is sharper than that I on the bathymétrie surface (Figure 2) suggesting that the linear ridge existed prior to deposition of the sediments deposited during, and following, the last glacial episode about I 12,000 years ago. To resolve whether or not the linear band and break in slope signify faulting, a thorough I investigation must be undertaken. If faulting is the cause, then it must be determined whether or not the faults are active and capable of generating a strong, near-field earthquake. Until this is done, the interpretation that the linear features are faults which I have moved geologically recently, and are capable of moving again and producing a rather sizeable earthquake, cannot be dismissed.

I During the summer of 1994, a series of four manned dives was made in the vicinity of the lineament off Darlington to attempt to identify its nature. Unfortunately, murky waters prevented drawing any conclusions, other than to say that there is a bathymetrically I expressed lineament, oriented about O3O°-O5O°. At the base of the lineament the rock is I limestone, presumably from the Lindsay Formation, whereas at the top a darkly colored I I -C50- limestone, which might represent the overlying Whitby Formation, was recovered. If those formations were correctly identified, it indicates a normal stratigraphie sequence, but that _ does not, a priori, rule out the possibility of a fault. I

The preliminary stratigraphie information, obtained from the dives, suggests that there m may be yet another fault, or even a fault zone. The bedrock strata in the area display a | southerly regional dip of 5 m/km (Ontario Hydro, 1981) and, at the Darlington site, the Lindsay-Whitby contact is at an elevation of 81 m above sea level. According to Ontario • Hydro maps, the elevations that they report are equal to the Canadian Geodetic Datum fl elevations, plus 22 m. With that condition, the true elevation of the Lindsay-Whitby contact, in the Darlington excavation, would be 59 m above sea level. Using the regional B dip of 5 m/km the contact, at a distance of 2.5 km southeast of the site, should be at 59- • 12.5 m or 46.5 m above sea level.

During the dives, the dark brown to black argillaceous limestone was obtained at a water ™ depth of 19.2 m, equivalent to an elevation of 55.80 m above sea level, and the Lindsay was sampled at a water depth of about 23 m, or 52 m above sea level. If it is assumed that I the exposed Lindsay is at the top of the unit, then the contact would also be at 52 m above ™ seal level, or about 5.5 m higher than it expected with a uniform regional dip. The situation is somewhat confusing, however, because the topographic map of the area I (30M/15) shows the elevation at the Darlington site to be between about 84 and 76 m, which means that the elevations given in the Ontario Hydro reports, for the Darlington a site, are equal to the Canadian Geodetic Datum. Again, assuming that the regional dip is I uniform then, at a southerly distance of 2.5 km from the nuclear station, the base of the Whitby should occur at an elevation of 81-12.5=68.5 m above sea level. As in the previous « example, this elevation is not in agreement with that predicted if the strata are inclined, | uniformly, but in this case the Lindsay-Whitby contact would be about 13 m below the predicted contact. Because the work to date is preliminary, it is not yet possible to • ascertain the reason for the apparently anomalous elevation of the Lindsay-Whitby contact I in the vicinity of the break in slope.

Inclinometers on board the submersible, though not functioning perfectly, did indicate a • gentle, yet noticeable dip of bedrock strata toward the south, consistent with the findings of Ontario Hydro. An explanation commonly given for this phenomenon is deposition of H sediments on an irregular basement surface. It must be pointed out, however, that the • depositional mechanism may be overworked. An examination of bedrock outcrops near Kingston, and Lyn, Ontario, respectively reveal outcrop-scale evidence that either • deposition on boulders, or sedimentation in a channel, need have no profound effect on the attitude of the overlying sediments. At Kingston, limy mud of the Gull River Formation _ was deposited on rounded granitic cobbles, derived from the Precambrian basement • (Figure 5). The effects of the topographic irregularity, represented by the granitic clasts, is visible in the limestone but, in several cases, they die out upward over a very short vertical « distance, such as beneath the hammer in Figure 5. The same applies to the deposition of | sand on Precambrian quartzite clasts in the Nepean Formation at Lyn and calcareous sand deposited in a channel in the March (?) Formation, just west of Brockville. None of this is • I I -C51-

I to imply that topographic irregularities will not result in the inclination of overlying strata. Rather it is to point out that alternative explanations are possible, and that no firm I conclusions should be drawn until the necessary studies are carried out. Despite the dives, it is still unclear whether or not the break in slope represents a fault, or an artifact of something else. Because of the seismic hazard implications, work should be I conducted to: a) verify the stratigraphy, b) determine whether or not the regional dip is uniform, or variable, and c) ascertain the true nature of the offshore lineaments. If faulting is the cause, then it must be determined whether or not the faults are active, and capable of I generating a strong, near-field earthquake. Until that is done, the interpretation that the linear feature is a fault which moved geologically recently, and is capable of moving again I and producing a rather sizeable earthquake, cannot be dismissed. If the 3 km long linear feature, in the immediate vicinity of the Darlington Nuclear I Generating Station, is a fault then, from empirical seismological data, it could produce an earthquake of up to M*6.5 (Figure 6). The results from all three attenuation relation equations show that earthquakes of magnitude 6.0 and 6.5, at a distance of 20 km, will I produce much greater peak accelerations than the similar magnitude earthquakes used to determine the Design Basis Earthquake (Table 1; shown in boldface). Though calculations can be made for distances less than 20 km, the resulting values are apparently unreliable I (Hasegawa, personal communication).

In order to determine the Design Basis Seismic Ground Motion for the Darlington Nuclear I Generating Station A, Basham (1975) calculated peak accelerations for earthquakes ranging in magnitude from 5.00 to 6.70 at distances of 20 to 200 km. Included in those calculations were peak accelerations for M=6.0 and M=6.50 earthquakes at distances of I 70 km and 110 km respectively. Identical magnitude-distance relations were employed to calculate peak accelerations using the equations of Hasegawa, Basham and Berry (1981) and Boore and Atkinson (1987). The results of all three equations are compared in Table 1 I and show that, except for the more distant earthquakes, the peak accelerations determined by Basham in 1975 are less than those calculated from the two subsequent equations for I the same magnitude-distance relations. It is clear that a set of linear features exists in the immediate vicinity of the Darlington I Nuclear Generating Station. One possible explanation is the presence of a family of faults. Because of their proximity to the nuclear power plant, that possibility cannot be dismissed I until a carefully conceived geoscientific investigation is carried out. It is beyond the scope of this paper, and the knowledge of the writer, to address the topic of response of any engineered structure to a near-field M=6.5 earthquake. The point is I simply to show that conditions favorable for such an earthquake may exist near the Darlington station and, if so then, irrespective of the equation used, peak accelerations would be far in excess of those considered in the DBSGM. Even though parameters such I as frequency and duration have not been addressed, peak accelerations well in excess of I the DBSGM must be a cause for some concern. I -C52- I

TABLE 1 I

CALCULATIONS OF PEAK GROUND ACCELERATION |

Using the equation of Hasegawa, Basham and Berry, 1981 : •

Magnitude Distance (km) Accel. (%g)

6.70 200 6 I 6.50 110 9 6.50 20 60 • 6.00 70 8 1 6.00 20 31 5.00 20 9 B

Using the equation of Boore, D.M. and Atkinson, G.M., 1987: I

Magnitude Distance (km) Accel. (%g) H

6.70 200 1 6.50 110 3 - 6.50 20 37 I 6.00 70 5 6.00 20 25 m 5.00 20 10 1

Using the equation of Milne (personal communication to Basham, 1975): I Magnitude Distance (km) Accel. (%g) I

6.70 200 2 6.50 110 4 6.50 20 25 I 6.00 70 4 6.00 20 16 5.00 20 6 I I I I I I -C53-

I Selected References

I Basham, P.W., 1975. Design basis seismic ground motion for Darlington Nuclear Generating Station A. Seismological Service of Canada, Internal Report 75-16, Division of Seismology and Geothermal Studies, Earth Physics Branch,

• Department of Energy, Mines and Resources, 34 pp. plus tables and figures. Boore, D.M. and Atkinson, G.M., 1987. Stochastic prediction of ground motion and I spectral response parameters, at hard-rock sites in eastern North America. Bulletin of the Seismological Society of America, v. 77, pp. 440-467.

I Hasegawa, H.S., Basham, P.W., and Berry, M.J., 1981. Attenuation relations for strong seismic ground motion in Canada. Bulletin of the Seismological Society of • America, v. 71, pp. 1943-1962.

Liberty, B.A., 1969. Paleozoic geology of the Lake Simcoe Area, Ontario. Geological • Survey of Canada Memoir 355, 201 pp.

Ontario Hydro, 1981. Darlington G.S. structural geology. Design and Development • Division, Report No. 81406.

Ontario Hydro, 1989. Darlington NGS 1989 high resolution marine geophysical survey. • Design and Development Division-Generation, Report No. 90108.

Slemmons, D.B., 1977. Faults and earthquake magnitude. State-of-the-Art For Assessing Earthquake Hazards in the United States, Report 6, Miscellaneous Paper S-73-1. I Prepared for Office, Chief of Engineers, U.S. Army, Washington, D.C., 129 pp. g plus appendix. I I I I I I I I -C54-

List of Figures

Figure 1. Contours on bedrock surface beneath Lake Ontario showing an ENE- £ trending linear break in slope. Site survey area offshore of the Darlington Nuclear Generating Station. (From OH Report 90108, Figure 15) •

Figure 2. Contours on the bathymétrie surface beneath Lake Ontario showing an ENE-trending linear break in slope. Site survey area offshore of the • Darlington Nuclear Generating Station (From OH Report 90108, Figure 9) I

Figure 3 Side-scan sonar mosaic of the bottom of Lake Ontario offshore of the M Darlington Nuclear Generating Station. Note that the light-colored, NNE- • trending linear band which passes between the "L" in "INFILL" and the number "20" separates distinctly different patterns. (From OH Report B 90108, Figure 16) •

Figure 4 Geophysical interpretation of aeromagnetic data showing linear magnetic I highs and interpreted faults. One of the interpreted faults lies just offshore of Darlington and is sub-parallel to the linear features seen in the preceding — figures. (From Paterson, Grant and Watson) I

Figure 5 Cross section showing the effect of rounded cobbles, representing an g irregular basement topography, on the overlying, formerly limy muds in the jjj Gull River Limestone. Note that the effect of the clasts on the limestone layers beneath, and to the right of, the hammer, is negligible. On the right a side of the photo, the effect is far more pronounced. |

Figure 6 "Relation of earthquake magnitude to length of the zone of surface • faulting" (From Slemmons, 1977, Figure 1) I I 1 I I 1 I Darlington Generating Station

Figure 1, Contours or* tadrock Surf act himA Lak Ontario offsW. à Darlington Generating Station

-m* ^^n

-Ç5.8-

rtt

LAKE ONTARIO

LEGEND Darlington Nuclear Generating Station

Magnetic High Scale 1:66,600 Interpreted Fault Interpretation by Patterson, Grant & Watson

Figure 4. Linear magnetic anomalies, one of which is interpreted as a fault parallel to the linear breaks in slope seen in Figures 1 & 2. I -C59- I I I I I I I I I I I I I 8

1 Figure 5 Cross section showing the effect of rounded cobbles, representing an irregular basement topography, on the overlying, formerly limy muds in the Gull River Limestone. Note that the effect of the clasts on the limestone I layers beneath, and to the right of, the hammer, is negligible. On the right side of the photo, the effect is far more pronounced. I B 10 -C60- I I 1000 1 5001— I I I I I I I I I I i I

23456789 I EARTHQUAKE MAGNITUDE

Figure 6. Relation of earthquake magnitude to length of the zone of surface faulting I (Modified from Slemmons, 1977) References on curves given in Slemmons, 1977 I I I -col-

I Faults in the Metropolitan Toronto Area; Active vs Dormant Faults I A.A. Mohajer I University of Toronto / Seismican i Background History of earthqfetiv, in a region is usually used as the guide to assess future potential seismic hazard. In .uoas of sparse or low seismic activity, such as southern Ontario with I a short history of less than three centuries of settlement, lack of destructive earthquakes in the past does not necessarily rule out the possibility of future big events. Earthquake occurrences in historical and/or geologically recent times signify the presence of a fault I plane or zone with a preferred orientation on which strain energy is repeatedly released under the ambient crustal stress. It is, therefore, crucial to search for paleoseismic evidence, to delineate exposed or subsurface faults, and to investigate their history of reactivation for I assessment of the potential future earthquakes.

An important question before the residents of Metro Toronto and for the responsible I government agencies is whether or not there is a fault under or near the Pickering nuclear power plant, and whether or not it is active. To answer these questions properly new site and feature-specific data are needed. A long term seismic program has, therefore, initiated I by the AECB since 1991, which involves detailed inland and marine fault investigation and site- specific microearthquake monitoring. The aim of the current work is to generate sufficient new data to help re-evaluation of the design basis seismic ground motion for the I nuclear power plants at Pickering and Darlington. I Guidelines, Standards and Definitions I There are no established guidelines in Canada for discriminating between active and dormant faults other than a set of recommendations in the National Standard of Canada CAN3-N289.2-M81, 1981 (clause # 3.3.2.1) regarding ground motion determination for I seismic qualification of Candu nuclear power plants. This standard requires regional investigations to provide information on: " the neotectonics of the region, with particular attention being given to the structures that may be revealed in Quaternary deposits, to I structures in older rocks that may be post-glacial in age, and to structures that have a spatial correlation with known earthquake activities". In the United States, the US Nuclear Regulatory Commission (1982) defines an active fault as the one which has experienced a I movement, at or near the surface, at least once in the last 35,000 years, or more than once in the last 500,000 years. There are many examples of major active faults, however, with I return periods that are much longer than those prescribed in the United States (Table-1). I I I -C62-

Fault movements recur episodically throughout time, in response to the build up of stress. • Thus, in order to show that a fault is dormant (inactive), it must be shown either that the stress no longer exists, or that the fault has not moved for a long period of time, at least I since the onset of the current stress field. Failing either, it is reasonable to assume that all • faults, in an area where seismicity has been documented, have the potential to be reactivated (Muir Wood and Mallard, 1992). In addition, direct evidence for or against surface faulting I displacement may have been removed by erosion or buried by recent sediments. Repeated glacial erosion and sedimentation has masked much of the surficial evidence. It is therefore _ necessary to define a third category of faults where knowledge of the activity remains I uncertain (Table 2). This category of faults include a wide range of indicators some of which may put more weight on the side of active or dormant fault status. The possibility _ for reactivation of these faults and their potential to generate future earthquakes can not 9 be ignored, unless and until they can be proved dormant on the basis of new data through further local investigations. g

Major Crustal Discontinuities in the Western Lake Ontario Region •

The presence of major discontinuities in the crustal rocks of southern Ontario are suggested by the recognition of linear aeromagnetic anomalies (Forsyth et al., 1987) and together with Oj gravity anomalies are defined as the Niagara-Pickering linear zone (NPLZ), and the I Georgian Bay linear zone (Wallach and Mohajer, 1990). A prominent regional terrane boundary, known as the Central Metasedimentary Belt Boundary Zone, CMBBZ (Hanmer, • 1988) coincides with the first linear anomaly which passes virtually under the nuclear I facilities at Pickering (Figure 1). This geological boundary was described by Hanmer (1988) as a major east dipping ductile shear zone formed under pressure and temperature of the fl mid-crustal depth, now exposed at northern limit of the Paleozoic cover about 100 km • north of Pickering. The existence of the faulted Precambrian basement rocks is confirmed by Ontario Geological Survey's core samples from deep wells (Carter and Easton, 1990), and È defined as a fault line on the geological map of Ontario (Ontario Geological Survey, 1991). • Deep seismic reflection data from the oil exploration surveys in Lake Ontario and Lake Erie provided further evidence of reactivation of basement structures in the form of listric and I planar faults dipping 15° to 30° to the east, which are traceable to the depth of at least 22 * km (Milkereit, et al., 1992). Several possible tectonic development models were proposed by Milkereit et al., (1992) to explain reactivation of this structure during Grenvillian I orogeny which involves lateral growth by accretion and post-orogenic collapses. The alternative models they have proposed include subduction, collision, multiple thrusting, and _ subsequent formation of a series of half-grabens which may be the result of extension during I pre-Appalachian Iapetan rifting.

Wheeler (1995) modelled lapetan rift structures as an active seismic source zones in eastern North America. He put the western limit of the Iapetan rift along the Clarendon-Linden fault system, which runs sub-parallel to the CMBBZ and shows similar aeromagnetic signature, at about 100 km to the east. If the CMBBZ can possibly be interpreted as an I I I I -C63-

lapetan structure as suggested by Milkereit et al., (1992) then it can be considered as a prominent seismogenic source. I The present data base can not resolve the processes involved with tectonic development of the CMBBZ, and needs extensive local investigations in consideration of its very important role in assessment of the nuclear site safety. Particularly, where there are many I documented indications of extensive faulting and fracturing in the Paleozoic cover strata (Sanford, 1985, 1993; Mohajer et al., 1992, 1995; Eyles et al., 1993; Wallach, this proceedings) which can be evidence of continued reactivation of basement structures to the I present geological times.

I Recent Faulting in the Greater Toronto Area

Possible evidence of reactivation is found along the Rouge River valley where several faults I are exposed only 7 km from the nuclear power plant at Pickering (Mohajer et al., 1992). These faults are possibly the result of reactivation of Precambrian basement faults and their upward propagation through the Paleozoic strata and into Pleistocene cover deposits. A I series of normal faults cut and displace soft Pleistocene sediments, and the underlying bedrock (Figure 2). The maximum displacement observed was 1.2 m. of vertical offset. This was interpreted by Mohajer et al., (1992) as a consequence of crustal tectonics at the I presence of glacial rebound. These faults were also briefly examined by Adams et al (1993 a, b). Several possible explanations for the origin of the normal faulting, were suggested but i the preferred hypothesis was glacial ice shove. Field investigation by Wallach on behalf of the AECB revealed that there are at least two periods of normal faulting and that the faults could not be explained by simple glacial processes (Wallach's internal AECB report, 1993, I personal communication). The age of the faulting is poorly constrained; structures are no older than 40,000 years and may postdate 13,000 years.

• Additional faulting was recognized, upstream, in the vicinity of the Metro Toronto Zoo (Hibbert et al., 1993). An apparent throw of nearly 4 m is indicated in this newly found section (Figure 3). Multiple normal faults at section # 13, in addition to evidence of • slumping and two different resedimentation texture found across the faults, would indicate repeated episodes of deformation during deposition of both the Sunnybrook Till (faults A, B B, C) and the overlying Thorncliffe Formation (fault D). I An Image of the Sub-surface Faulting More than 1.5 km of high-resolution shallow seismic reflection (SSR) data were collected along four variously oriented profiles within the Rouge River valley (Mohajer et al., 1995). I Seismic profiles were acquired across the projected strike of fault zones mapped in I Pleistocene deposits and bedrock outcrops exposed along the Rouge River.

I I -C64-

The geometry of bedrock reflectors demonstrates a low relief (< 5 m), gently undulating B bedrock surface which, at some localities (e.g. site A), shows a zone of discontinuous and offset reflectors (Figure 4). Across site A, a 220 m profile was acquired on the table lands • immediately adjacent to a faulted Quaternary outcrop on the eastern bank of the Rouge B River (outcrop #13; Figure 3). The profile is oriented at 350° across the projected strike (120°) of the main fault mapped at outcrop #13. At this site, the bedrock reflector is offset H by about 12 msec, indicating a maximum vertical displacement of the bedrock surface of 15 • m. The offset of Quaternary strata is approximately 4 m near the surface, which may indicate an upwards decrease in displacement along the fault. If this situation prevails in the B region, it suggests a tectonic origin for the faults, as opposed to the postulated glacial ice B shove mechanism proposed by Adams et al (1993a; 1993b).

This information necessitates a detailed study of the fault displacements at the bedrock and • Quaternary interface and near the surface across the CMBBZ, at least at two localities, north of Pickering and south of Lake Scogug, using high-resolution seismic reflection and I ground magnetic surveys. ™ I I I I I I I I I I I I I -C65- I I Table 1 - Durations of recurrence intervals for displacement events on active Faults I Slip-rates (years) References San Andreas (at Pallet Creek) 50-250 Sieh (1984) I North Anatolian Fault (Turkey) 100-300 Ambraseys (1970) Tabas-Doruneh (Eastern Iran) 300-900 Mohajer & Nowroozi (1979)

I Irpinia (Italian Apennines) 1700 Boschiet al (1990)

Wasatch Fault (Basin and Range, 2000 Schwartz & Coppersmith (1989) I Western U.S.) Charleston, Carolina est. 3000 Obermeieret al (1987) I -three events in Holocene Lost River Fault (Idaho, U.S.) 6000-8000 Hanks & Schwartz (1987)

I Erft Sprung System (cologne, est. 40000 Ahorner(1962) north of Alp) I Rio Grande Rift System avg. 100,000 Machette (1978) (Albuquerque, New Mexico) I Meers Fault (Oklahoma, U.S.) est. >1,000,000 Crone & Luza (1990) Reelfoot Fault, 1811-1812 avg 1,000,000 Russ(1979) (New Madrid, Missouri) episodic activity Sexton & Jones (1986) B 600-2000 ? Pratt (1994) Williams et al (1995) I I I I I I I State Of Dormant Uncertain Active Faulting (Extinct) (Unproven) Evidence Fault does not displace Fault is a small secondary Fault has appropriate dimensions preplio-Quaternary materials fracture and is uniquely implicated by well-located or structures large earthquake(s) Fault does not displace late The mineralogy of mechanically Quaternary materials or Fault coincides with accurately located continuous fault gouge is structures hypocentre(s) from local network and is incompatible with the current consistent with parameters from well- stress/temperate regime Fault style and orientation constrained focal mechanism(s) makes displacement unlikely in Current Tectonic Reaime Fault displaces ground surface or late Quaternary deposits Spatial association with small and/or structures ^ macroseismic earthquake or instrumental earthquake located by less accurate regional network

Fault has undergone multiple post-gfacial reactivation

Fault has a close analogue proved active Faults within seismotectonic province of moderate to high activity

Table - 2 Criteria for discrimination of dormant, active and uncertain (unproven) faults for the areas with moderate seismicity (modified after Mallard, 1992). CENTRAL GNEISS BELT CENTRAL METASEDIMENTARY BELT

Reported earthquake locations o of 20th century, 44° 00'- BOUNDARY ZONE O [I] Historical events

|> Buried Thrusts DARUNGTON O NGS

O 0 LAKE O ONTARIO

o i cO o

D o o N O A o o -43° 00" • o 20 40 60 O -I o °. i i_ i o 43° 00' 80° 00' 79° 00" kilometres I 78° 00' Figure 1 - Seismctectonic framework of the Western Lake Ontario region. I -C68-

SECTION 3 I NORTHEAST SOUTHWEST RECENT TERRACE GRAVELS I I I

0 5m I I Bedrock Don Formation Scarborough Formation I

SCHMIDT POLE CONCENTRATIONS I % of total per 1.0% area I w- I I N=250 N=23 I JOINTS IN BEDROCK FAULTS IN BEDROCK & QUATERNARY I SECTION 1 NORTH SOUTH I I 1

0 10m I Don Formation Scarborough Formation i Figure 2. Faulted strata at sections 1 and 3 along Rouge River. Stereonets (lower hemisphere) show I poles to first-order joints in bedrock and normal faults cutting bedrock and overlying Quaternary strata at section 3. Thick lines represent averaged planes; note correspondence between those of first-order joints and faults. I NW SE

MID-HOLOCENE (?) TERRACE GRAVELS & SANDS

LATE WISCONSIN TILL

RESEDIMENTED SUNNYBROOK UNIT 1 (RSU-1) UNIT 2 (RSU-2)

LOWER TERRACE n o°— a o oo — ^o

. SCARBOROUGH FORMATION

ROUGE RIVER

Figure 3 - Detailed stratigraphy and structural elements in soft Pleistocene sediments at Section #13, along the Rouge River Valley; northwest corner of the Metro Zoo. -C70- I METRO ZOO WEST I PROFILE A I CDP 75 125 175 375 I I JE. LU I 6 I I I N I

CDP 75 125 175 225 275 325 375 1 1 l 0 -, I I I PLEISTOCENE SEDIMENTS I

m •- •" ~" ™" *" " 20 - .. \ I 40 - WHOBY SHALE * ^ —— -w

•t < 60 - * I X * t 80 - ai - Q I 100 ~

* I 120 "

140 ~ • - I t 160- 1 I

15 30 45 I I I metres Figure 4 - Shallow seismic reflection profile along Rouge River Fault at the northwest corner of the Metro Zoo. I I -C71-

I REFERENCES I Adams, J. and Bashain, P.W., 1991. The seismicity and seismotectonics of eastern Canada; Chapter 14 in Neotectonics of North America (ed.) D.B. Slemmons, E.R. Engdahl, M.D. Zoback, and D.D. Blackwell; Geological Society of America, Decade Map Volume 1 : p. B 261-276. Adams, J., L. Dredge, C. Fenton, D.R. Grant, W.W. Shilts, 1993a. Comment on "Neotectonic faulting in Metropolitan Toronto: implications for earthquake hazard assessment in the I Lake Ontario region, by Mohajer et al., 1992": GEOLOGY, September 1993, p. 863.

Adams, J., L. Dredge, C. Fenton, D.R. Grant, W.W. Shilts, 1993b. Late Quaternary faulting in I the Rouge River Valley, southern Ontario: Seismotectonics or glaciotectonics? Geological Survey of Canada Open File 2652, 58p.

I Ahorner, L., 1962. Intesuchungen zur quartaren Bruchtektonik der Niederrheinischen Bucht. Eiszeitalter und Gegenwart, 13, 24-105.

I Ambraseys, N.N., 1970. Some characteristic features of the Anatolian Fault Zone. I Tectonophysics, 9, 143-165. Atkinson, G., 1990. Updated seismic hazard estimates at Ontario Hydro sites. Report No. I 90275, Ontario Hydro, Toronto, Ontario. Atkinson, G. and Stagg, M., 1987. Seismic Hazard at Ontario Hydro Dam and Plant Sites. I Report No. 87337, Ontario Hydro, Toronto, Ontario. Basham, P.W., 1976. Design basis Seismic ground motion for Darlington Nuclear Generating I Station A. Report. 75-16, Department of Energy Mines and Resources, Ottawa, Ontario. Basham, P.W., 1989. A Paleozoic-Mesozoic rift framework for seismic hazard assessment in eastern North America; in Current Research, Part F; Geological Survey of Canada, Paper . 89-1F, p. 45-50.

Basham, P.F. and Adams, J., 1989. Problems of seismic hazard estimation in regions with few I large earthquakes: examples from eastern Canada; Tectonophysics, v. 167, p. 187-199. I Basham, P.W., Weichert, D.H., Anglin, F.M. and Berry, M.J., 1985. New probabilistic strong seismic ground motion maps of Canada; Bulletin of the Seismological Society of America, I v. 75, p. 563-595. Boschi, E., Pantosti, D. and Valensise, G., 1990. Paradoxes of Italian seismicity. EOS, I November 13th 1990, 1787-1788. I I -C72- I

Carter, T.R., and M.S. Easton, 1990. Extension of Grenville basement beneath southwetern I Ontario: Lithology and tectonic subdivisions. In Carter, T.R.,ed., Subsurface geology of southwestern Ontario: A core workshop: London Ontario Petroleum Institute, p. 9-28.

Crone, A.J. and Luza, K.V., 1990. Style and timing of Holocene surface faulting on the Meers • Fault, southwestern Oklahoma, Geological Society of America Bulletin, 102, 1-17. fl

CSA CAN3-N289.2-M81, 1981 (reaffirmed 1992 without change). Ground Motion • determination for seismic qualification of CANDU Nuclear Power Plants. Canadian • Standards Association, Rexdale, Ontario. I Eyles, N., J. Boyce, and A. A. Mohajer, 1993. The bedrock surface of the western Lake Ontario • region: Evidence of reactivated basement structures?, Géographie Physique et Quaternaire, v. 47, n. 3, pp. 269-283. I Hanmer, S., 1988. Ductile thrusting at mid-crustal level, southwestern Grenvile Province: Canadaian Journal of Earth Sciences., V. 25, pp 1049-1059. I

Hanks, T.C. and Schwartz, D.P., 1987. Morphologic dating of the pre-1983 fault scarp on the — Lost River Fault at Doublespring Pass Road, Custer Country, Idaho. Bulletin of the • Seismologcial Society of America, 77, 837-846.

Hibbert, J, A. A. Mohajer, N. Eyles, 1994. Faulting in unconsolidated sediments and bedrock east | of Toronto. Atomic Energy Control Board Report No. 2.263.2., 50p.

Jackson, J., 1980. Reactiviation of basement faults and crustal shortening in orogenic belts. | Nature, 283, 343-346.

Johnston, A.C., 1989. The seismicity of'Stable Continental Interactions'; in Earthquakes at North • Atlantic Passive Margins: Neotectonics and Postglacial Rebound, (ed.) S. Gregersen and P.W. Basham; Kluwer Academic Publishers Dordrecht, p. 299-327. •

Machette, M.N., 1978. Dating Quaternary faults in the southwestern United States by using calcic paleosols. US Geological Survey Journal of Research, 6, 369-381. I

Mallard, D. J., 1991. Learning to cope with faults. Proceeding of the International Conference on Seismic Hazard Determination in Areas with Moderate Seismicity. Saint-Remy-les- I Chevreuse, France, October 22-23, PP 111-122.

Milkereit, D.A. Forsyth, A.G. Green, A. Davidson, S. Hanmer, D.R. Hatchinson, W.J. Hinze, • R.F. Mereu, 1992. Seismic images of a Grenvillian terrane boundary. Geology, V. 20, N. 11, pp 1027-1030. m I I I -C73-

I Mohajer, A.A., Boyce, J., N. Eyles, 1995. Subsurface characterization of the Rouge River Quaternary faults, using high-resolution shallow-seismic reflection profiles. Atomic I Energy Control Board, Project No. 2.263.3. I Mohajer, A.A., 1987. Relocation of earthquakes and delineation of seismic trends in Southern Ontario, prepared for Ontario Hydro, also presented to the Geological Association of Canada, annual meeting, St John's, Nfld, May 23-25, 1988. I Muir Wood, R. and D.J. Mallard, 1992. When is a fault 'extinct1? J. Geol. Soc, V. 149, PP. 251- 254. Ontario Geological Survey, 1991. Bedrock geology of Onatrio, southern sheet. Ontario I Geological Survey, Map 2544, Scale 1:1,000,000. I Obermeier, S.F., Weems, R.E. and Jacobson, R.B., 1987. Earthquake-induced liquefaction features in the coastal South Carolina region, in Proceedings of the Symposium on I Seismic Hazards in Eastern North America, Oct. 29-22, State University of N.Y., Buffalo. Technical Report NCEER 87-0025, 480-493. Russ, D.P., 1979. Late Holocene faulting and earthquake recurrence in the Reelfoot Lake area, I northwest Tennessee, Geological Society of America Bulletin, 90, 1013-1018. Sanford, B.V. 1993. Stratigraphie and structural framework of upper Middle Ordovician rocks in I the Head Lake-Burleigh Falls area of south-central Ontario. In Wallach and Heginbottom, J.A., (eds.) Neotectonics of the Great Lakes Area, Géographie Physique et Quaternaire, I V. 47, pp253-268. Sanford, B.V., Thmpson,F..T., and McFall, G.H., 1985. Plate tectonics; a possible controlling mechanism in the development of the hydrocarbon traps in southwestern Ontario. I Bulletin of Canadian Petroleum Geology, V.33, pp52-71.

Schwartz, D.P. and Coppersemith, K.J., 1984. Fault behaviour and characteristic earthquakes: I examples from the Wasatch and San Andreas Faults. Journal of Geophysical Research, 89, 5681-5698.

1 Sexton, J.L. and Jones, P.B., 1986. Evidence for recurrent faulting in the New Madrid seismic I zone from Mini-Sosie high resolution reflection data. Geophysics, 51, 1760-1788. Sieh, K.E., 1984. Lateral offsets and revised dates of large prehistoric earthquakes at Pallet I Creek, southern California. Journal of Geophysical Research, 89, 7641-7670. US Nuclear Regulatory Commission, 1982. Appendix A Seismic and geologic siting criteria for nuclear power plants. Code of Federal Regulations — Energy, Title 10, Chapter 1, Part I 100. (app.A,10, CFR 100), 1 Sept. 1982. I I -C74-

Wallach, J.L. and A.A. Mohajer, 1990. Integrated geoscientific data relevant to assessing seismic hazard in the vicinity of Darlington and Pickering nuclear power plants: Proceedings, m Canadian Geotechnical Conference, October 1990, Quebec City, P. 679-686. |

Wells, D.C. and K.J. Coppersmith, 1994. New empeirical relation among magnitude, rupture m length, rupture width, aipture area, and surface displacement, BSSA, V. 84, N. 4, PP. • 974-1002. • Wheeler, R.L., 1991. Earthquakes and the Iapetan passive margin in eastern North America; in • Proceedings, Geological Survey of Canada workshop on eastern Seismicity Source Zones for the 1995 Seismic Hazard Maps, (comp.) J. Adams; Geological Survey of Canada, H Open File 2437, p. 73-85 and 272-283. I

I 1 I I I I -C75- On the interpretation of the Rouge River Faulting, Metropolitan Toronto — I What would an active fault with an unknown return period contribute to seismic hazard assessment?

I John Adams National Earthquake Hazards Program, Geological Survey of Canada, I 1 Observatory Cres, OTTAWA, K1A 0Y3 Canada

In 1992 (and in abstracts presented at several meetings in the preceding few years) A. Mohajer I and colleagues published a description and interpretation of two outcrops in the Rouge River valley where normal faults displace both the bedrock surface and the lower strata of the overlying Pleistocene deposits. These authors claimed they had found "neotectonic" faulting and speculated I that the faults had "seismotectonic implications for metropolitan Toronto and the nearby nuclear facilities". Irrespective of the truth of these claims, a sensible scientific discussion of these faults was not helped by the scope and tone of the paper. To my mind, the published interpretation of I the faulting was open to dispute and the burden of proof required for such serious allegations was I not satisfied. It is certain that evidence of a seismogenic surface rupture will be found in Ontario. Calculations I have made with C. Fenton that are based on the worldwide rate of the larger earthquakes in I stable cratons suggest that of the order of 20-40 M>6 earthquakes might have occurred in Ontario during the 10,000 to 14,000 years since déglaciation; a number of these (perhaps 5 - 20, dependent on assumptions) would probably have broken the surface. This is not a large number, I as is consistent with the generally low level of seismicity. Indeed one might argue, given the very limited exposure of bedrock in the Toronto area, that unless Mohajer and colleagues were just incredibly lucky to find seismogenic faulting in the Rouge River, such faults must be I extremely common under the drift deposits of Toronto (a conclusion which conflicts with the current rate of seismicity).

I In eastern Canada there is a knowledge gap between the earthquakes, that chiefly happen 5-20 km underground, and the geologically observable surface. As shown by geological mapping, the Precambrian rocks of the Canadian shield are disrupted by many faults formed 1 to 2 billion I years.ago. Under southern Ontario similar faults must exist, but they are concealed by the Paleozoic limestones and shales, Quaternary sediments, and the waters of the lakes. Some fault or other in the Precambrian basement will be the origin of the next moderate earthquake in I southern Ontario, but it is impossible to map all faults in the top 25-35 km of crust. Detailed mapping of the surface faults is possible where the Precambrian rocks are exposed, and the I position and orientation of the largest faults can be projected down into the earth with some confidence. This is of limited value because i) it does not take a very large fault (a few km is enough) to cause a sizeable earthquake; ii) even where mapped at the surface, not one fault in I the Precambrian of Ontario has been identified as "active"; and iii) even if all the "active" faults were identified, there is no guarantee that a hitherto inactive fault will not move in the future. In fact, some of the larger earthquakes in California in the past 20 years have occurred on faults I that were not recognized until after the earthquake. I I e -C76- One measure of possible fault activity is a significant displacement (tens to hundreds of metres) _ in geologically recent times (last few hundred million years). Some southwest-striking faults B offset the Paleozoic bedrock in the region between Orillia and Kingston. Perhaps these faults will be seismically active in the future, but there has been no proof, pro or con. Still younger m faults might be found by searching the Quaternary strata for tectonically faulted strata, but as g noted above, the current evidence is equivocal. In the case of Paleozoic or Quaternary fault offset, if found in a particular region, there remains the troubling question: even if a few active m faults are found, as expected, how do we take into account the unknown active faults and the B potentially-active "inactive" faults? Unless geologists can identify the majority of future active seismogenic faults, placing the hazard emphasis on a few known sources misdirects attention • from the unknown sources. 9

A further problem will be using the results from the discovery of an active fault. How do we fl estimate the size of the past earthquake(s), particularly when the outcrop exposures will almost • certainly be extremely limited? How do we estimate the likelihood that the active fault will move again in the future? For example, if the displacement of the Quaternary beds on the Rouge fl River Faults represents 1) a single slip episode that 2) occurred coseismically on a tectonic fault • (these are contentious statements), then the best estimate of the return time is 125,000 years (the age of the oldest faulted Quaternary strata) with estimated limits of 60,000 years (an event I occurred just before the oldest strata were laid down, or will occur tomorrow) and infinity (this • is the only movement that will occur on this feature).

Instead of relying on incomplete or chance discoveries, robust estimates of earthquake occurrence can be made on seismological and seismological/geological grounds. _ ! For example, the current earthquake history of the region (ca. 150 years) can be assumed to represent what is to be expected over the next hundred years. We then define a region large m enough to give a statistically meaningful sample while staying within a region geologically £ similar. Then we use a relationship between large and small earthquakes to estimate the rates of future large earthquakes and represent the results as occurrence probabilities (for example, for • an earthquake larger than magnitude x occurring within y kilometres of a site) or as seismic £ hazard levels. Alternatively, the combined earthquake history of continental areas that have had a similar I geological history to southern Ontario can be used to constrain the rates of large earthquakes. Thus, though there have been no earthquakes larger than 6 in the Lake Ontario region during I recorded history, it is possible to make robust estimates of their rate. •

A sensible approach is to make several estimates under different hypotheses. Paleoseismic fl evidence may have a role to play even though the associated errors are likely to be very large. • The estimates should be weighted using professional judgement and this weighted value reported together with its uncertainty. The rate and its uncertainty then feeds into the seismic hazard fl assessment. If such a probabilistic hazard estimate is soundly based its conclusions would not " automatically require adjustment if evidence of a coseismic surface rupture was found nearby. _

Prepared on 950530 for AECB Workshop on Seismic Hazard Assessment in southern Ontario, Ottawa 19-21 June, 1995. m I Cfc'(r I -C77- CHARACTER AND REACTIVATION HISTORY OF THE I CLARENDON-LINDEN FAULT SYSTEM: EVIDENCE FROM SOUTHWESTERN NEW YORK STATE

I Robert D. Jacobi and John Fountain I The Clarendon-Linden Fault System (CLF) trends approximately north-south I across western New York, Lake Ontario and southern Ontario Province. This report presents the results of our four-year study that focussed on the previously unidentified southern extension of the CLF in Allegany County, southwestern I New York State. The study involved 11 different, integrated tasks, including reconnaissance geology, and analyses of stratigraphy, structure, soil gas, VLF, geochemistry, remote sensing (lineament), well logs, paleoseismology, seismic I reflection profiles, and seismicity. Data gathered for each task was entered on a GIS (ARC/INFO), and both the data and the maps constructed from the data were I QA/QC'd. A total of 2.8 km of stratigraphie section were measured at the 0.5 cm scale on I over 1000 outcrops. The study area involves Upper Devonian units from the upper West Falls Group to the top of the Canadaway Group. Lithostratigraphic correlations of several marker units, including the contacts of the black shales I (Dunkirk Member and Hume Member) and distinctive sandstones (e.g. the thick sandstones of the Rushford Formation) allowed us to identify regions where I anomalous dips occurs between outcrops. Although the anomalous dip could have several origins, data from other tasks (e.g. structure and seismic reflection profiles) suggest a faulted monoclinal origin for the observed offset of surface I bedrock along north-trending lineaments. Nine cross-sections in northern Allegany County provide evidence for at least 5 CLF faults associated with north-trending lineaments. The maximum offset of the surficial bedrock is on the I order of 60 m (200 ft.). i In the structural analyses task we observed at least 4 north-south zones of north-striking structural features, including steeply dipping layering, increased density of north-south fractures, and series of normal step-faults, each fault with I minimal offset (generally < 15 cm stratigraphie offset). Each of these zones occurs where stratigraphie analyses indicate anomalous local dip between outcrops. We described the character and abutting relationships of the fractures 1 exposed in the 1000+ outcrops examined, and digitally analyzed the fracture patterns, including fractal analyses, in the lab (e.g. Fig. 1). We were able to I identify eight distinctly oriented fracture sets, including a NS set. We found that outside the NS zones of increased NS fracture density, abutting relationships

I I -C78- I imply that the NS fractures postdate the regional NW-trending fracture set and are generally not a master set. However, within the NS zones of increased NS • fracture density, the NS fractures are a master set. It appears that the NS fracture • generation associated with NS faulting was quite restricted spatially in its early generation history. The variable density of NS fractures therefore can be mapped I to identify zones where NS faulting exists; these NS zones correlate with areas where other lines of evidence (e.g. lineaments and seismic reflection profiles) « suggest NS trending faults do occur. We were also able to establish other fault | trends that interact with the NS fault zones: 1) NW fault zones which offset the trend of the NS fault zones, and 2) NE thrust zones with two distinctly different • trends that appear to both abut and offset the trend of NS fault zones. The • Alleghanian NE thrusts in outcrop are associated with 1) stacked shuppen zones of stratigraphically restricted, NE-trending, scaly-cleavage or pencil cleavage, fl and dismembered/deformed bedding, 2) NE-striking folds, 3) deformed fossils, and 4) series of NE-striking listric faults. g

We performed about 80 km (50 mi) of soil gas analyses across northern Allegany County. Soil gas anomalies are recognized by elevated amounts of gas in the soil; H geochemical analyses show mat the gas is thermal gas, not swamp, or biogenic H gas. The soil gas anomalies generally occur in relatively narrow zones (<200 m). These zones imply that narrow pathways~fractures~are open down to Devonian I black shale source rocks, significantly, most soil gas anomalies occur along " lineaments that other tasks, including stratigraphy, structure, lineament analyses, _ and seismic reflection profiles have suggested are fault zones. Thus, soil gas | analyses confirmed the location of fault zones, and suggested that the zones contain fractures that are presently open down to the source rock. Both NS fault • zones and NE fault zones emit high amounts of thermal gas 9

We analyzed topographic maps, air photos, Landsat, and SLAR images for H lineaments. We found a number of lineament orientations, similar to the fracture B trends found in the structure survey. Furthermore, the fractal dimension of the _ topographic and air photo lineaments agrees very closely to that found for the £ fractures observed on outcrops. It therefore can be shown mathematically mat the lineaments are related to the fractures observed in outcrop. The lineaments • allowed us to integrate observations from different sites and different tasks all • along the same lineament. Important results were 1) the clearly defined NS lineaments allowed us to map the NS fault zones from sporadic small outcrop and fi from seismic reflection lines into regions where we had no data, 2) in regions where we had no outcrop data, NS lineaments allowed us to suggest which » structures observed on seismic reflection profiles were definitely related to CLF | structures, and 3) die map pattern of major NS lineaments shows that the main central CLF fault is highly segmented, with NS faults spatially offset by • NW-trending lineament (fault/fracture) zones. The western fault (Rawson Fault) • I I -C79-

I shows a much less well-developed segmented nature, i.e., the NS lineaments are much longer.

I Well log analyses in north central Allegany County confirm the locations and amount of stratigraphie offset of major central CLF faults inferred from other I tasks, including stratigraphy, structure, soil gas, remote sensing, and seismic reflection profiles. However, an insufficient number of well logs in other parts of Allegany County are in the public domain to resolve the complexity now I recognized to be probable from interactions among NW-, NÉ- and NS-trending faults; i.e., in areas where well logs and outcrops are few, the observed offset between wells may be due to a number of faults, only some of which we have I recognized. Well log analyses show that the CLF faults have a growth fault geometry for much of the Upper Devonian section. From very subtle changes in thickness across the assumed traces of the CLF faults, it is possible to reconstruct I the sense-of-motion history for the CLF in Late Devonian times. The well logs suggest that the motion for the central fault was down-to-the-east in Tully to I Genesee time, and up-to-the-east in later times. We shot east-west seismic reflection profiles across the CLF in four different I locations in Allegany County. Older seismic reflection profiles were also procured from industry. The seismic reflection profile near the northern boundary of Allegany County displays an east-directed thrust faulted monocline that affects 1 the entire section-from the Cambrian Theresa to the top of the section, although reflectors above the Tully are difficult to resolve. The maximum cumulative I stratigraphie offset is on the order of 90 m (300 ft), down-on-the-east. The next seismic line to the south (central seismic line) displays no major CLF structure in the units above the Trenton reflector. This absence of structure is surprising, I since the line crosses the NS lineaments that extend south from the CLF fault imaged on the northern line. However, the Trenton and older reflectors do display a structural sag with about 45 m (150 ft) structural relief; the asymmetry I of the sag suggests minor west-directed thrusting. This structure occurs below the NS lineaments, and so is thought to be a continuation of the major structure 1 imaged on the northern line. The major, high-stratigraphic level CLF structures imaged on the northern line could exist east or west of the central seismic line, but such a situation has two requirements: l)the faults above the Silurian salt section I are decoup ,ed from the faults below the salt, 2) the upper faults are displaced spatially along NW-trending fault zones. Both requirements are compatible with the data from the other tasks, as well as close analyses of the seismic line. The I third seismic line crosses the western main CLF fault (Rawson Fault) and displays a thrust-faulted monocline with -82 m (270 ft) stratigraphie offset, I down-on-the-east. Variations in reflector interval thicknesses across faults on all the seismic lines generally confirm the growth-fault geometries inferred from well I log analyses. I I -C80- I Paleoseismological analyses by Tuttle failed to find clear evidence for liquefaction I events in the glacial section overlying the CLF faults. Tuttle inferred that seismic events larger than a magnitude 6 has not affected the CLF in Allegany County in g Pleistocene and Holocene times. However, the methods utilized by Tuttle cannot | discern smaller events (e.g., on the order of magnitude 5 to 5.9). The energetic gas seep at Pike, New York, that was initiated during or shortly after the B Saguenay earthquake, suggests that it is possible to reactivate fractures of the • CLF with quite distant seismic events.

SUMMARY •

We have established that the CLF extends farther south man previously j| supposed--into southwestern New York State. Stratigraphie and structural analyses show that CLF faults affect exposed bedrock. Seismic reflection profiles • demonstrate that these surface bedrock faults are located above faults that affect 8 the entire Paleozoic section. However, faults above the Silurian salt section appear to be spatially more discontinuous than the deeper faults beneath the salt. The B traces of both the shallow and the deep faults apparently are offset by NW-striking fault zones. The CLF faults in AUegany County have been active — for much of Paleozoic time, and recent initiation of gas seeps suggests mat g fractures associated with the CLF can be reactivated by distant seismic activity. 1 I I I I I I I I -C81-

FIGURE 1. An example of the over 120 structure maps constructed for northern Allegany County displaying both modified rose diagrams and detailed maps of fracture patterns. The modified rose diagrams show the number of fractures for each petal orientation in the upper half circle and the normalized length of the fractures for each petal orientation in the lower half circle. The detailed fracture maps generally are digitized from photo mosaics of the outcrop. HOUGHTON QUADRANGLE- CREEK 3A Session D

Methods and Approaches for Determining Seismic Sources Based on Geological and Seismological Information METHODS AND PROCEDURES FOR DETERMINING SEISMIC SOURCES BASED ON GEOLOGICAL AND SEISMOLOGICAL INFORMATION

Kevin J. Coppersmith Geomatrix Consultants 100 Pine Street San Francisco, CA

The purpose of this talk is to define the concept of a seismic source in the context of probabilistic seismic hazard analysis (PSHA), to examine the types of interpretations that are necessary to define sources, and to offer some recommendations for methods to define seismic sources using geologic and seismicity data.

Seismic sources are depicted in map form and represent locations within the earth's crust having relatively uniform seismiciry characteristics. Seismicity characteristics, in this case, refers to those parameters of importance to PSHA: probability of activity, recurrence rate, and maximum magnitude. These characteristics for an identified seismic source are distinct from those for the surrounding region, which itself might be represented by one or more seismic sources. In all cases, the question that the source-characterizer must make answer is: Are there sufficient differences across the region of interest in the location, recurrence rate, and/or maximum magnitude that a seismic source boundary should be identified? The answer to this question differs with the type of seismic source being defined, as discussed below.

A simple example of a seismic source is an active (e.g., Holocene) fault zone. Clearly, the maximum earthquake potential as well as the rate of earthquake occurrence associated with the fault is distinct from that of the adjacent crust. Thus, identification of the fault as a seismic source is well-justified for hazard analysis. It is common for fault sources to be defined from geologic data and for assessments of activity (or 'capability') to come from a combination of geologic data (recency of slip) and seismicity data. Common uncertainties are: evaluations of the recency of slip associated with mapped faults and associations of observed seismicity (e.g., epicenter lineations) with candidate geologic structures.

A second example of a seismic source is a source zone whose boundary encloses a zone of concentrated seismicity separating it from adjacent regions having lower rates of observed seismicity. The zone of concentrated seismicity may or may not have a known association with geologic structure. For example, the New Madrid seismic zone is associated with a multiply- reactivated rift structure, while the Central Virginia Seismic Zone has not yet been associated unequivocally with geologic structure. Because the results of a seismic hazard analysis are highly sensitive to earthquake recurrence rates, the differentiation of these types of seismic sources may be very significant to the results, depending on the location of the site. However, whether or not the different rates of observed seismicity indicate that the maximum magnitude (maximum possible magnitude) within the source and is different from the adjacent region is often not clear. Usually, the maximum magnitude is judged to be related more to the type and dimensions of geologic structure than to the rate of earthquake occurrence. -D2- I

A third example of a seismic source-and the most common within stable continental regions g (SCR) such as eastern North America—is a source zone identified primarily on the basis of geologic, geophysical, and tectonic information, with the consideration of patterns of observed | seismicity as well. The mechanism for the generation of earthquakes—that is, sudden | displacement along faults-is believed to be universal within the earth's crust. However, in stable continental regions the identification of such faults is difficult and a seismic source zone is the 1 way that uncertainty can be displayed and incorporated into a hazard analysis. That is, a 'seismic I source zone' is not a physical reality in the sense that an active fault is clearly physically associated with earthquake processes. Instead, a source zone is an artificial construct devised for 1 seismic hazard analysis to represent a region within which one or more unknown faults are • postulated to exist. The likelihood that at least one seismogenic fault exists within a seismic source zone can be expressed as the 'probability of activity' of the source, discussed further I below. *

Experience in a large number of PSHAs conducted within the central and eastern United States I over the past decade has shown that the configuration and physical basis for seismic sources differs from one source characterizer to the next. Fortunately, it has also been shown that in * many cases the differences in seismic source configurations do not usually have a pronounced 1 effect on the calculated seismic hazard results. This is because, despite differences in the locations of source boundaries, the interpreted maximum magnitudes and recurrence rates are • usually not greatly different among analysts. Nevertheless, the interpretation of seismic source g zones remains one of the few places in PSHA where notions of tectonics can be incorporated into the analysis, both in the interpretation of the source geometry and in the probability of activity. g

Interpretations of the locations of seismic source boundaries includes interpretations of the tectonic features and structures that can serve to localize earthquakes. A wide range of geologic I and geophysical data can be used to identify surface and subsurface structures. For example, I regional tectonic maps identify large-scale fault systems, folds, sutures, etc. that are profound crustal features that might be interpreted to exist throughout the seismogenic crust (upper I approximately 15-20 km). As such, these features might be interpreted to control the location ! of earthquakes. Likewise, deep geophysical data such as isostatic and Bougher gravity and deep seismic reflection data have found application in the interpretation of crustal structure (e.g., deep fl rift faults related to plate extension) that may localize seismicity. The assessment of which • tectonic structures might be most successful in localizing seismicity is usually highly interpretive but is most influenced by the association of a feature with observed seismicity. 1

Once a potential seismic source has been identified that is associated with a particular tectonic - feature (e.g., a seismic source enclosing a system of faults and folds related to Triassic fl continental separation), an evaluation must be made of the probability that the seismic source is 'active.' Active in this sense means that the source is seismogenic or capable of generating • significant earthquakes (M>5) in the present tectonic environment. The Electric Power Research g Institute's (EPRI, 1989) seismic hazard analysis for the central and eastern United States provided an assessment of the criteria for evaluating the probability of activity of identified tectonic g features. In general, the following criteria were identified: | 1 I -D3-

• Spatial association with observed large-magnitude seismicity • Spatial association with observed microseismicity • Evidence for geologically-recent displacement • Existence through the entire depth of crust • Favorable orientation relative to the stress regime • Evidence for brittle slip in the present tectonic regime • Geologic evidence for multiple episodes of reactivation • Paleoseismic evidence for prehistorical earthquakes

Clearly, the relative importance of these (and other) criteria for evaluating the seismogenic potential of particular tectonic features is different. A Mesozoic rift system that can be spatially associated (if not causally associated) with a large-magnitude historical earthquake would, in most experts' assessments, have a higher probability of being seismogenic than an identical rift lacking observed seismicity. Therefore, the criteria for evaluating activity can be ranked and rated. Further, each tectonic feature having some potential for being seismogenic can be evaluated relative to these criteria. The result is an assessment of the probability of activity, which can be used in the hazard analysis. In the EPRI study and other similar studies in the U.S., this assessment was made explicit. In other studies, the assessment of the seismogenic potential of tectonic features or, more generally, of a particular potential source region is evaluated implicitly when defining the configuration of seismic source zones-and in defining the configuration of alternative seismic source zones in the same region. In either case, there is usually considerable uncertainty in evaluating seismic sources within SCRs and this uncertainty should be incorporated into a PSHA.

Increasingly, seismicity data are being supplemented by geologic paleoseismic data in the evaluation of the locations, size, and recurrence rates of large-magnitude earthquakes. Examples are studies in the Charleston, South Carolina area and in the Wabash Valley, Illinois. An important aspect of both of these study areas is that both the presence and absence of paleoliquefaction is being mapped in order to constrain the size and location of prehistorical earthquakes. Studies of active SCR faults, such as the Meers fault in Oklahoma, are providing insights into deformation rates and processes within SCR. No doubt the integration of geologic, geophysical, and seismicity data will continue to provide insights into the nature of earthquake occurrence and, equally important, provide information with direct application in seismic hazard analysis.

REFERENCES

Electric Power Research Institute (EPRI), 1989, Probabilistic seismic hazard evaluations at nuclear power plant sites in the central and eastern United States: Resolution of the Charleston earthquake issue: EPRI Report NP-6395-D. AECB Workshop on Seismic Hazard Assessment in Southern Ontario, Ottawa, June 19-21, 1995

Seismicity of the Lake Ontario Region to 1994 .

Anne E. Stevens Geological Survey of Canada, 1 Observatory Crescent, Ottawa, Canada K1A 0Y3 I

Abstract I

Earthquakes in the Lake Ontario region (1840-1994) were examined for geographical groupings of I possible significance for seismic hazard assessment. Prior to the 1970s, the earthquake locations - were too inaccurate to support any geographical pattern of epicentres or to justify any correlation ] with other proposed geological or geophysical trends.

For the period of best earthquake monitoring (1970 to 1994), most of the natural earthquakes occurred under and south of Lake Ontario, with very few near its north shore and no earthquakes I located on the Niagara Peninsula. No reliable alignments of epicentres could be identified. For the earthquakes located under Lake Ontario in this period, an accurate calculation of depth was I impossible, and remains so even at present. Earthquakes more than 20 km offshore are too far from ' the nearest seismograph station to be able to distinguish between shallow, mid- and lower-crustal g focal depths. j

Thus, the earthquake locations are either too inaccurate (pre-1970) or too scattered or both to define j any patterns, linear or otherwise. If the epicentres by themselves do not show any well-defined linear trends, then they cannot be said to coincide spatially with any linear geologic or geophysical j feature. Earthquakes in the Lake Ontario region are characterized by intermittent scattered activity, with no preferred trends identifiable to date. I

Reference

Stevens, A.E., 1995. Earthquakes in the Lake Ontario region: intermittent scattered activity, not persistent trends, Geoscience Canada, accepted 28 February 1995, currently in press. -D5-

SEISMOGENESIS AND STRUCTURE IN THE LAKE ERIE-LAKE ONTARIO REGION OF THE U.S. FROM A GLOBAL PERSPECTIVE Leonardo Seeber and John Armbruster Lamont-Doherty Earth Observatory of Columbia University, Palisades N.Y. 10964 Overview The region south of Lake Ontario and Lake Erie (Figure 1) is part of a continental platform in the interior of the North American plate and can be classified as a stable continental region (SCR; Johnston, 1989). This region is characterized by Paleozoic sedimentary rocks lying nearly flat on Precambrian (Grenville) basement. The thickness of these rocks vary from a few hundred meters north of Lake Ontario to several kilometers in the Appalachian foreland. Several distinct zones of seismicity can be identified from historic earthquake catalog, particularly after the catalog is improved with systematic searches of archival material. In this work we consider the relation between seismicity and structure in the Lake Ontario-Lake Erie region and compare it with characteristics of SCR worldwide.

Intraplate continental areas generally exhibit geologic evidence of neotectonic stability, nevertheless, earthquakes, which represent deformation events, do occur in these areas. The rate of seismicity in stable continental regions (SCR) is low in comparison to plate boundaries. But, SCR earthquakes can be sufficiently large and frequent to pose a substantial hazard (e.g., Johnston, 1989). Furthermore, they indicate contemporary tectonic deformation that needs to be considered in light of the geologic characterization of these areas as "stable" (Coppersmith and Youngs, 1989; Dawers and Seeber, 1991; Seeber and Armbruster, 1989). The apparent inconsistency between short- and long-term manifestations of the tectonic process (Seeber and Armbruster, 1993) can now be confronted with paleoseismic data from faults that generated recent large SCR earthquakes and ruptured the surface. Results suggest extremely low rates of activity on each of these faults and the tendency for activity to be concentrated in time/space clusters with very wide temporal separations (Crone et al., 1992). In order to account for the observed moment release rate, many more sources than the currently active ones must exist. The low rate on individual faults makes it difficult to concentrate spatially the long-term moment release and seismogenic faults are probably widely scattered over SCR (Seeber and Armbruster, 1993). A long-term pattern of diffused seismicity is consistent with the geologic stability because moment rate is diffused on many small-displacement faults, but may be problematic for the interpretation of historic seismicity as representative of the long-term distribution. Further complicating this interpretation is the triggering of earthquakes by human activities. Thus, conclusions on future seismicity based solely on past seismicity under the assumption of stationarity may be misleading.

An Improved Earthquake Catalog Accuracy, uniformity of coverage, and continuity are desirable characteristics of an earthquake catalog. We reexamine archival sources, mostly newspapers in the eastern US, and recover new macroseismic data which improve constraints on source parameters and remove 'noise' by identifying unreliable events (Seeber and Armbruster, 1991,1993; Mohajer, 1987). Our reexamination effort covers the period preceding seismic instrumentation as well as the early period of instrumentation because source parameters from early instruments can be quite misleading and constraints can be often improved with felt reports. By improving macroseismic coverage, magnitudes can be obtained from intensity distribution (e.g., from felt area) rather than maximum intensity or from a single seismogram, which are notoriously unreliable measures of earthquake size. Finally -D6- I earthquakes with sufficient data are relocated using MACRO (Figure 2), an algorithm that I yields epicenters and magnitudes from intensity data (Arrnbruster and Seeber, 1987). We consider it encouraging that the spatial clustering of epicenters is enhanced after our revision of the historic catalog (Seeber and Armbruster, 1993)

Seismicity and Geology in the Lake Erie-Lake Ontario Region 1 The most prominent seismic zones in the area south of Lake Erie and Lake Ontario in Figure 1 are the Northeastern Ohio seismic zone (NEOSZ) and the Attica seismic zone J (ASZ). They contain the two known M>5 earthquakes in the region, the 1986 M5.0 Leroy I (Figure 3) and the 1929 M5.2 Attica earthquakes, respectively. The NEOSZ and the ASZ are associated with the Akron Magnetic lineament (Figure 4; Hildebrand and Kucks, 1984; I Lidiak et al., 1985; Lucius and Frese, 1988) and with the Clarendon-Linden fault (Figure I 5; Chadwick, 1919; Van Tyne, 1975; Hutchinson et al., 1979; Fakundiny et al., 1978). " Both these features are recognized as expressions of regional basement structures of Grenville age (King and Ziez, 1978; Culotta et al, 1990; Pratt et. al., 1989). The I Clarendon-Linden fault represents a reactivation in the brittle regime of a Precambrian I ductile structure which is now in the upper crust. This fault was reactivated twice during the Paleozoic, once in extension and once in compression, and appears to be currently i active as a source of historic seismicity. The Clarendon-Linden fault has been episodically g active and possibly seismogenic over hundreds of millions of years, yet the total accumulated displacement is only a few tens of meters. Thus, geologically subtle small- . displacement faults can be the source of potentially damaging earthquakes. Such small- I displacement faults could be undetected in areas without detailed geologic data and a hereto • unmapped small displacement fault or set of faults could be associated with the Akron Lineament. Furthermore, many other geologically subtle but potentially seismogenic small- I displacement faults could exist. | Seismic zones in northeastern North American platform appear to be associated with « regional structural features, yet brittle faults associated with these features have very little 1 accumulated displacement. These faults cannot account for the observed moment release extended over the neotectonic period, unless the maximum earthquakes on these feature have been sampled historically. Assuming, instead, that the typical maximum magnitude I limit for seismogenic structures in the region is similar to the magnitude of large SCR I earthquakes worldwide (M6-7), and assuming that the rate of historic seismicity is representative of the long-term rate, then small accumulated displacements on the active i faults can be maintained only if there are a lot of these faults, most of which can be | expected to be aseismic during historic time. Is the combination of apparent geological stability and earthquake activity that seems to characterize the Lake Erie-Lake Ontario region representative of S CR? j

Seismogenesis and Structure in SCR j A number of large SCR earthquakes have occurred in the last few decades worldwide which have been studied both seismologically and geologically. Crone et al. (1992) . summarizes Paleoseismic results from several recent earthquakes in Australia by J concluding that times to the most recent prehistoric events are estimated to be 100,000 ' years or longer. In North America, the 1990 Ungava earthquake was found to have ruptured a fault that coincided with an Archean ductile structural boundary, which, | however, showed no evidence of prior brittle reactivation (Adams et al., 1991). The Meers I fault ruptured twice during the Holocene, but the total accumulated displacement on Holocene and Pleistocene strata is the same, indicating that this fault had not ruptured at the i -D7- surface for at least 100,000 years prior to the Holocene ruptures (Crone and Luza, 1990). In peninsular India, the 1993 Killari earthquake ruptured to the surface; preliminary evidence suggests no prior rupture in the 65 my old cap rocks of the Deccan Traps (Seeber et al., JGR, in review). The 1967 Koyna earthquake also ruptured to the surface, and basalt flows showed no significant accumulated displacement across the rupture (Sahasrabudhe et al., 1969).

The PDE lists 12 M>6.0 earthquakes during the last 30 years in the combined SCR of Australia, India and Eastern North America (Figure 6 and Table 1), where most of the geologic studies have been carried out. Of these 12 earthquakes, 9 have centroid depths less than 5 km deep and ruptured the surface. The 1970 Ms5.7 Calingiri event also ruptured the surface and fits in the same category. A common pattern emerges from recent paleoseismic studies of these ruptures; penultimate ruptures are either non existent, i.e., the fault had not had a prior brittle rupture, at least where it was observed at the surface (e.g., 1990 Ungava, 1993 Killari), or have occurred long ago (>100,000 yr.; Crone et al., 1992). Thus, the following may be generalized from the last 30 years of SCR earthquakes: 1.) A prominent class of large SCR earthquakes are shallow and rupture the surface; 2.) Recurrence rates are very long, if they exist at all; 3.) These earthquakes occur on faults that may be associated with structures inherited from tectonic phases preceding the current SCR, but they exhibit only subtle or no sign of neotectonic activity. If the available geologic data on seismogenic faults and the historic rate of seismicity are representative of the long-term behavior of SCR, then the rate of earthquakes from individual faults can only be reconciled with the rate of intraplate seismicity by the presence of a large number of potentially seismogenic faults, the vast majority of them still unidentified (Seeber and Armbruster, 1993). Let F be the number of faults that can produce surface ruptures, R be the average rate of these ruptures on an individual fault, and N be the number of surface ruptures in the time T. Then,

F=N/RT.

If N=10, T=33 years, and R=10"5/y (Table 1), then F=3xlO4 faults over the SCR of Australia, India, and eastern North America (Figure 6). A rate of 10"5/y may be at the high end of the range typical of SCR (Crone et al.,1992; Machette et al., 1993) and 3X104 may be a minimum number of faults. Such a large number of faults would need to be scattered over most of these SCR, if the spacing between them is allowed to be similar to the rupture dimensions of a M6-7 earthquake (e.g., a 20x20 km area) for each of these faults. These numerical results are very uncertain, but they point out that the low rate of paleoseismicity on individual faults combined with the historic rate of M>6 earthquakes suggests many low-displacement but potentially seismogenic faults scattered widely over SCR. Thus, tentative conclusions derived from the contrast between earthquake and geologic data in the Lake Ontario-lake Erie region (Seeber and Armbruster, 1993) are substantiated by data from SCR worldwide. The tectonic regime in SCR is very different from the concentration of moment release on few large faults observed at plate boundaries. While along plate boundaries the determination of earthquake hazard on the basis of individual active faults may be a realistic goal, such an approach is likely to be misleading in SCR. In these areas, not only potentially active faults may be numerous and ubiquitous, but they are also likely to remain unknown until they produce a surface-rupturing earthquake because they are very difficult to identify from regional geological investigations. -D8- ' I Triggered Seismicity The sedimentary rocks of the eastern North American platform are being mined throughout • the region for oil, gas and salt and are being used as repository of waste. Some of these activities were already under way early this century. These engineering endeavors often involve pumping large volumes of fluids in or out of deep wells; such activities alter ji circulation, pressure, and chemistry of fluids in the sedimentary rocks and in the upper part M of the basement at seismogenic depths. Artificial changes in the hydrology of the upper crust alter mechanical conditions and can trigger seismicity. The closer the environment is • to failure in the natural state, the more likely that artificially induced changes will trigger | earthquakes (Nicholson and Wesson, 1990). Several recent cases of artificially triggered seismicity have been well documented in the I Lake Erie-Lake Ontario area. Seismicity has been associated with salt brine recovery at Dale ™ (Attica), western N.Y. (since 1971; Fletcher and Sykes, 1977), oil/gas recovery in Gobies, southern Ontario (since 1979; Mereu, 1986), and waste disposal in Ashtabula, northeastern • Ohio (since 1987). Several other less convincing instances of induced seismicity have also fl been reported (e.g., Nicholson and Wesson, 1990). These include recent well documented earthquakes where circumstances are ambiguous, such as the 1986 M=5.0 Leroy A earthquake, and earlier cases where engineering activities could have triggered seismicity, I but available data are inconclusive. The 1929 M=5.2 Attica earthquake may be an example of the latter. Confidence limits on its location (Dewey and Gordon, 1984) allow it to have originated near brine fields which were already active in 1929. Thus, it is possible that the A two largest events in the study area were artificially triggered. I Fluid Injection and Seismicity in Ashtabula, Ohio g The 13 July 1987 Mb=3.6 Ashtabula earthquake (Figures 1 and 4) is one of the clearest examples of a macroseismic event triggered by fluid injection in a deep waste disposal well - (Nicholson and Wesson, 1990; Seeber and Armbruster, 1993). This event generated I relatively abundant aftershocks, which we locally monitored for ten days beginning two • days after the main shock. All 36 well-determined hypocenters are clustered in a narrow east-striking vertical zone about 1.5 km long and extending from a depth of 1.7 to 3.5 km I (Figures 7). This distribution suggests a single active fault. First-motions fit a well- f constrained composite fault-plane solution characterized by a left lateral nodal plane that matches closely the attitude of the hypocentral zone. The combined hypocentral and first- I motion data delineate a vertical active fault; we refer to this feature as the Ashtabula fault. I The motion on this fault is left-lateral and is consistent with the east-northeast direction of horizontal compression in the regional stress (Zoback and Zoback, 1989; Sbar and Sykes, 1973). The Appalachian Plateau in northeastern Ohio is characterized by sub-horizontal J Paleozoic sediments (4.5 km/sec) above Precambrian basement (6.0 km/sec). The depth of I the unconformity in the epicentral area is 1.8 km. Thus, seismicity is concentrated near the top of the basement; considering location uncertainties (Figure 7), most and possibly all of j the July 1987 hypocenters are located below the unconformity. I The July 1987 Ashtabula sequence occurred near an injection well that had been in operation for only a year, since July 1986. The zone of injection for this and for other deep waste-disposal wells in northeastern Ohio is in the Mt Simon formation, a permeable sandstone at the base of the sediments. This well is less than a km from the July '87 hypocenters (Figure 7). No prior earthquakes are known to be located within 30 km from I Ashtabula. The close spatial and temporal correlation between injection and seismicity is strong evidence for a triggering effect. -D9-

In an isotropic medium, mechanical changes induced by injection are expected to decrease rapidly away from the injection point. Since the Ashtabula fault is located 0.7 km away from this point, a distance far exceeding the confidence limits of the locations (Figure 7), it is likely to be a reactivated pre-existing fault rather than a new fault which may be expected to radiate from the well. No evidence for such a fault has yet been reported from geologic data. A reflection survey for the purpose of studying the structural environment of the well apparently did not resolve the seismogenic fault (Ashtabula Star Beacon, 9/24/92). According to the same report, the well operators concluded from the reflection data that "fault lines are not located near the area and the earthquakes are not related to the operation of the injection well". We conclude that seismicity was triggered on a preexisting fault and that accumulated displacement on this fault was probably small in order for it to be invisible in reflection profiles.

The possible concentration of Ashtabula seismicity on a single basement fault has important implications for earthquake hazard assessment. First, the combined data from 1987 (Figure 7) and from 1992 (Seeber and Armbruster, 1993) are consistent with a single active fault that may be 5-10 km long and is slipping coherently in response to the regional stress field. Conceivably, such a fault could rupture in a single event and generate an earthquake in the M=5-6 range. Secondly, if the fluid is migrating away from the injection area by a quasi 1- dimensional flow along the upper reaches of a steep basement fault, then effects of the injection, including induced seismicity, may occur at distances from the well much larger than for two-dimensional flow (Nicholson and Wesson, 1990). Thirdly, induced slip on the vertical basement fault may extend into the overlying sedimentary rock locally increasing permeability. The opening of fractures caused by the deformation may breech flat-laying aquicludes and may increase the risk of contamination of shallow aquifers by the injected waste fluid. Conclusions The geologic setting and the seismicity of the Lake Ontario-Lake Erie region are typical of a SCR and the relation between seismicity and structure in this region leads to conclusions similar to the conclusions reached about space-time distribution of SCR earthquakes in general. Thus, in view of the scarcity of available local data, important conclusions regarding the distribution of future seismicity in this area may be drawn from worldwide characteristics of SCR seismicity (e.g., Johnston, 1989). A prominent class of large SCR earthquakes is characterized by shallow ruptures that reach the surface. Paleoseismic data from these ruptures reveal faults that have very low rates of neotectonic activity. The low seismogenic rate per fault requires a large expected number of faults to keep up a moment rate as observed in the last 3 decades. Most of these faults are quiescent at any one time and are expected to be widely scattered over SCR. Another important characteristic of SCR earthquakes is that they tend to occur in time-space clusters. By assuming a stationary behavior of the seismicity, current methodologies for delineating the expected distribution of earthquake sources focus on the spatial component of this clustering. The temporal component tends to be neglected because it is poorly constrained by the short historic record and by the sparse paleoseismic record. Yet, it may be very important, as indicated by the large percentage of SCR earthquakes located in areas without obvious seismological or geological symptoms of prior activity. Thus, future SCR seismicity can be expected from historically active sources as well as from previously quiescent sources that, at the current state of knowledge, we can assume to start up with equal probability anywhere. Relative weights to these two components of future seismiciry can be assigned on the basis of worldwide SCR data, assuming uniform behavior in these regions. -D10- I

A substantial but poorly constrained proportion of current seismicity is triggered by human activities. This seismicity may significantly bias the historic rate in terms of non-triggered 1 seismicity. Furthermore, the probability of a damaging earthquake may be significantly m modified by the onset of activities known to effect mechanical conditions at seismogenic H depths. This is the one parameter effecting future seismicity that is available for calibration m and monitoring, at least in theory. Recognizing and quantifying the component of the seismicity which is triggered is important from the hazard viewpoint: first, because it offers ft the opportunity to reduce the hazard by reducing the seismicity; secondly, because it • permits an unbiased record of natural seismicity; and thirdly because induced earthquakes occur in a partially controlled physical environment and offer a unique opportunity to m explore the mechanical behavior of the upper crust. £

Finally, large SCR earthquakes are often preceded by bursts of small earthquakes, 8 particularly in cases where this seismicity is in a previously quiescent area. Of course, " many other bursts of seismicity are not precursors to large earthquakes. Nevertheless, the probability of a large earthquake can be envisioned to increase after such a burst. Short • term foreshocks are evaluated statistically in California (e.g., Jones, 1995). A similar fl approach could be used to evaluate changing patterns of current seismicity in terms of a time-dependent distribution of future large earthquakes. m References Adams, John, R.J. Wetmiller, H.S. Hasegawa, and J. Drysdale, 1991, The first surface M faulting from an intraplate earthquake in North America, Nature, V. 352, p 617. V Armbruster, J.G., and L. Seeber, 1987, Seismicity 1886-1889 in the Southeastern United States: The aftershock sequence of the Charleston, S.C. earthquake, U.S. Nuclear m Regulatory Commission, Washington, D.C., NUREG/CR-4851, 153 pp. I Chadwick, G.H., 1919, Large faults in western New York, Geol. Soc. Amer. Bull, V. 30, p. 117-120. Coppersmith, K.J. and R. R. Youngs, 1989, Issues regarding earthquake source I characterization and seismic hazard analysis within passive margins and stable continental • interiors, in S. Gregersen and P.W. Basham (eds.), Earthquakes at North-Atlantic Passive Margins, Kluwer Academic Publishers, Boston, Mass, p.601-631. I Crone, A.J. and K.V. Luza, Style and timing of Holocene surface faulting on the Meers f fault, southwestern Oklahoma, Geol. Soc. Am. Bull., 102, 1-17, 1990. Crone, A.J., M.N. Machette, and J.R. Bowman, Geologic investigations of the 1988 m Tennant Creek, Australia, earthquakes ~ Implications for seismicity in stable continental | regions, U.S. Geological Survey Bulletin 2032-A, 51 p., 1992. ™ Culotta, R.C., T. Pratt and J. Oliver, 1990, A tale of two sutures: COCORP's deep seismic surveys of the Grenville province in the eastern U.S. midcontinent, Geology, v. I 18, p. 646-649. | Dawers, N. and L. Seeber, Intraplate faults revealed in crystalline bedrock in the 1983 Goodnow and 1985 Ardsley epicentral areas, New York, Tectonophysics, 186, 115- c 131., 1991. I Dewey, J.W. and D.W. Gordon, 1984, Map showing recomputed hypocenters of earthquakes in the eastern and central United States and Canada, 1925-1980: U.S. Geological Survey Miscellaneous Field Studies Map, MF1699, scale 1:2,500,000. I Fakundiny, R.H., J.W. Pferd, and P.W. Pomeroy, 1978, Clarendon-Linden Fault System 1 of western New York: Longest(?) and oldest (?) active fault in eastern United States. Northeastern Section of the Geological Society of America, 13th Annual Meeting, | Abstracts with Programs, p. 42. | Fletcher, J.B., L.R. Sykes, 1977, Earthquakes related to hydraulic mining and natural seismic activity in western New York State: Jour. Geophys. Res., v. 82, p. 3767-3780. . -Dll-

Gordon, F.R. and J.D. Lewis, The Meckering and Calingiri earthquakes October 1968 and March 1970, Geological Survey of Western Australia Bulletin 126, 229 p., 1980. Hildebrand, T.G. and R.P. Kucks, 1984, Residual total intensity map of Ohio, Map GP- 961, U.S. Geological Survey, Reston Virginia. Hutchinson, D..R., P.W. Pomeroy, RJ. Wold and H.C. Halls, 1979, A geophysical investigation concerning the continuation of the Clarendon-Linden fault across Lake Ontario: Geology, v. 7, p. 206-210. Johnston, A.C., The seismicity of 'Stable Continental Interiors', in S. Gregersen and P.W. Basham (eds.), Earthquakes at North-Atlantic passive margins, Kluwer Academic Publishers, Boston, Mass, p. 299-327, 1989. Jones, L.M., Foreshocks and time-dependent hazard assessment in southern California, Bull. Seism. Soc. Am., 75, 1669-1680, 1985. King, R.R,. and I. Zietz, 1978, The New York-Alabama lineament: Geophysical evidence for a major crustal break in the basement beneath the Appalachian basin: Geology, v. 6, p. 312-318. Lidiak, E.G., WJ. Hinze, G.R. Keller, J.E. Reed, L.W. Braile and R.W. Johnson, 1985, Geologic significance of regional gravity and magnetic anomalies in the east-central midcontinent, in Hinze, WJ., éd., The utility of regional gravity and magnetic anomaly maps: Tulsa, Oklahoma, Society of Exploration Geophysicists, p 287-307. Lucius, J.E., and R.R.B. von Frese, 1988, Aeromagnetic and gravity anomaly constraints on the crustal geology of Ohio: Geological Society of America Bulletin, v. 100, p. 104- 116. Machette, M.N., A.J. Crone, and J.R. Bowman, Geologic investigations of the 1986 Marryat Creek, Australia, earthquake - Implications for paleoseismicity in stable continental regions, U.S. Geological Survey Bulletin 2032-B, 29 p., 1993. Mereu, R.F., J. Brunet, K. Morrissey, B. Price, and A. Yapp, 1986, A study of the microearthquakes of the Gobies oil field area of southwestern Ontario: Bulletin of the Seismological Society of America, v. 76, p. 1215-1223. Mohajer, A.A., 1987, Reappraisal of the Seismotectonics of Southern Ontario Task 1: Relocation of Earthquakes and Seismicity Pattern, Atomic Energy Control Board, Energy, Mines and Resources and Ontario Hydro, Research Report INFO-0293, 62 p. Nicholson, C. and R.L. Wesson, 1990, Earthquake hazard associated with deep well injection-A report to the U.S. Environmental Protection Agency, U.S. Geological Survey Bulletin 1951,74 pp. Pratt, T., Culotta, R.C. Hauser, E., Nelson, D., Brown, L., Kaufman, S., Oliver, J., and Hinze, W., 1989, Major Proterozoic basement features of the eastern midcontinent of North America revealed by recent COCORP profiling: Geology, v. 17, p. 505-509. Sahasrabudhe, Y.S., V.V. Rane, and S.S. Deshmuckh, Geology of the Koyna Valley, Proceedings of Symposium on Koyna Earthquake, Indian Journal of Power and River Valley Development, 47-54,1969. Sbar, M.L., and L.R.Sykes, 1973, Contemporary compressive stress and seismicity in eastern North America - An example of intraplate tectonics: Geological Society of America Bulletin, v. 84, p. 1861-1882. Seeber, L. and J.G. Armbruster, 1989, Low-Displacement Seismogenic Faults and Nonstationary Seismicity in the Eastern United States, in Annals of the New York Academy of Sciences, v. 558 edited by K.H. Jacob and C.J. Turkstra, p. 21-39. Seeber, L. and J.G. Armbruster, 1991, The NCEER-91 Earthquake Catalog: Improved intensity-based magnitudes and recurrence relations for US earthquakes east of New Madrid, NCEER Technical Report 91-0021, 98 pp. Seeber, L. and J.G. Armbruster, Natural and induced seismicity in the Erie-Ontario region: Reactivation of ancient faults with little neotectonic displacement, Géographie Physique et Quaternaire, 47, 363-378, 1993.

I -D12- S

Seeber, L., G. Ekstrom, S.K. Jain, C.V.R. Murty, N. Chandak, and J.G. Armbruster, The 1993 Killari earthquake in central India: A new fault in Mesozoic basalt flows?, Jour. Geophys. Res, in reviw, 1995. „ Simpson, D.W., Triggered earthquakes, Ann. Rev. Earth. Planet. Sci., 14, 21-42, 1986. • Van Tyne, A.M., Clarendon-Linden structure, western New York, New York State • Geological Survey, Albany, Open File Report, 1975. Zoback, M.L., and M.D. Zoback, 1989, Tectonic stress field of the continental United States, in Pakiser, L., and Mooney, W., eds., Geophysical framework of the continental United States: Geological Society of America Memoir 172, p. 523-539. I I a i i i i i i

i -D13-

TABLE 1

PDE 1963 - 1993 M >= 6.0 or surface rupture

EARTHQUAKE LAT LON DATE centroid mag Surface depth Mb Ms Rupture range

Australia Meckering -31.518 116.971 1968 10 14 1-5 6.0 6.8 SR Calingiri -31-010 116.540 1970 3 12 5.0 5.7 SR -21.981 126.682 1970 3 24 5-11 6.2 5.9 -22.126 126.721 1975 10 3 6.0 .0 Cadoux -30.812 117 .179 1979 6 2 2-4 6.0 6.1 SR Marryat Creek -26.194 132.767 1986 3 30 0-3 5.8 5.8 SR Tennant Creek -19.847 133.803 1988 1 22 2-6 6.1 6.3 SR Tennant Creek -19.798 133.910 1988 1 22 3-4 6.1 6.4 SR Tennant Creek -19.829 133.882 1988 1 22 4-5 6.5 6.7 SR

India

Rann of Kutch +23.000 070.000 1956 7 21 6.0 not SCR Koyna +17.700 073.900 19G7 12 10 3-6 6.0 .0 SR Killari +18.066 076.450 1993 9 29 2-4 6.3 6.2 SR

N America

Saguenay +48.050 288.900 1988 11 25 25-30 5.9 6.0 Ungava +60.080 286.555 1989 12 25 1-3 6.2 6.3 SR

Range of centroid depths either from range of multiple determinations by different authors or from uncertainties of single determinations. LAKE ONTARIO

.c- i

NCEER 0 100 km o o O 3 4 5 MAGNITUDE 40° 83° o Figure 1. Epicenters of earthquakes listed in the NCEER-91 catalog (Seeber and Armbruster, 1991) with additions and modifications (from Seeber and Armbruster, 1993). Events with M<3 and/or reported felt only at one place and/or suspected to be non-earthquakes are excluded. Earthquiikes inferred to be artificially triggered are represented with filled squares (A=Ashtabula; G=Gobles; D=Dale-Attica; ASZ=Attica seismic zone; WLOSZ=Western Lake Ontario seismic zone; NSZ=Niagara seismic zone; NEOSZ=Northeast Ohio seismic zone). 80 77 80 78 1857-10-23 lat 42.71 Ion -78.98 Imax 5 Magnitude 4.0 1873-07-06 lat 42:69 Ion -79.14 Imax 5 Magnitude 4.2

Figure 2. New macroseisniic data and relocated epicenters for two historic earthquakes in the eastern end of Lake Erie. Epicenters, magnitudes and error bars (one standard deviation) are obtained with the algorithm MACRO (see text). X's denote previous epicenters. -D16-

H 41°40'

Figure 3. Map and two perpendicular vertical sections of the aftershock hypocenters of the 1986 Mb=5.0 Leroy, Ohio earthquake (plotted in Figure 4) as located with data from a temporary local network. Early aftershocks suggest a vertical rupture about 1 km across and striking northeast. The depth range of the aftershocks places the rupture in the Precambrian basement, significantly below the unconformity at the base of the Paleozoic sedimentary rocks (From Seeber and Armbruster, 1993). The northeast striking nodal plane is parallel to the Akron Lineament. 42°N MAGNITUDE

42°N

81°W

Figure 4B. 1986-1991 epicenters from the network operated by1 John Carroll University in Cleveland. Inferred rupture planes are shown for the 1986 Leroy earthquake (Mb=5.0; Figure 3) and for the 1987 Ashtabula earthquake sequence (M=3.9; Figure 7; from Seeber and Armbruster, 1993).

82°W

Figure 4A. The Akron Lineament as represented by aeromagnetic data and M>3.0 epicenters. Macroseismic epicenters and related error bars have been determined with MACRO (see text). Error bars are smaller than the circles for most recent instrumental epicenters (from Seeber and Armbruster, 1993). -D18- I I I I I I I I I

Figure 5. The Clarendon-Linden fault as mapped in Paleozoic sedimentary rocks (Van Tyne, 1975); aeromagnetic data (Hildebrand and Kucks, 1984), and epicenters in die Attica area of western New York; same epicenters as in Figure 1 (open circles) plus regional network data from the Lamont-Doherty Observatory in New York and the Geologic Survey in Canada (Filled circles). The 1929, M=5.2 epicenter (to the south) and the 1844, M=3.5 epicenter (to the north) are the only ones with error bars (80% confidence, from Dewey and Gordon, 1984; and 75% confidence, from MACRO, respectively)..These confidence limits allow them to be on the Clarendon-Linden fault zone. All other epicenters in this figure are similarly or better constrained and could also be in the fault zone (from Seeber and Armbruster, 1993). 80

0°

-40°

-80 40 120 160 200 240 280° 320°

Figure 6. PDE epicenters of all 1963-1993 M>6.0 events in the boxed stable continental regions (listed in Table 1). -D20- I 41* EARTHQUAKES ASHTABULA, OH. JULY 16-26,1987 55' I

INJECTION WELL MIDDLE ROAD I I

41° I 54'

A km I _L 80°4S' 8O°43' I I c 1 INJECTION WELL-

Figure 7. Map, vertical sections and focal mechanism (composite, upper hemisphere) of the seismicity in Ashtabula following the June 1987 M=3.6 event. The position of a class 1 injection well is indicated by the triangle and the vertical bar. This facility pumped waste fluids into the formation just above the unconformity between Precambrian basement and platform rocks at a depth of 1.8 km (dashed line). The 35 hypocenters were quality selected from a total of 54 events recorded by a temporary network operated for ten days. Line segments in A and C connect two hypocenters obtained for each of the events with two independent sets of phase readings. The HYPOINVERSE error ellipse represents an event recorded in August 1989; it is representative of formal confidence limits for the other hypocenters (from Seeber and Armbruster, 1993). -D21-

The Southern Ontario Seismic Network - Application to Seismic Hazard and Earthquake Engineering

R.F. Mereu and H.W. Asmis

Abstract

In recognition of the wealth of useful data obtained from regional seismic networks, such as the Geological Survey of Canada's Eastern Canadian Telemetred Network (ECTN), Ontario Hydro Nuclear (OHN) recently invested in the development of a new seismographic network focused on the southern Ontario region, home of major nuclear plants.

The main principles guiding the network design were:

(i) the network must have scientific credibility for detecting and accurately locating all significant regional earthquakes;

(ii) the network must have engineering significance, by being able to recover the free-field ground motions from significant regional earthquakes.

The chosen design for the Southern Ontario Seismic Network (SOSN), obtained from Nanometrics, was a modern digital system with wide dynamic range, featuring three-component S-13 Teledyne-Geotech seismometers. This configuration is expected to stay on-scale for nearly all motions that are reasonably likely to occur. The signals are digitized directly at the source, at 100 samples/sec, and sent directly, in real time, to a central server at the University of Western Ontario, via leased digital phone lines.

Site selection and preparation were key components of the design, particularly since rock sites are not available. All selected locations are in quiet farm fields, far away from trees, buried in ten-foot deep steel vaults, on firm glacial till. The SOSN currently has 4 sites around the western end of Lake Ontario; two new sites are being added in 1995. OHN also contributes to the operation of a GSC ECTN station (SADO) in the region of overlap between the SOSN and the ECTN networks.

The results from the SOSN so far have exceeded expectations. Several significant regional earthquakes have been recorded, including the mN 4.4 earthquake of 94/01/06 near Reading, Pennsylvania, and the recent 95/05/25 mN 3.1 Buffalo earthquake, which was strongly felt, despite its modest magnitude. New information has been gathered on the site response and ground motion characteristics of typical southern Ontario sites (Atkinson, 1995). A 'rapid-response' package has been developed to estimate site ground motions throughout the region following any significant local or regional earthquakes, thus ensuring the network can deliver timely information of direct engineering relevance. The ability to calibrate site responses through repeated recordings has great engineering relevance; this eliminates the problems associated with 'once-in-a-lifetime' recordings from strong ground motion equipment, located on sites with assumed or poorly-known properties. -D22- I

From many small local events, and numerous quarry blasts, it has been found that SOSN can ™ detect and locate earthquakes of (as low as) a magnitude 1-1.5, within the covered region. Improvement in this detection threshold is expected when the SOSN is expanded by two 1 stations this summer, but there may be a lower limit to detection in Southern Ontario, due to " considerable cultural noise. However, the extremely accurate earthquake locations produced by the SOSN will allow correlations to known geological structures, if at all possible. I

The SOSN will also become an essential component of the new Ontario Hydro Nuclear _ in-plant seismic monitoring. Using computer networking technology, the existence of the I SOSN as an accurate earthquake locator (using exact atomic time), will allow an efficient design, that will produce good engineering strong ground motion data. + I I I I I

1 I I I I I I r i n -D23-

Local Seismic Monitoring East of Toronto

A. A. Mohajer

University of Toronto / Seismican

Background and Standards

Microearthquake monitoring in a dense local network is one of the techniques, usually used to delineate currently active seismic sources. The conventional wisdom is that smaller but more frequent seismic events normally occur on the active fault planes and a log linear empirical relation ties their frequency to various magnitude range in a unit of time.

The National Standard of Canada #CAN3-N289.2-M81, 1981 for the seismic qualification of Candu nuclear power plants, requires site vicinity microearthquake monitoring. According to a stipulation of clause # 3.2.2. of this standard, in areas of unusual geological complexity or for which the historical seismicity information is considered unreliable or incomplete, it may be necessary to supplement the available data by operating a Microearthquake recording system in the vicinity of the site. International Atomic Energy Agency (IAEA) safety guide series # 50-SG-S1, 1979 also has a provision for microearthquake monitoring to assist in seismotectonic interpretation in the vicinity of the nuclear power plants.

East of Toronto Network

A program of site-specific seismic monitoring has been supported by the AECB since 1991, to investigate the feasibility of microearthquake detection in suburban areas of east Toronto in order to assess the rate of activity of local events in the vicinity of the nuclear power plants at Pickering and Darlington. The network coverage was also constrained by the inland extension of the Niagara-Pickering Linear Zone (NPLZ), an inferred crustal structure which passes virtually beneath the nuclear power plant site (Wallach and Mohajer, 1990). The NPLZ was characterized by a lack of seismicity in the present century, although a possible correlation with historical earthquakes in the Niagara Falls region may suggest potential activity of its southern segment (Mohajer 1993). Therefore, monitoring evidence of seismic activity or lack of it, along this structure, is extremely important to the site safety evaluation of the nuclear facilities. In particular, the NPLZ is known to coincide with the prominent CMBBZ terrane boundary, which may also join the Akron magnetic boundary, a candidate seismogenic feature to the south (Seeber and Armbruster, 1993).

After an extensive background noise (ambient vibration) survey of 25 test sites, a five station seismic network with 30 km aperture was deployed at the most favourable locations between the Pickering and Darlington nuclear power plants on Lake Ontario to the south and Lake Scugog to the north (Figure 1). The portable four channel digital recording system (PRS-4), developed by the GSC, was used in this survey. The threshold settings of the triggered mode of operation were optimized to obtain the best possible signal to noise -D24- • ratio for each site. The detection threshold obtained for two of the stations allow recording of local events of ML 0-2, a magnitude range which is usually not observed by regional seismic networks. An analysis of several thousand triggered signals resulted in a t discrimination of cultural disturbances from natural events, such as blasts at the St. Mary's | quarry near Darlington. The remaining 110 events, mostly at night time, can not be assigned to any source other than the natural release of crustal stresses (e.g. Figure 2). The j recurrence frequency of these microearthquakes shows a linear relationship which matches I that of larger events in the last two centuries in this region (Figure 3). This confirmation of the slope of such relations justifies a more confident extrapolation of the magnitude- j return period lines to the future which, collectively, serves as a statistical basis for the I prediction of the recurrence interval of the magnitude 6 and 7 earthquakes within the area (Figure 3). |

The preliminary results, to date, suggest that the stress is currently accumulating and is partially released within clusters of small earthquakes with episodic stress release ] characteristics. A high level of cultural noise in some of the stations reduced the number ' of multiple triggers (common recordings between stations). But there is sufficient information, based on the seismic phases (S and P travel times), to estimate the earthquake ] focus distances. It appears that most of the recorded events nucleated within the hanging- wall (eastern block) of the Niagara-Pickering segment of the CMBBZ. Microearthquakes at each cluster show identical wave form and spectra, which indicate a common origin for l each group. The closely spaced clusters were deduced from the fixed distance arrivals at one station, based on the S-P time plot versus number of events (e.g. Figure 4). *

The results obtained to date show that there is an average of 25 to 30 local events per year which were being missed in the past. These small shocks are below the detection « threshold of the regional and national networks and can provide useful information for J resolving the seismotectonic relations in the area. The initial attempt for deployment of the EOTN was the most cost effective way of demonstrating the feasibility of Microearthquake * monitoring in east Toronto area. This seismic network needs to be upgraded to provide a | clearer picture of the depth distributions, pattern recognition of the event clusters, and possibly infer the causative focal mechanisms of the local seismic activity at the vicinity of f the nuclear power plants. I I I

I I I 80 -45e Events Reported By GSC: 1983-1994

Events Recorded By GSC Common With EOTN Since 1991 O } Additional Events Reported By U.S. Seismic Networks: 1979-1989 Events Within EOTN Coverage (triggered at one or two stations; since 1991) Listed By Station A EOTN Stations

-44e LAKE ONTARIO

( y") •- Clarendon-Linden -43' Fault System

Figure 1 - Seismic events recorded by modern seismographic networks in the Lake Ontario Region over the last 15 years. -D26- I

Trace 5 660 T~|5 I 0.0

i.4e+03 -660 I 1700034000310006800035000 Ms. 1700034000510005800085008 Ms Û3"fi 1.9e+03 Trace 8 6.8e+03 Trace 11 • |8 Ti \ 0.0 0*0- - . . '. • . * *. . 1.9e+03 I I t \ -6.8e+03 17000340005100B6800035000 Ms 1708034080518006800085000 Ms 2.6e+03 I 0.0

2.6e+03 170B03400B510006800035000 Ms-3.le+83 1700034000510085888085000 Ms Trace 23 23

-2.5e+83 2.1e*03 170B0340B051080680BBB5000 Ms I 17000340005100B6800085000 Ms

Figure 2 - Typical plot of the velocity time history of the vertical component of local events recorded in one or two stations of the east of Toronto network (EOTN).

! I -D27-

t Microearthquakes 100- Monitored Since 1991 _X Seismicity Data Of \ ^? T ~T The Last 150 Years •k — — — — Future Earthquakes 10-

\ CD > i CD 1- a. CO +-> CD 0.1- K CD .Q log(N)=3.785-1.110M V \

13 V CD 0.01- I ' \\

3 . \ E X o \ 0.001 - \ \ ? 9N \ \

1 1 1 1 I I 3 4 Magnitude

Figure 3 - The magnitude-recurrence relationship for the western Lake Ontario region compared to that for the microearthquakes recorded east of Toronto since 1991, with extrapolation to the future probable earthquakes. S-P vs N, Solina, S-P<12s

Number Of Events Figure 4 - Microearthquakes recorded at Solina station showing clustering of events from specific distances. -D29-

References:

Mohajer, A.A. 1993. Seismicity and seismotectonics of the western Lake Ontario region: Géographie physique et Quaternaire, v. 47, p. 353-362.

Seeber, L. and J. G. Armbruster 1993. Natural and induced seismicity in the Lake Erie-Lake Ontario region: Reactivation of ancient faults with little neotectonic displacement. Géographie physique et Quaternaire, v. 47, p. 363-378.

Wallach, J.L. and A. A. Mohajer, 1990. Integrated geoscientific data relevant to assessing seismic hazard in the vicinity of Darlington and Pickering nuclear power plants: Proceedings, Canadian Geotechnical Conference, October 1990, Quebec City, P. 679-686. -D30- B AECB Workshop on Seismic Hazard Assessment in Southern Ontario i

Seismic Source Zone Modelling in Eastern Canada for National Building Code Applications 1

P.W. Basham Geological Survey of Canada 1

ABSTRACT

The best current hypothesis concerning the sources of the large earthquakes in I eastern North America is that most stable continental earthquakes occur through reactivation of relatively young rift faults that break the integrity of the continental crust. 1 Johnston, Coppersmith, et al. in their Electric Power Research Institute study of I worldwide analogues for the eastern North American continent showed that 71% of the seismicity of stable continental regions was associated with extinct intra-continental rifts I or continental passive margins (one-sided rifts). Further, all of the 17 earthquakes of 1 magnitude 7 and larger in their compilation are closely associated with the embedded rifts or passive margins. This hypothesis provides a good explanation for the locations of jj most of the larger eastern Canadian earthquakes. Two one-sided rifts are important for I earthquakes in southeastern Canada: the Atlantic margin, which was formed by the opening of the Atlantic Ocean in Triassic to Cretaceous times (the modern "passive" J margin); and the Iapetan paleo-margin along the ancient edge of the Precambrian I continent formed by the opening of the Iapetus (Proto-Atlantic) Ocean about 600-550 m.y. ago, and later partially overthrust by the Appalachians. I

Although this broad geological framework provides an excellent rationale for definition of "geological" earthquake source zones for seismic hazard assessment, I historical experience suggests that the rift features are not equally active along their I entire lengths. The geological processes causing these earthquakes will operate very slowly within continents, while for National Building Code purposes we are interested in I seismic design criteria that are valid for the next about 50 years. Thus we cannot let the | geology completely dominate historical experience, which is accomodated, for example, by restricting the extent of the Iapetan Margin source zone in the regions of Labrador f and Pennsylvania where there have been few significant historical earthquakes. I

Within this rift framework, the eastern portion of Lake Ontario is in the transition I zone between the Iapetan rifted margin zone and the stable continental core, comparable | to the Mesozoic Rifted Basin zone inboard of the Atlantic Rifted Margin (significant crustal extension but not complete rifting); the western portion of Lake Ontario and I Lake Erie are within the stable continental core, since they lie inboard of the Paleozoic I normal faults at Picton and Attica. The historical seismicity in and around these two lakes is concentrated in the Niagara-Attica, southern Lake Erie and Anna Ohio regions, I which are modelled as source zones for the historical-dominated model. With no known I geological controls on these earthquakes, one source zone for Lakes Ontario and Erie includes both lakes in a single zone with higher earthquake rates than the adjacent I background zones. ( -D31-

Seismic Source Zone Interpretation in Southern Ontario

A. A. Mohajer

University of Toronto / Seismican

Background and Rationale

Recent seismic hazard analyses for critical facilities have relied on interpretations developed by experts to assess the range of interpretations and associated earthquake potential in a variety of inter-and intra-plate tectonic environments (for example, LLNL, 1985; EPRI, 1986; GSC 1991). A significant aspect of these expert opinion studies is that the range of opinions and the uncertainties are large and the studies are focused primarily on the present understanding of the problem and often not on the gathering of new data to resolve or reduce the uncertainties.

The latest revisions of the International Atomic Energy Agency (IAEA) Safety guide 50-SG- Sl requires detailed investigations for all nuclear power plants sites, regardless of the presumed level of seismicity. Furthermore, a minimum horizontal ground acceleration value of 0.1 "g" is recommended for a nuclear power plant site anywhere in the world (Gurpinar , 1991).

The current long term seismic program, which is supported by the Atomic Energy Control Board (AECB), has focused on gathering, analyzing, and interpreting geoscientific data germane to assessing seismic hazard in vicinity of Darlington and Pickering. Coupled with this program has been the evaluation of existing data and interpretations developed by others, complemented by extensive interactions with the scientific community throughout the course of the work. Emphasis is placed on the possible implications of these data to earthquake source characterization and site ground motions. For example, following identification of the Niagara-Pickering fault zone, east of Toronto, studies were performed to estimate the maximum magnitude earthquake that could be generated along various segments of that structure..

Relating earthquake size and the fault characteristics, such as dimensions and displacement, is an important input any realistic hazard assessment. These relations are particularly crucial for areas with limited historical records or areas of current low seismicity, as the only means of estimating the maximum potential earthquake magnitude (Mx). In intraplate environments (stable continental regions), where active faults are not readily visible on the ground surface, indirect observations such as the extent of the aftershock zone and seismicity pattern recognition have been utilized to establish the size of credible future earthquakes ( Johnston, 1989; Wells and Coppersmith, 1994). -D32-

Local seismicity and source zones

The seismicity pattern of eastern Canada is presently dominated by the location of I moderate to large magnitude earthquakes which occurred in 1663, 1791, I860,1870, 1988, { 1925 near Charlevoix, 1732 near Montreal, 1988 in Saguenay, 1935 near Timiskaming, 1944 in Cornwall, 1929 in Attica New York, and 1982 Miramichi. There is an apparent I seismicity gap between the activity in the lower St. Lawrence river and that of the southern I and western Lake Ontario regions. This gap may be due to the short settlement history in southern Ontario or due to a relatively long return period between major earthquakes. I Well documented historical accounts of large destructive earthquakes in the Middle and Far 1 East reveal repeat intervals on the order of 3 to 9 centuries (Ambraseys, 1988; Mohajer and Nowroozi, 1979). It is, therefore, impossible to affirm a lower earthquake hazard potential 1 for southern Ontario on the basis of the absence of considerable activity. Nevertheless, in • a speculative seismic source model (Adams and Basham 1989) the St. Lawrence rift zone has been extended across Lake Ontario and Lake Erie to possibly join the Mississippi river 1 valley. If this seismotectonic model is ultimately confirmed by further investigations, such " as that undertaken by Thomas et al. (1993), then the whole Lake Ontario region would have the potential to experience large magnitude earthquakes, similar to other zones along the j St. Lawrence river, in the future.

Adams and Basham (1991 and 1995) have since postulated two new models, one limiting I the extension of the St. Lawrence rift to the Cornwall area and the latter defining a new zone along the Iapetan rift margin (Wheeler, 1991), the western boundary of which is along j the Clarendon-Linden fault zone. Nonetheless, it is also suggested that the CMBBZ has I a similar geophysical signature and has geological characteristics to be considered as part of the Iapetan rifted margin. j

In the western Lake Ontario region, over 58 small to moderate magnitude earthquakes have been compiled and used for seismotectonic interpretation. An update of the data file, in i addition to a recomputation of some of the locations where ever possible, has been carried | out by Mohajer (1987). More than 80% of the known events in the western Lake Ontario region are apparently confined within a 20 km wide zone between Toronto and Hamilton 1 (THSZ) which is also bounded by a magnetic and a gravity lineaments ( Mohajer, 1993). | The THSZ activity is due to it's favourable orientation with respect to the current principal compressive stress direction. The NPLZ, however, makes a wider angle with the principal I stress axis, and is subject to higher friction and consequently, a higher strain build up. | Release of this greater accumulation of energy would be expected to produce a large magnitude earthquake given sufficient time. Nevertheless, the important consideration 1 is to estimate the rate of deformation, if any, along this structure which needs extensive I geological and geophysical investigations.

The depth distribution of the existing earthquake files show a relatively shallow * concentration ranging from 1 to 20 km. Nevertheless, the better located group confirm that more than half of the local earthquakes are generated at more than 2 km depth, and I I I -D33- therefore, can not simply be assigned to surficial stress release phenomena.

Proposed Fault Segmentation and the Maximum Potential Earthquakes in Southern Ontario

One of the better established relations in observational seismology is that between earthquake magnitude and the fault rupture parameters. It is now common knowledge that a rupture does not usually propagate through out the entire fault length during a single event; but it tends to stop at sharp bends, or intersections with other faults, commonly called tectonic-knots. This has provided a basis for definition of fault segments and for estimation of their largest potential earthquake.

In order to estimate the Mx in southern Ontario it is necessary to define the segment lengths of the potentially capable faults in the area. Lack of observable surface faulting makes this task rather difficult for this region. Nonetheless, there is sufficient geological information to identify linear structural segments of the potentially active features in the area. This information comprises field geological data, geomorphology, bedrock surface topography, and geophysical potential field data. These data reveal bedrock discontinuities, structural intersections, or sharp changes in fault orientation along strike. These features are used to define the segmentation of the CMBBZ and other important fault systems which could impact on the Greater Toronto Area (GTA) hazard estimate. The measured fault segment lengths and their associated Mx are summarized in Table 1. Each segment is characterized by its associated largest potential earthquake using suggested empirical relations of Wells and Coppersmith (1994).

Deterministic Hazard Estimate for the Greater Toronto Area (GTA)

Areas of sparse seismicity with a relatively short history of earthquake lack a sufficient data base for statistical manipulations. The deterministic approach is, therefore, adapted to estimate potential hazards on the basis of a longer term seismotectonic environment within the current tectonic regime. This method was used to estimate the potential of each segment of the CMBBZ fault zone within the GTA as presented in Table 1. Using the attenuation relation of Atkinson and Boore, (1995), the peak ground acceleration of 300 cm/sec2 can be estimated, assuming a 30 km depth for a M7 earthquake as the maximum potential event. The corresponding ground velocity will be in the order 7.5 cm/sec . A typical response spectra based on the above deterministic input parameters is developed for 5% damping and is compared with the probabilistic estimates and with that of design response spectra of Pickering and Darlington nuclear power plants for reference. These values my be too conservative to be used for ordinary building however may serve as a minimum requirement for design or seismic margin studies of critical safety related facilities in the GTA, unless and until it could be proven beyond any reasonable doubt that the faults incorporated in this study are dormant. -D34- I

Probabilistic Hazard Estimate For the GTA I

The probabilistic method for seismic hazard assessment has been widely used since the mid- j 1970's because it provides a basis to compare the results with those of other natural and • man-made hazards and engineering design decisions (McGuire, 1977). In addition, this approach facilitated a consistent treatment of uncertainties by testing the sensitivity of the I results to various input parameters. This method may not be readily applicable to local • sources of the western Lake Ontario region, because there are a limited number of earthquake which could be considered above the lower bound magnitude limit (Mmin, I usually above M4.5 to 5). Therefore, the probabilistic hazard assessment in this region is usually influenced by distant sources with a history of larger magnitude earthquakes. A - simple probability function is therefore used, by assuming a statistical random distribution 1 (a commonly used Poisson process) for small to moderate magnitude earthquakes, to estimate the likelihood of the future earthquake activities in the GTA. All the shocks - recorded or reported within a 60 km radius around Toronto have been used to estimate the I annual risk of various magnitudes. The probability of exceedence of each earthquake magnitude in the next 50 years have been calculated and presented in Figure 4. I Concluding Remark I The existing data base is not sufficient to prove or disprove the seismogenic capability of | the local and regional faults passing near nuclear power plants in southern Ontario. The | workshop objectives as expressed in the program announcement of February 1995, indicate that " The AECB wishes to ensure that severe or damage frequencies to nuclear power fl plants are less than 10'5 per annum". This level of low risk may only be achieved either J by a conservative deterministic approach allowing a credible magnitude 7 earthquake near the site, or by site-specific detailed surface and subsurface geological and geophysical I investigations to demonstrate that the suspected faults are not capable of further movements I to generate damaging earthquakes in the future.

I I I I I I -D35-

Table 1 : Proposed Fault segmentation and earthquake potential for the western Lake Ontario region.

Fault zone, Location Orientation Length Mx segment # (end points) (km)

CMBBZ 1.1 Pembrook- Kamaniskeg Lake- NE 70 7.12 1.2 Dorset- (OGS map) E-W 95 7.32 (1.2) (Babtist-(aeromag) E-W 80 7.20 1.3 Balsam Lake- NNE 75 7.17 1.4 Claremont NNE 80 7.20

NPLZ 2.1 Pickering- Dundas fault cross- NNE 45 6.84 2.2 Lake Erie shore NE 70 7.12

THSZ 3.1 Toronto- Hamilton NE 65 7.08

GBLZ 4.1 Georgian Bay- Oshawa (uncertain) NW >130 7.5

WPLZ 5.1 Wilson- Dundas fault cross- NE 45 6.84 5.2 Port Hope NE 40 6.80 DFZ 6.2 Hamilton- NPLZ cross ENE 60 7.05 6.3 Port Hope cross- ENE 65 7.08 6.4 Pres'quile ENE 50 7.00 5 *tti

>:fc\i

to. lm s-;:

^••tfra^V S ?..'-&

' "-""maqnè"'1lineamen"1"'""1"1t'" ": ^

-/ Q Historical èùents .^ ^ W • S'^-: v^ jû' Nuclear generating station^! :\f 0-

( •,:-• 1

•> ..V • V'.-j I : "» .'--7 I J'IlVÎ

.<=••

;•

% •-

^'-i! 'YÂ -D37-

Probablilistic Seismic Hazard Assessment for the Greater Toronto Area

Probability of Exceedence ML Annual Risk in 50 years

4.0 1.9x10-1 99% 4.5 5.9x10-2 95% 5.0 1.7x10-2 57% 3 5.5 4.7x10- 21% 3 6.0 1.3x10- 6% 4 6.5 3.7x10" 1.8% 4 7.0 1.0x10" 0.5%

90H

70H

o in

4 5 6 7 Magnitude

Figure 1 - Probability of exceedence, in 50 years (P50), of a magnitude (M) earthquake within a radius of 60 km around Toronto is computed based on the random variable distribution (poisson process). -D38-

References:

Adams, J. and Basham, P.W., 1991. The seismicity and seismotectonics of eastern Canada; Chapter 14 in 1 Neotectonics of North America (ed.) D.B. Slemmons, E.R. Engdahl, M.D. Zoback, and D.D. Blackwell; Geological Society of America, Decade Map Volume 1: p. 261-276. ..

Adams, J., P.W. Basham, and S. Halchuk, 1995. Northeastern North American earthquake potential - new challenges for seismic hazard mapping, In current research, 1995-D. Geological Survey of Canada, i 91-99. I

Ambraseys, N.N., 1988. Magnitude-fault length relationships for earthquakes in the Middle East, Historical j Seismograms and Earthquakes, ed. W.H.K. Lee, pp. 309-10, Academic Press.

Atkinson, G.M. and D.M. Boore, 1995. Ground-motion relations for the eastern North America. BSSA, V.85, N.I, pp 17-30. I Basham, P.W., 1989. A Paleozoic-Mesozoic rift framework for seismic hazard assessment in eastern North j America; in Current Research, Part F; Geological Survey of Canada, Paper 89-1F, p. 45-50. '

Basham, P.F. and Adams, J., 1989. Problems of seismic hazard estimation in regions with few large I earthquakes: examples from eastern Canada; Tectonophysics, v. 167, p. 187-199. '

Basham, P.W., Weichert, D.H., Anglin, F.M. and Berry, M.J., 1985. New probabilistic strong seismic I ground motion maps of Canada; Bulletin of the Seismological Society of America, v. 75, p. 563-595.

EPRI, 1988. Engineering model of earthquake ground motion for eastern North America, EPRI NP-6074, 1 Electrical Power Research Institute, Palo Alto, California.

GSC, 1991. Geological Survey of Canada workshop on eastern Seismicity Source Zones for the 1995 | Seismic Hazard Maps, (comp.) J. Adams; Geological Survey of Canada, Open File 2437, p. 73-85 and 272-283. I

Gurpinar, A., 1991. IAEA safety guide with respect to NPP sites in areas of moderate seismicity. In proceedings of the International Conference on Seismic Hazard Determination in Areas With Moderate Seismicity. Saint-Remy-les-Chevreuse, France, October 22-23, 1991.

Johnston, A.C., 1989. The seismicity of'Stable Continental Interactions'; in Earthquakes at North Atlantic | Passive Margins: Neotectonics and Postglacial Rebound, (ed.) S. Gregersen and P.W. Basham; I Kluwer Academic Publishers Dordrecht, p. 299-327.

McGuire, R., 1976. FORTRAN computer program for seismic risk analysis, U.S. Geological Survey, Open- File Report 76-67. McGuire, R., 1977. Seismic design spectra and mapping proceedures using hazard analysis based directly on oscillator response, Int. J. Earthquake Eng. struct. Dyn. V.5, pp J 211-234. I j -D39-

Mohajer, A.A., 1987. Relocation of earthquakes and delineation of seismic trends in southern Ontario. AECB/MAGNEC contribution 87-01, also presented to the Geological Association of Canada, annual meeting, St John's, Nfld, May 23-25, 1988.

Mohajer, A. A. and A. A. Nowroozi, 1979; Tabas earthquake of 1978 in East-Central Iran. Geophysical Research Letters (AGU), Vol. 6 No.9, PP 689-692.

Mohajer, A. A. and A.A. Nowroozi, 1978; observed and Probable intensity zoning of Iran; Tectonophysics, Vol. 49. pp 149-160.

Mohajer, A. and Y. Bozorgnia, 1984. Ground acceleration distribution in Iran; a probabilistic approach. Proceedings of the 8th World Conference of Earthquake Engineering, San Francisco U.S.A. VOL. 1, PP. 45-52.

Mohajer, A.A., N. Eyles, C. Rogojina, 1992. Neotectonic faulting in Metropolitan Toronto: Implications for earthquake hazard assessment in the Lake Ontario region, in GEOLOGY, v. 20, p. 1003-1006, November, 1992.

Mohajer, A.A. (1993) Seismicity and seismotectonics of the western Lake Ontario region: Géographie physique et Quaternaire, v. 47, p. 353-362.

Thomas, R.L. J.L. Wallach, R.K. McMillan, J.R. Bowlby, S. Frape, D. Keyes and A.A. Mohajer, 1993. Recent deformation in the bottom sediments of western and southeastern lake Ontario and its association with major structures and seismicity, Géographie Physique et Quaternaire, v. 47, n. 3, pp. 325-335.

Wallach, J.L. and A.A. Mohajer, 1990. Integrated geoscientific data relevant to assessing seismic hazard in the vicinity of Darlington and Pickering nuclear power plants: Proceedings, Canadian Geotechnical Conference, October 1990, Quebec City, P. 679-686.

Wells, D.C. and K.J. Coppersmith, 1994. New empeirical relation among magnitude, rupture length, rupture width, rupture area, and surface displacement, BSSA, V. 84, N. 4, PP. 974-1002.

Wheeler, R.L., 1991. Earthquakes and the Iapetan passive margin in eastern North America; in Proceedings, Geological Survey of Canada workshop on eastern Seismicity Source Zones for the 1995 Seismic Hazard Maps, (comp.) J. Adams; Geological Survey of Canada, Open File 2437, p. 73-85 and 272- 283. C I -D40- The Tectonic Stress Field in Eastern Canada:- I its use in seismic source interpretation '

John Adams ' National Earthquake Hazards Program, I Geological Survey of Canada, I 1 Observatory Cres, OTTAWA, K1A 0Y3 Canada I Various observations (pop-ups, direct measurements, oilwell breakouts, earthquake focal i mechanisms, etc) of crustal stresses indicate that to a first approximation southeastern Canada | and adjacent U.S. are being compressed by the tectonic movement of the plates. These observations show a fair consistency with depth (at least down to 10 km - see below) and I between observation type. The two horizontal stress components are larger than the vertical (i.e., I a thrust regime dominated by reverse faulting), and the larger horizontal lies in the ENE octant. Such compressional tectonic stresses oriented parallel to plate motion are common to most | continents, and it is misleading to term those in southeastern Canada "High". I

Some stress indicators show "anomalous" directions, but it is uncertain if these are real, second I order features of the stress field rather than "noise" on the regional signal. The strongest • evidence for the deep stress field comes from the larger earthquakes: Timiskaming (1935, M6.2), Cornwall (1944, M5.8), Miramichi (1982, M5.7), Saguenay (1988, M6.5), and Mont-Laurier I (1990, M5.0) all had mechanisms consistent with the regional field, but Charlevoix 1925 did not. "

Earthquakes occur as deep as 27-33 km in the Saguenay and Charlevoix regions, and therefore | stress levels capable of causing earthquakes exist at these depths (i.e. the crust is brittle and not " plastic). Within the seismogenic crust there is equivocal evidence for a change in the variability . of stress orientation with depth: earthquakes shallower than 10 km are usually consistent with 1 ENE compression; deeper earthquakes are less consistent.

Near-surface stresses were different from the present during the déglaciation of southeastern I Canada, as is suggested by the orientation of some pop-ups and the cm-throw "postglacial faults". However, the state of stress at seismogenic depths at that time is unknown. Contemporary • changes in the stress field (e.g., postseismic changes near large earthquakes) have not been | studied in Canada; some spatial anomalies might represent such temporal stress anomalies - ghosts of large earthquakes? g

Where determined, the stress magnitudes are a few MPa to a few tens of MPa near surface, increasing to 100's MPa at depths of a few kilometres. Stress differences (and deviatoric • stresses) are much smaller than the absolute stresses. Stress drops during earthquakes have been fl suggested to be as high as 100 MPa (e.g., for Saguenay), but such "high stress drop" earthquakes can be modelled successfully with normal (ca. 10 MPa) stress drops, and indeed their time É histories preclude the compact rupture areas implied by 100 MPa stress drops. Measurements 1 in the top 2.5 km suggest that stress differences greater than 10 MPa exist almost everywhere

I -D41- in the Canadian Shield below about 1 km. Hence, on the basis of available stress levels, every part of the craton is potentially seismogenic, and the stress information in southeastern Canada forms no basis for delimiting seismic source zones. Thus other factors, such as high pore-fluid pressures or abnormally weak faults, must control the release of strain energy in the crust.

Considering the implications for the nature of the earthquake sources themselves, it seems that reverse faulting on NW-striking planes should be considered typical, reverse faulting on NE- striking planes occurs, but less frequently, and while NW-striking faults might be preferred, failure of reverse faults with other strikes is not precluded. While the reverse faults may involve a moderate component of strike-slip, pure strike-slip faulting is less likely than reverse faulting north of 44°N, but cannot be ruled out to the south (e.g., L. Erie basin).

Conclusions

• The Canadian craton is being compressed from the ENE octant. • Deviatoric stresses exceed those necessary to generate earthquakes throughout the shield. • Nothing currently known about the stress field suggests that any regions are especially "highly-stressed" and so pose a threat on this basis. • Earthquake sources modelled for hazard estimation should emphasize earthquakes involving reverse faulting on NW planes, but examine other reverse/strike-slip mechanisms to test the sensitivity of the results.

Prepared on 950530 for AECB Workshop on Seismic Hazard Assessment in southern Ontario, Ottawa 19-21 June, 1995. -DA 2- I Pop-ups and offset boreholes as geological indicators of earthquake-prone areas in intraplate eastern North America I

J.L. Wallach |

The assessment of seismic hazard in eastern Canada relies mainly upon the evaluation of historical and current seismic data. However, seismic data have only been collected for a I maximum period of about 200-300 years, which is too short to provide a representative | picture of where major seismicity has taken place in the geologically recent past. Consequently moderate to large earthquakes (M>4.5) are expected to occur, as they have J in the past, where no previous moderate to large earthquakes have been recorded. Among • the most recent examples of such earthquakes are the 1980 mb=5.1 Sharpsburg and the 1988 mbLg=6.5 Saguenay earthquakes. The occurrence of these and others, in unexpected I areas, underscores the need to identify areas susceptible to major seismicity, particularly ™ where seismicity is not known to have occurred in the past.

Finding areas that may be susceptible to potentially damaging earthquakes would be * facilitated if earthquake-related features could be recognized on the surface. The most obvious would be fresh fault scarps, such as the one which formed during the Ungava I earthquake of December 25, 1989 (Lamontagne, personal communication). Unfortunately such fault scarps in eastern North America are either very rare, or have not been j recognized. Other surficial geological indicators of paleoseismicity are liquefaction I features, though they are not usually obvious on the surface. Pop-ups and offset boreholes, however, are surficial neotectonic structures, which are generally accessible and easily recognized. I Stresses have been measured in eastern North America, examples of which are | summarized here. Around Lake Ontario, the maximum principal stress is oriented | generally east-northeast (Lindner, 1985), though trends of 310°, 325°, and 350° were measured at depths of 8 to 22 m (Dames and Moore, 1973; 1974a). At the Darlington f excavation site, stress tests were conducted in a borehole, 303 m deep (Haimson and Lee, | 1980). In the sedimentary rock sequence, the orientation of the maximum horizontal compressive stress ranges from 070°, at a depth of 74.7 m, to 032°, at a depth of 207.4 m. I In the underlying Precambrian basement, to a depth of 299.5 m, orientations of 021°, 025° ' and 024° were measured. 1 In the epicentral area of the 1982 Miramichi earthquake, the average P-stress orientation is ' 055° (McKay et al., 1985). Measurements in the Roblindale Quarry in southern Ontario, g revealed an average P-stress trend of 050°, in the limestone, and 066° in the underlying I gneissic granite (McKay and Williams, 1988). The predominant trend in northern Michigan is 075° (Kim and Smith, 1980), whereas in eastern Canada, the Appalachian . Basin and the Illinois Basin the respective orientations are 054°±7°, 058°±8°, and 271 °±5° I (Plumb and Cox, 1987). • f I I -D43-

Moderate to large earthquakes are not uncommon in eastern North America. Among the largest are the 1811-1812 New Madrid (mb=7.2, M^.2 on December 16, 1811) and the 1988 Saguenay earthquakes (mbLg=6.5). Several have been spatially associated with the St. Lawrence valley system, a major fault system named by Kumarapeli and Saull (1966), and its projected extension upstream through the lower Great Lakes toward New Madrid. Others have been centered in the Maritime Provinces of eastern Canada, in northern Canada, and in the northeastern, southeastern and north-central USA. The hypocentral depths of the moderate to large events are scattered throughout the thickness of the continental crust ranging from <4 km for the Ungava earthquake (Wetmiller, personal communication) to 29±1 km for the Saguenay earthquake (Du Berger et al., 1991) (Table 1).

Focal mechanisms show a pattern of fault movements consistent with the current stress field in which the greatest principal horizontal compressive stress is commonly oriented in the northeast quadrangle. For example, the preferred nodal plane for the Saguenay earthquake is oriented 325°/67°NE, and the mechanism is reverse faulting with a small strike-slip component (North et al., 1989; Roy, 1991). Fault plane solutions from ten small to moderate earthquakes in southern Québec and northern New York State show, with one exception, reverse faulting along northwest striking nodal planes (Yang and Aggarwal, 1981). In southeastern New York State and adjacent New Jersey fault plane solutions for ten seismic events, ranging in depth from 1-10 km, show the predominant mechanism to be reverse faulting, but for each mechanisms at least one, if not both, of the nodal planes is oriented north to northeast (Aggarwal and Sykes, 1978). In the Charlevoix Seismic Zone, microseismic activity in 1974 occurred commonly along north-northeast to northeast oriented reverse faults, with or without strike-slip components (Lamontagne, 1987). Fault plane solutions compiled for several events in southeastern Canada all show reverse fault mechanisms with, or without, a minor strike-slip component on either northwest- or northeast-striking surfaces (Adams and Basham, 1989).

Recent reverse faulting, at the surface, is marked by offset boreholes, consequent upon slip along bedding or foliation surfaces. These structures have been documented in nearly flat-lying limestone in the Saguenay-Lac St-Jean region (location 1), in the Ottawa-Hull area (location 2), north of Napanee, Ontario (location 3), and in the excavation of the Darlington NGS (location 4). The respective directions of slip seen in the aforementioned examples are 074° at location 1, 070° at location 2, 320° and 075° at location 3, and 230° at location 4. Reverse displacement of offset boreholes was also seen on the Canadian Shield in southwestern Québec. There slip occurred on foliation surfaces oriented 040719°SE and 016718°E with the hanging walls having moved in the directions of 309° and 290°, respectively. Borehole offsets display evidence of recent reverse faulting in southeastern Connecticut with slip of the upthrown blocks having been to the southeast on surfaces striking about 225° and dipping from 5° to 30° to the northwest (Block et al., 1979).

Pop-ups are elongate anticlinal structures which form in rocks with horizontal planes of weakness enabling the upper units to decouple from the lower ones. They may occur as -D44-

individual structures with a uniform trend or they may be composed of differently • oriented, but connected segments (Wallach and Chagnon, 1990). At some locations in | eastern North America pop-ups trend exclusively northwest. At others they may trend north-northeast to northeast and, still elsewhere, both northwest and northeast to east- i northeast oriented pop-ups may coexist (Wallach and Chagnon, 1990). Thus far they have | been documented almost exclusively in sedimentary rocks, with the sole exception being the pop-ups in a massive, horizontally fractured diorite in the epicentral area of the 1982, I Miramichi earthquake. Pop-ups have been recognized in open fields, stream valleys, lake I bottoms and on quarry floors (Dames and Moore, 1974b; Wallach and Chagnon, 1990; Thomas et al., 1993). J

In eastern North America pop-ups have been identified in a broad belt which extends in a generally northeast direction from north-central Kentucky to New Brunswick. Most of I them occur along the St. Lawrence fault zone, and its projected extension into the lower * Great Lakes, as proposed by Adams and Basham (1989). They are particularly abundant in the area of western Lake Ontario, with several occurrences in the heavily populated areas I of southern Ontario and western New York State. "

In glaciated areas pop-ups have always been interpreted as being post-glacial in age (e.g. I Dames and Moore, 1974b). In quarries pop-ups form following excavation of the overlying rock, thus they are obviously young. Pop-ups which are found in open fields, m stream valleys and on lake bottoms are also considered to be post-glacial structures, but I their ages relative to, at least, the last glaciation should be proved on a case-by-case basis.

Pop-ups that rise above the adjacent rock surface are generally considered to have formed | subsequent to the last glaciation, which terminated in southern Ontario about 12,000 years ago. This interpretation is logical, but should also be regarded with caution because high I fluid pressures could be generated at the base of an ice sheet. Consequently it is I conceivable that some surficial pop-ups may have formed prior to, and survived the effects of, the last glaciation. 1

Various mechanisms have been proposed to explain the formation of these structures. They include the removal of salt and gypsum from underlying strata (Gilbert, 1892), and I isostatic rebound (Adams, 1989). Many others, however, have explained the formation of " pop-ups as resulting from horizontal compression, which is the explanation favored by this writer. 1

The presence of seven pop-ups in the Roblindale Quarry of southern Ontario led to | measurements of stress in both limestone and the underlying Precambrian basement (e.g. I McKay and Williams, 1988). The average P-stress trend is 124°, with 70 of the 85 measured axes trending nominally northeast, which is approximately orthogonal to the « average pop-up orientation. 1

From stress measurements throughout eastern North America, it is clear that the I predominant orientation of the maximum horizontal stress is generally northeast (e.g. I I -D45-

Haimson and Lee, 1980; Kim and Smith, 1980; Lindner, 1985; Plumb and Cox, 1987 and McKay and Williams, 1988). In determining the orientations of pop-ups, particularly those composed of differently oriented, but connected segments, all segments were measured. About two-thirds of the 337 pop-ups and pop-up segments trend northwest-southeast.

The presence of mutually perpendicular pop-up and P-stress orientations, both on local and regional scales, strongly suggest that pop-ups form in response to horizontal compression. However, besides the presence of high horizontal stress there is a need for the upper rock mass to be able to decouple from the lower. Thus the presence of horizontal planes of weakness is a prerequisite for the formation of pop-ups. If planes of weakness do not exist, or are steeply dipping in which case they effectively do not exist, pop-ups will not form even if stress conditions are, otherwise, favorable. But for two exceptions, pop-ups have only been documented in essentially flat-lying sedimentary rocks.

In intraplate eastern North America, pop-ups have been recognized in several areas in which moderate to large earthquakes have occurred (Table 1). Pop-ups have also been observed in proximity to surface rupture associated with large magnitude earthquakes in Iran (Nowroozi and Mohajer-Ashjai, 1985), northern Algeria (Meghraoui, 1992) and on the Ungava Peninsula (Adams et al., 1992 and poster session).

Both pop-ups and earthquakes are stress-relief phenomena which form in response to the current ambient stress field. Therefore, it is logical that evidence of high horizontal compressive stress on the surface in a particular area would imply the presence of similar stress conditions, and an accumulation of strain energy, at greater depths in the same area. Due to inhomogeneities in the earth's crust, the sudden release of variable amounts of stored strain energy, as a consequence of co-seismic fault movements at different depths, would generate a sequence of earthquakes of varying magnitudes at those depths. This may be expressed either as aftershocks of a single event or, in the course of time, as several distinct earthquakes. Focal depths of earthquakes fall within a depth range of from 4 to 29 km (Table 2). Furthermore, there are documented cases of surficial seismicity resulting from the formation of a pop-up (Swinford, 1985).

In summary recent moderate to large eathquakes in eastern North America have been centered in areas in which none had previously been documented, and at depths ranging from 4 to 29 km. Fault movements responsible for earthquakes at those depths are kinematically consistent with both the measured contemporary stress field, and with recently formed pop-ups and reverse faults, expressed by displaced boreholes, visible at the surface.

The similar geometric, kinematic and dynamic conditions both near the surface and at depth, and the depth distribution of moderate to large earthquakes, indicate that the contemporary tectonic processes acting on the surface are not de-coupled from those operative at depth. Adams et al. (1991b) also implied as much by stating, "Together with the unusually deep (29 km) Saguenay earthquake of 1988, the Ungava earthquake -D46- I I indicates that almost the entire thickness of continental shields and not just the deeper 1 parts are (sic) capable of generating damaging earthquakes." Bell (1994) stated "Common 1 stress orientations testify to the widespread attachment between the Paleozoic sediments of southern Ontario and the underlying Canadian Shield". Therefore, if there is evidence of I high horizontal compressive stress at the surface, it is logical to expect the same at depth. 9 Consequently, for the purposes of seismic hazard assessment, no distinction should be made between pop-ups and small near-surface earthquakes on the one hand, and moderate I to large, generally deeper earthquakes on the other. Instead, because pop-ups and shallow, " small-magnitude earthquakes are stress-relief phenomena, they should be regarded as signals of the probable accumulation of strain energy at depth and its eventual relief to I produce a large-magnitude earthquake. "

The existence of lake-bottom pop-ups immediately south of Toronto, and the presence of I small to moderate magnitude earthquakes in the western Lake Ontario area, suggest that it is not unreasonable to expect a large magnitude earthquake in the region encompassing « the nuclear power plants. This is despite the absence of a previously documented, large I magnitude earthquake in that area. i

i i

I 1 I I -DAT-

References

Adams, J., 1989. Postglacial faulting in eastern Canada: nature, origin and seismic hazard implications. In Môrner, N-A and Adams, J. (eds.) Paleoseismicity and Neotectonics, Tectonophysics, 163, 321-331.

Adams, J. and Basham, P., 1989. The seismicity and seismotectonics of Canada east of the Cordillera. Geoscience Canada, 16, 3-16.

Adams, J., Wetmiller, R.J., Drysdale, J., and Hasegawa, 1991a. The first surface rupture from an earthquake in eastern North America. In Current Research, Part C; Geological Survey of Canada, Paper 91-1C, 9-15.

Adams, J., Wetmiller, R.J., Hasegawa, H.S. and Drysdale, J., 1991b. Nature, 352, 617- 618.

Adams, J., Percival, J.A., R.J. Wetmiller, Drysdale, J.A. and Robertson, P.B., 1992. Geological controls on the 1989 Ungava surface rupture: a preliminary interpretation. In Current Research, Part C; Geological Survey of Canada, Paper 92-1C, 147-155.

Aggarwal, Y.P. and Sykes, L.R., 1978. Earthquakes, faults and nuclear power plants in southern New York and Northern New Jersey. Science, 200, 425-429.

Bell, J.S., 1994. Attached and detached stress regimes in Canadian sedimentary basins. Geological Association of Canada, Program with Abstracts, 19, A10.

Block, J.W., Clement, R.C., Lew, L.R. and De Boer, J., 1979. Recent thrust faulting in southeastern Connecticut. Geology, 7, 79-82.

Dames and Moore, 1973. Standardized nuclear unit power plant system SNUPPS-initial report. Rochester Gas and Electric Corporation, Addendum Questions, Item No. 5,1 p.

Dames and Moore, 1974a. In-situ rock stress measurements. Preliminary Safety Analysis Report. Somerset Nuclear Generating Station, New York State Electric and Gas Company, Appendix 2A, 13 p.

Dames and Moore, 1974b. Seismo-tectonic conditions in the St. Lawrence River Valley Region, Phase 1,1973 Geologic Investigations. Cranford, New Jersey. I -D48-

Du Berger, R., Roy, D.W., Lamontagne, M., Woussen, G., North, R.G. and Wetmiller, I R.J., 1991. The Saguenay (Quebec) earthquake of November 25, 1988: " seismologic data and geologic setting. In Mareschal, J.-C. (éd.), Intraplate Deformation, Neotectonics, Seismicity, and the State of Stress in Eastern North I America, Tectonophysics, 186, 59-74. *

Gilbert, G.K. 1892. Post-glacial anticlinal ridges near Ripley, N.Y. and near Caledonia, J N.Y. Proceedings of the American Association for the Advancement of Science, 249-250. |

Haimson, B.C., and Lee, CF., 1980. Hydrofracturing stress determinations at Darlington, Ontario. In: Underground Rock Engineering, 13th Canadian Rock Mechanics Symposium (The H.R. Rice Memorial Symposium), Toronto. The Canadian I Institute of Mining and Metallurgy, CM Special Volume 22, 42-50.

Kim, K. and Smith, C.C., 1980. Hydraulic fracturing stress measurements near the | Keweenaw Fault in Upper Michigan. In Underground Rock Engineering, 13th Canadian Rock Mechanics Symposium (The H.R. Rice Memorial Symposium), I Toronto. The Canadian Institute of Mining and Metallurgy, CIM Special Volume • 22, 24-30. 1 Kumarapeli, P.S. and Saull, V.A., 1966. The St. Lawrence valley system: a North J American equivalent of the East African rift valley system. Canadian Journal of Earth Sciences, 3, 639-658. I

Lamontagne, M., 1987. Seismic activity and structural features in the Charlevoix region, . Quebec. Canadian Journal of Earth Sciences, 24, 2118-2129. j

Lamontagne, M., Hasegawa, H.S., Forsyth, D.A., Buchbinder, G.G.R. and Cajka, M.G., | 1994. The Mont Laurier, Québec earthquake of 19 October 1990 and its I seismotectonic environment. Bulletin of the Seismological Society of America, 84, 1506-1522. I Lindner, E.N., 1985. In situ stress indications around Lake Ontario. In: 26th US Symposium on Rock Mechanics, Rapid City South Dakota, 575-590. 1

Mauk, F.J., Christensen, D., and Henry, S., 1982. The Sharpsburg, Kentucky earthquake, 27 July 1980: Main shock parameters and isoseismal maps. Bulletin of the I Seismological Society of America, 72, 221-236. •

McKay, D.A., Williams, J.B., Bowlby, J.R. and Grass, J.D., 1985. Miramichi epicentral I area-in situ stress pilot project. Ontario Hydro Research Division, Report No. 85- ^ 185-K, 88 pp. plus appendices. . -D49-

McKay, D.A. and Williams, J.B., 1988. Roblindale Quarry in situ stress measurements- Phase 2. Ontario Hydro Research Division, Report No. 88-113-P, 90 pp.

Meghraoui, M, 1992. Active folds and their seismotectonic implications in the Tellian Mountains of Algeria. In Môrner, N-A., Owen, L.A., Stewart, I. and Vita-Finzi, C. (eds.), Neotectonics-Recent Advances: Abstract Volume. Quaternary Research Association, Cambridge, England, 36.

North, R.G., Wetmiller, R.J., Adams, 1, Anglin, F.M., Hasegawa, H.S., Lamontagne, M., DuBerger, R., Seeber, L. and Armbruster, J., 1989. Preliminary results from the November 25, 1988 Saguenay (Quebec) earthquake. Seismological Research Letters, 60, 89-93.

Nowroozi, A.A. and Mohajer-Ashjai, A., 1985. Fault movements and tectonics of eastern Iran: boundaries of the Lut plate. Geophysical Journal of the Royal Astronomical Society, 83, 215-237.

Nuttli, O.W., 1973. The Mississippi Valley earthquakes of 1811 and 1812: intensities, ground motion and magnitudes. Bulletin of the Seismological Society of America, 63, 227-248.

Plumb, R.A., and Cox, J.W., 1987. Stress directions in eastern North America determined to 4.5 km from borehole elongation mesurements. Journal of Geophysical Research, 92, 4805-4816.

Roy, D., 1991. The Saguenay earthquake and geology. In Adams, J. compiler, Proceedings, Geological Survey of Canada Workshop on Eastern Seismicity Source Zones For The 1995 Seismic Hazard Maps, Geological Survey of Canada Open File 2437, 351-355.

Seeber, L. and Armbruster, J.G., 1986. A study of earthquake hazards in New York State and adjacent areas: Final report covering the period 1982-1985. U.S. Nuclear Regulatory Commission, NUREG/CR-4750, 98 p.

Swinford, M., 1985. Stream anticlines. Ohio Geology Newsletter, Ohio Department of Natural Resources, Winter, 1985, 8 pp.

Thomas, R.L., Wallach, J.L., McMillan, R.K., Bowlby, J.R. Frape, S., Keyes, D and Mohajer, A.A. 1993. Recent deformation in the bottom sediments of western and southeastern Lake Ontario and its association with major structures and seismicity. In Wallach, J.L. and Heginbottom, J.A., (eds.) Neotectonics Of The Great Lakes Area, Géographie physique et Quaternaire, 47, 325-335.

Wallach, J.L. and Chagnon, J-Y., 1990. The occurrence of pop-ups in the Quebec City area. Canadian Journal of Earth Sciences, 27, 698-701. -D50-

Weston Geophysical Corporation, 1986. Investigations of confirmatory seismological and j geological issues, northeastern Ohio earthquake of January 31, 1986, 70 p. plus J diagrams.

Wetmiller, R.J., Adams, J., Anglin, F.M., Hasegawa, H.S. and Stevens, A.E., 1984. I Aftershock sequences of the 1982 Miramichi, New Brunswick, earthquakes. Bulletin of the Seismological Society of America, 74, 621-653. I

Yang, J.P. and Aggarwal, Y.P., 1981. Seismotectonics of northeastern United States and adjacent Canada. Journal of Geophysical Research, 86, 4981-4998. I I I I I 1 I I I -D51-

Table 1 Locations of known pop-ups and spatially related (<80 km) earthquakes (M>4.5) Pop-ups Earthquakes Location Latitude1 Longitude1 Latitude1 Longitude Magnitude CANADA Ungava Peninsula, Que. 60.12°N 73.60°W 60.12°N 73.60°W Ms=6.3 Miramichi, N.B. 47.00°N 66.60°W 47.00°N 73.60°W mb=5.7 Québec City, Que. 46.80°N 71.30°W St-Marc des Carrières, Que. 46.70°N 72.10°W ——•••• Terrebonne, Que. 45.70°N 73.60°W 45.50°N 73.60°W mb=5.6-6.0 2 St. Eustache, Que. 45.50°N 73.60°W 45.50°N 73.60°W mbs:5.6-6.0 Aylmer, Que. 45.40°N 75.90°W 45.40°N 75.40°W M*5.0 Ottawa, Ont.2 45.30°N 75.90°W 45.40°N 75.40°W M=5.0 Hallville, Ont. 45.10°N 75.60°W Cornwall, Ont. 45.00°N 74.70°W 44.98°N 74.90°W M=5.9 Georgian Bay area, Ont. 44.70°N 79.50°W Balsam Lake area. Ont. 44.60°N 78.80°W Burleigh Falls, Ont. 44.50°N 78.30°W Roblin, Ont. 44.40°N 77.00°W Youngs Point, Ont. 44.40°N 78.20°W Kingston, Ont. 44.30°N 76.50°W Wellman, Ont. 44.20°N 77.70°W Prince Edward County, Ont. 44.00°N 77.00°W near Toronto, Ont. 43.50°N 79.60°W near Burlington. Ont. 43.30°N 79.90°W Port Colborne, Ont. 42.90°N 79.20°W USA Chippewa Bay, N.Y. 44.40°N 75.80°W Alexandria Bay, N.Y. 44.30°N 75.90°W Omar, N.Y. 44.30°N 76.00°W Gasport, N.Y. 43.20°N 78.60°W 42.90°N 78.30°W M=4.63 Niagara Falls, N.Y.2 43.10°N 79.00°W 42.90°N 78.30°W M*4.63 Alden, N.Y. 42.90°N 78.50°W 42.84°N 78.24°W mb=5.2 2 Caledonia, N.Y. 42.90°N 77.90°W 42.84°N 78.24°W mb=5.2 2 Vaiysburg, N.Y. 42.70°N 78.30°W 42.84°N 78.24°W mb=5.2 2 Letchworth Gorge, N.Y. 42.60°N 78.00°W 42.84°N 78.24°W mb=5.2 Ripley, N.Y. 42.30°N 79.70°W northeastern Ohio 41.65°N 81.16°W 41.65°N 81.16°W mb=4.9 southwestern Ohio 39.40°N 84.70°W east-central Kentucky 37.80°N 84.40°W 38.18°N 83.94°W mh=5.2 '-Approximate latitudes and longitudes, rounded off to nearest 0.10° 2-Indented data refer to a second set of pop-ups proximal to the same earthquake as the previous set. 3-M=5.0. Geological Survey of Canada files.

10 -D52-

Table 2 Depths of moderate to large earthquakes in eastern North America (M>4.5) I Lat. Long. Magnitude Depth Date Location Reference I (N) (W) (km)

60.12° 73.60° Ms= 6.3 £4.0 12/25/1989 Ungava, Québec Adams et al., 1991a 1

48.12° 71.18° mbLg=6.5 29.0±1.0 11/25/1988 Saguenay, Québec Du Berger et al., 1991

47.00° 66.60° mb= 5.7 7.0±3.0 01/09/1982 Miramichi, N.B. Wetmilleret al., 1984 j

46.47° 75.59° mbLg=5.0 ll.Oil.O 10/19/1990 Mont Laurier, Québec Lamontagne et al., 1994. I

43.94° 74.26° ML= 5.1 *6.0±1.0 10/07/1983 Adirondack Mtns.N.Y. Seeber& Armbruster, 1986 j

41.65° 81.16° mb= 4.9 4.0-5.0 01/31/1986 Leroy, Ohio Weston Geophysical, I986 \ 38.18° 83.94° mu= 5.1 15.5±2.6 07/27/1980 Sharpsburg, Ky Mauk et al., 1982 mb Body-wave magnitude using P or S peak amplitude of regional events. I mbL Body \va\'e magnitude of regional earthquakes in eastern North America using 1 Hz Lg amplitude (Nuftli, 1973). I

Ms Surface-wave magnitude •

ML Local magnitude ' I 1

H O

-D53- Hazards Associated with Seismically Active Linear Zones Christine A. Powell, Department of Geology, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-3315

Seismic activity in eastern North America is not randomly distributed but tends to occur in distinct zones. Many of these zones are linear and appear to coincide with ancient intracontinental rifts and other zones of weakness associated with rifted margins. For example, almost all significant earthquakes in the continental portion of southeastern Canada can be spatially associated with the Paleozoic rift system along the St. Lawrence River (Adams and Basham, 1991). The two most active zones of seismicity in the eastern United States, the New Madrid zone and the eastern Tennessee zone, are also associated with Paleozoic rift and continental margin structures (Johnson et. al, 1985; Bollinger et al., 1991). The purpose of this paper is to discuss the geophysical and geological features of one of these linear zones, the eastern Tennessee seismic zone (ETSZ), and the seismic hazard it may pose. Similarities can be found between the ETSZ and seismic zones in southeastern Canada. The eastern Tennessee seismic zone (ETSZ) is a pronounced seismic feature in the central and southeastern United States (Johnston et al., 1985; Bollinger et al., 1991, Powell et al., 1994). Close to 900 events were recorded during the period 1981 to 1991. The maximum historical event I is approximately magnitude 5 and the largest instrumentally recorded event is the 1973 m,, 4.6 earthquake just south of Knoxville, Tennessee (Bollinger et al., 1976). The zone trends NE I through eastern Tennessee and parts of North Carolina and Georgia. Spatial dimensions are roughly 300 km long by 50 km wide; focal depths range between 5 and 25 km. Using seismic I moment release per unit crustal volume (25,000 km^ by 20 km depth) over the last decade as a I basis of comparison, the ETSZ has produced the second highest release of seismic strain energy in the United States east of the Rocky Mountains. Only the New Madrid zone has been more tseismogenic. New England has experienced more earthquakes but events in eastern Tennessee occur in a more concentrated zone. Seismicity in the ETSZ has been monitored by the Southern Appalachian Regional Seismic | Network (S ARSN) since 1981 and by Tennessee Valley Authority (TVA) stations since 1983. Station spacing is 45 to 80 km throughout eastern Tennessee and western North Carolina and I station locations have not varied. Event locations are accurate and relatively unbiased (Bollinger et al., 1991). Recently compiled one and three-dimensional velocity inversions based upon Ï regionally recorded P and S arrival times show little deviation from the starting model and relocated hypocenters display approximately the same spatial distribution as the original locations (Vlahovic I et al., 1995). Except for the southernmost part in Georgia, the ETSZ lies within the network. i -D54-

Thus, the clustered nature of se'"smicity within the ETSZ cannot be attributed to nonuniform station J distribution, an inadequate crustal velocity model, or to variable detection capabilities. ETSZ epicenters lie within the Valley and Ridge and Blue Ridge provinces. These I provinces are thrust and fold complexes underlain by a master decollemont (Cook et al., 1983). Maximum depth to the decollemont within the ETSZ is approximately 5 km. In contrast, mean Jj focal depth within the ETSZ is 15 km (Johnston et al. 1985; Teague et al., 1986; Bollinger et al., 1991); thus, most of the earthquakes occurred in crystalline basement rocks of inferred Grenville- age, and are not associated with the decollemont or overlying, detached rocks. ï Instrumentally located epicenters in the ETSZ generally lie close to and east of the New York- « Alabama (NY-AL) aeromagnetic lineament between latitudes 34.3° and 36.5°N. Epicenters also lie | west of the Clingman aeromagnetic lineament. Both lineaments trend to the northeast; the NY-AL lineament extends for more than 1,600 km from the northeast corner of Alabama to Albany, New I York while the Clingman lineament extends for roughly 1000 km from northwest Georgia to Maryland. The long wavelengths of the NY-AL and Clingman lineaments suggest that they are I associated with structural or mineralogical variations in basement rocks (King and Zietz, 1978; Nelson and Zietz, 1983). These lineaments define a basement block with a distinctly different • aeromagnetic signature (Nelson and Zietz, 1983; Hatcher et al., 1987). This block has been named • the Ocoee by Johnston et al. (1985). Between the lineaments, the signature is characterized by • relatively low intensities and numerous gradients trending roughly N15°E. Outside of the block, J magnetic features have higher amplitudes and trend roughly northeast-southwest. The NY-AL magnetic lineament is coincident with a distinct gradient in the Bouguer and isostatic gravity maps I in eastern Tennessee; anomalously low gravity values are present to the northwest of the magnetic lineament. The seismogenic block between the NY-AL and Clingman lineaments is characterized | by slightly negative gravity values. The implication is that the seismogenic block consists of high density, low susceptibility rocks, relative to rocks located northwest of the magnetic lineament. 1 Earthquakes within the ETSZ cannot be attributed to known faults. Focal mechanism solutions ' indicate sub-vertical strike-slip faulting with either right-lateral motion along north-south striking g planes or left-lateral motion along east-west striking planes (Johnston et al. 1985; Teague et al., i 1986; Bollinger et al., 1991), consistent with the regional stress field (Zoback, 1992). The axis of - maximum horizontal compressive stress (si) is oriented roughly northeast-southwest in eastern fl Tennessee (Zoback, 1992). Neither set of fault planes strikes parallel to the trend of the ETSZ; fault planes are oriented roughly 45° to the overall northeasterly trend of the seismic zone. Results I of a new compilation of focal mechanism solutions, using the one-dimensional velocity model determined by Vlahovic et al. (1995), are similar to those obtained previously. 1 Most strike-slip fault zones trend at large angles to si. For example, the low stress-bearing ' San Andreas fault system trends almost 90° to si (Mount and Suppe, 1987) and the south-west 1 a -D55- trending branch of the New Madrid seismic zone trends roughly 30° to 40° to si (Mitchell et al. 1991). In contrast, the ETSZ trends at a small angle (5 to 15°) to si. The ETSZ appears to have narrowed to its present horizontal dimensions within the last 15 to 20 years (Bollinger et al., 1991). Seismicity for 1698 to 1977 was equally distributed between eastern Tennessee and western North Carolina whereas after 1977 seismicity was concentrated in eastern Tennessee. One possibility is that the apparent narrowing of the zone is due to increased earthquake detection and location capabilities provided since 1981 by SARSN and TVA stations. This possibility was investigated by considering the spatial and temporal distribution of felt earthquakes: such intensity reports are independent of seismic station operation and can be extended to the period before instrument observations. Seismic activity appears to have narrowed over the last twenty years on the basis of the felt earthquake distribution, despite the large error bars associated with events. Event relocations, currently in progress, will help clarify this issue. The ETSZ may be an evolving seismic zone in which slip on north- and east-striking surfaces is slowly coalescing into a northeast-trending zone. Whether the associated faults are new or represent a reactivation of ancient faults that were partially healed during prolonged past inactivity is uncertain. Seismicity has occurred most recently in the Ocôee basement block. Orientations of focal planes within the block are appropriate for failure in the regional northeast-southwest oriented, midplate stress field. These fault orientations differ from the northeast-southwest trend of the seismic zone because this trend is controlled by the subsurface geometry of the Ocoee block. Concentration of seismicity along the NY-AL magnetic lineament is also controlled by the geometry of the Ocoee block. Thus, the ETSZ represents seismic activity that results from the regional stress field and is coalescing near the juncture between a relatively weak, seismogenic block (Ocoee block) and relatively strong crust to the northwest. Seismicity levels may be high in eastern Tennessee because this portion of the Ocoee block contains preexisting faults, foliations, or compositional layerings that are oriented favorably for failure in the present-day, regional stress field. Faults within the Ocoee block may date from rifting associated with formation of the Iapetos Ocean (Johnston et al., 1985) and may have been modified by Paleozoic compression or Mesozoic extension. Similarly, earthquake activity in the adjacent Giles County, Virginia seismic zone which is remarkably similar to the ETSZ in focal depths, focal mechanism solutions, and association with the Ocoee block, has been attributed to compressional reactivation of Iapetan normal faults (Bollinger and Wheeler, 1983). Eastern Tennessee also corresponds to the portion of the NY-AL lineament with the steepest gradient and the most distinct separation of magnetic basements. Basement crust northwest of the lineament may be strengthened by the presence of mafic rocks associated with an inferred Keweenawan-age (1100 Ma) rift (Keller et al., 1982). The presence of mafic rocks is suggested by magnetic, gravity, and seismic (Owens et al., 1984) data and the inferred rift runs parallel to -D56- • and just northwest of the NY-AL lineament in Tennessee The proposed rift does not appear to be ™ seismogenic although a few earthquakes have been associated with its western boundary. _ A possibility exists that deformation within the ETSZ may evolve eventually into a through- | going, strike-slip fault running along or near the entire northwest boundary of the Ocoee block in eastern Tennessee. This evolution is suggested by several observations. First, seismic strain B energy release is highest along this boundary, as evidenced by concentration of the largest events near the NY-AL magnetic lineament. Second, the aeromagnetic signature associated with the NY- I AL lineament suggests that this feature is a sharp (vertical?) boundary separating two distinct rock types and, thus, may facilitate strike-slip motion. Third, orientation of the boundary is more north- • south than the orientation of s i ; a shear couple exists along this boundary facilitating right-lateral J strike-slip motion. Fourth, seismicity appears to have concentrated near the boundary recently. « Specification of a mechanical model for a through-going fault may be possible following event | relocation analysis, presently underway. Relocated events may reveal en echelon strike-slip movement in agreement with wrench zone models or, tabular zones striking parallel to the I northwest boundary of the Ocoee block, suggesting general weakening of the crust leading to possible failure. fl Even if a through-going fault does not develop along the northwest boundary of the ETSZ, the zone poses a significant seismic hazard. The NS - EW oriented focal planes determined from focal • mechanism solutions suggest that only segments of the zone may be capable of failure at any time. S However, each segment is large enough to produce a damaging earthquake. The potential danger g of the zone is also indicated by the great vertical extent of the seismic activity; large eastern North P America events are often deep (e.g. the 1968 Dlinois event (25 km) and the 1988 Saguenay event (29 km)). The lack of a large event in historical time within the ETSZ is often used to downplay I the potential hazard associated with the zone. However, the Gutenberg-Richter recurrence relationship does not truncate at low magnitudes for any seismic zone of comparable size anywhere 1 in the world. A simple calculation assuming Poisson behavior and catalog statistics for eastern Tennessee derived from the historical and instrumental record (Bollinger et al., 1989) indicates that |j the best estimate probability of an mb 6.5 or larger event in eastern Tennessee during an exposure * period of 300 years is only about 0.2. Thus, the absence of a large event in the historical record is • to be expected and cannot be used to rule out the possibility of a major shock in the near future. | Properly assessing the seismic hazard is critical. Contained within the ETSZ are numerous nuclear power reactors and hydroelectric projects, Oak Ridge National Laboratory, and population centers J including Knoxville and Chattanooga. The close association of the ETSZ with potential field anomalies and the lack of a large I earthquake in the historical record are features held in common with recently proposed linear I I -D57- seismic zones in southern Ontario. A major difference is in the rate of seismic activity; the ETSZ is characterized by a much higher rate of seismicity than the zones proposed in southern Ontario.

References

Adams, J. and Basham, P., 1991, The seismicity and seismotectonics of eastern Canada, in Neotectonics of North America, D.B. Slemmons, E.R. Engdahl, M.D. Zoback, D.D. Blackwell, Eds., Decade Map Volume 1 (Geological Society of America, Boulder, CO.), p. 261-276. Bollinger, G.A., Langer, C.J., and Harding, ST., 1976, The eastern Tennessee earthquake sequence of October through December, 1073, Bull. Seism. Soc. Am., v. 66, p. 525-547. Bollinger, G.A., and Wheeler, R.L., 1983, The Giles County, Virginia seismic zone, Science, v. 219, p. 1063-1065. Bollinger, G.A., Davison, F.C., and Sibol, M.S, 1989, Magnitude recurrence relations for the southeastern United States and its subdivisions, J. Geophys. Res., v. 94, p. 2857-2873. Bollinger, G.A., Johnston, A.C., Talwani, P., Long, L.T., Shedlock, K.M., Sibol, M.S., and Chapman, M.C., 1991, Seismicity of the southeastern United States; 1698 to 1986, in Neotectonics of North America, D.B. Slemmons, E.R. Engdahl, M.D. Zoback, D.D. Blackwell, Eds., Decade Map Volume 1 (Geological Society of America, Boulder, CO.), p. 291-308. Cook, F.A., Brown, L.D., Kaufman, S., Oliver, J.E., 1983, COCORP seismic reflection profiling, Amer. Assoc. Pet. Geol. Stud. Geol. Ser. 14, 1. Johnston, A.C., Reinbold, D.J., Brewer, S.I., 1985, Seismotectonics of the southern Appalachians, Bull. Seism. Soc. Amer., v. 75, p. 291-312. Keller, G.R., Bland, A.E., Greenberg, J.K., 1982, Evidence for a major late Precambrian tectonic event (rifting?) in the eastern mid-continent region, United States, Tectonics, v. 1, p. 213-223. King, E.R., and Zietz, I., 1978, The New York-Alabama lineament: geophysical evidence for a major crustal break in the basement beneath the Appalachian basin, Geology, v. 6, p. 312-318. 24, p. 147-157. Hatcher, R.D., Zietz, I., Litehiser, J.J., 1987, Crustal subdivisions of the eastern and central United States and a seismic boundary hypothesis for eastern seismicity, Geology, v. 15, p. 528-532. Mount, V.S., and Suppe, J., 1987, State of stress near the San Andreas fault: implications for wrench tectonics, Geology, v. 15, p. 1143-1146. Mitchell, B.J., Nuttli, O.W., Herrmann, R.B., 1991, Seismotectonics of the central United States, in Neotectonics of North America, D.B. Slemmons, E.R. Engdahl, M.D. Zoback, D.D. -D58- *

Blackwell, Eds., Decade Map Volume 1 (Geological Society of America, Boulder, CO.), p. | 245-260. Nelson, A.E., and Zietz, I., 1983, The Clingman lineament, other aeromagnetic features, and I major lithotectonic units in part of the southern Appalachian mountains, Southeastern Geol., v. Owens, T.J., Zandt, G., Taylor, S.R., 1984, Seismic evidence for an ancient rift beneath the I Cumberland Plateau, Tennessee, J. Geophys. Res., v. 89, p. 7783-7796. ' Powell, C.A., Bollinger, G.A., Chapman, M.C., Sibol, M.S., Johnston, A.C., Wheeler, R.L., | 1994, A seismotectonic model for the 300-kilometer long eastern Tennessee seismic zone, • Science, v. 264, p. 686-688. - Teague, A.G., Bollinger, G.A., Johnston, A.C., 1986, Focal mechanism analysis for eastern | Tennessee earthquakes (1981-1983), Bull. Seism. Soc. Am., v. 76, p. 95-109. Vlahovic, G., Powell, C.A., Sibol, M.S., and Chapman, M.C., 1995, The crustal velocity J structure for the eastern Tennessee seismic zone, Seism. Res. Lett., v. 66, p. 48. Zoback, M.L., 1992, First and second order patterns of stress in the lithosphère: the world stress 1 map project, J. Geophys. Res., v. 97, p. 11703-11728. I I I 1 I \

I I 1 I I I I I I

I Session E I I

| Methods and Data for Characterizing the Seismicity Parameters • of Seismic Sources I I I I I I I I I Workshop on the Assessment of Seismic Hazard in Southern Ontario 19-21 June 1995. Ottawa I

Stable continental earthquakes and seismic hazard " in Eastern North America 1_ Arch C. Johnston • Center for Earthquake Research & Information (CERI) University of Memphis, Memphis, TN USA 38152 •

Canada and the United States east of the Rocky Mountains make up Eastern North America, fl one of the nine stable continental regions (SCR) of the world. Its crust is old and for the most part tectonically stable. A recent study for the Electric Power Research Institute has shown • that the seismic potential of SCR crust is not uniform but varies according to the degree of | rifting or crustal extension that it underwent in the geologic past. This paper identifies three types of Eastern North America crust—un rifted, failed intracontinental rift complexes, and • Mesozoic rifted passive margins—and provides an accounting of the known largest | earthquakes and seismic activity of each. Such an approach allows seismic hazard and risk estimates for Eastern North America to be refined. _

An understanding of the seismic hazard and risk of midplate regions such as eastern North America is aided by comparison with better-known seismic regions. Japan, for example, divides its earthquakes, hence its seismic hazard, into two categories. The first and I best known originates from the major and great earthquakes of the highly active subduction • zone plate boundaries lying offshore. The second, less studied and less understood, is what they call their inland earthquakes, which arise from active faulting in the active intraplate I region that is onshore Japan. The recent devastating Kobe earthquake was of this latter type. •

An analogous dual classification also serves well for eastern Canada or, more broadly, • eastern North America (ENA). The analog to the offshore plate-margin events of Japan are I the offshore passive margin events of ENA. And the inland earthquakes of Japan have an analog in stable continental regions (SCR) such as ENA: the cratonic or Appalachian fold • belt earthquakes that reactivate a tiny proportion of the abundant old faults of these regions. I Thus for seismic hazard analysis inland SCR earthquakes may be divided into those occurring in crust that has not suffered rifting (extension) since the Precambrian (-570 mya) • and those occurring in Paleozoic or Mesozoic rifted crust. A generalized mapping of the | three types of continental crust is shown in Figure 1 for ENA.

THE ENA OFFSHORE PASSIVE MARGIN " The most obvious characteristic of the world's continent?! passive margins is that their n outboard margin marks the transition boundary between conth' ntal and oceanic crust. This | boundary is a first-order feature of the Earth's surface layers. True oceanic crust is basaltic, no more than 10-15 km thick, and averages only -7 km. Typical continental crust is granitic, n of -40 km average thickness with extremes of -70 km and -20 km. Typically the continental- I oceanic boundary (COB) is abrupt, occurring over a lateral dimension of 10-20 km, but it can be transitional, extending over 100 km or more. The crust of the continental shelf out to the _

I -E2-

COB is typically thinned and faulted due to the rifting process and often intruded with magma. These effects may extend for hundreds of kilometers inboard of the COB.

The analogy between the offshore subduction zones of Japan and ENA's passive margin COB cannot be taken too far. The fundamental difference is the sustained relative motion between tectonic plates that takes place at the offshore Japan zones guarantees sustained generation of large earthquakes there. The COBs do not have this component of relative movement between plates; rather they are the sites of profound change in crustal properties within a single plate (in this case the North American plate). This change cannot guarantee large earthquakes in the sense of the Japanese offshore subduction zones. It does, however, seem to concentrate midplate earthquake activity for reasons that are not completely understood. Stein et al. (1989) provide a comprehensive analysis of stress concentration mechanisms at passive margins that may be important in explaining COB earthquakes. Alternatively, earthquakes may concentrate there because of the abundant faults—originally formed in the rifting process that opened the ocean basin—available for reactivation. An important question for the ENA COB is whether déglaciation stresses make formerly glaciated passive margins more susceptible to large earthquakes than never- glaciated margins.

Based on fairly sparse focal mechanism data (Johnston et al. 1993), most of the SCR passive margins of the world appear to be under compression with P-axes roughly aligned with absolute plate motion vectors. This suggests that locally generated stresses such as from déglaciation, sediment loading or crustal density contrasts are secondary to the tectonic stress regime imposed by global plate interactions. However, these secondary stresses are localized at or near the COB and produce there a concentration and/or perturbation of the regional stress field not found in the continental interior.

Worldwide, the COB and its inboard passive margins are the sites of some of the largest SCR earthquakes. As tabulated by Johnston et al. (1994), nine of the 15 known SCR earthquakes of moment magnitude M > 7.0 occurred in passive margin crust: four of the nine were COB events. The largest of these was 1933 Baffin Bay (Figure 1) for which an early seismic moment determination (Stein et al. 1979) yields M7.7, but M7.3 estimated from its surface-wave magnitude (Ms 7.3) is probably more accurate. Bent (1995a) has recently determined that the 1929 Grand Banks earthquake was a complex strike-slip rupture of M7.2. The two largest inboard passive margin earthquakes in the world were both historical: the 1604 event in the Taiwan straits, estimated M7.6 (Johnston et al. 1994), and the 1886 Charleston event (Figure 1), estimated M7.4 (Johnston 1995).

The seismic activity rate of the COB cannot be estimated with any confidence. Seismographic coverage sufficient to monitor the COB for small and moderate events has been inadequate for most of this century. I will, however, provide estimates for the COB and Mesozoic margins in combination with the interior rift complexes in a later section.

The passive margins discussed in this section are the current ones for ENA, produced by the Mesozoic (-200 mya) opening of the present-day Atlantic Ocean. Wheeler (1995) recognizes a 'fossil' passive margin limit (Figure 1) produced by the rift-opening of the lapetan Ocean at the beginning of the Paleozoic Era (-550-600 mya). It was the closure of the lapetan Ocean that produced the ENA Appalachian fold belt during the Paleozoic; the crust between the lapetan and Mesozoic margin limit consists mainly of Appalachian fold belt and Grenville province (1.0-1.2 by) rocks. Its largest earthquake is the 1897 M5.8 Giles I -E3-

County, Virginia event. Other notable events include a pair of mid-M5 shocks in New Hampshire in 1940 and the M5.5 New Brunswick earthquake in 1982. Major M>7 - earthquakes, such as Grand Banks, Baffin Bay, New Madrid, or Charleston, are not observed I in this type of crust. In this respect the lapetan margin crust does not differ from unrifted SCR • crust. CRATONS AND FOLD BELTS: UNRIFTED SCR CRUST I In September 1993 in the Killari-Latur region of peninsula India an M6.1 earthquake killed approximately 10,000 people. In magnitude, shallow focus, thrust faulting mechanism, I surface rupture and geologic setting in unrifted Precambrian craton, it was a virtual twin of • ENA's largest known earthquake in unrifted crust, the 1989 M6.0 Ungava earthquake in northern Quebec, which killed no one. This stark contrast highlights an aspect of ENA I seismic risk that both comforting and troubling. Since European settlement Canada and the • U.S. combined have experienced at least 15 SCR earthquakes of M > 6.0. Yet the deadliest, 1886 Charleston, killed only -60 people. In this century not a single life has been directly I taken by an ENA earthquake (the tsunami from the 1929 Grand Banks event did kill -30 I people). Is the ENA seismic risk truly insignificant as this century's record suggests or have we just been lucky? Presently there is no scientific means to gauge whether significant ENA I earthquakes will continue to miss urban areas. Our only useful approach to assess ENA I seismic hazard is to quantify ENA seismic activity rates.

Earthquakes such as Ungava and Killari-Latur—with magnitudes M > 6 and foci in | ancient Precambrian cratons—are perhaps the rarest and least explainable type of earthquake that we know of. About the best we can say is that such events must be • reactivations of ancient faults or shear zones stressed to failure by tectonic stresses \ transmitted from the plate boundaries to the interiors. It is possible but not demonstrated that stress localization or concentration mechanisms are a factor, as they almost certainly are at • the COB. Continental cratons worldwide are densely packed with these ancient shear zones; I at present it is impossible to forecast which may be the more likely sites for future Ungava- type earthquakes. Therefore the best way to characterize the seismic potential of unrifted . SCR crust is with a seismic activity rate normalized to a given unit of crustal area. This is I done in the Conclusions section in comparison with rifted crust.

RIFTED SCR CRUST I Only two intracontinental (or intracratonic) rifted regions are shown in Figure 1 : the St. Lawrence rift complex, which includes the Saguenay and Ottawa grabens, and the Reelfoot rift complex, which is drawn to include the Rough Creek and Wabash Valley grabens, I although it is controversial whether they do indeed link up with Reelfoot. Other minor ENA ' rifts or grabens that probably are not through-going crustal features are not shown, nor are Precambrian rifts such as the huge Midcontinent rift system of Grenville age. Because of their I age these latter features are considered to be incorporated into the craton; worldwide, in all ' SCRs, Precambrian rifts, where identified, are aseismic to background seismicity levels.

Both rift systems have produced major earthquakes. The St. Lawrence rift includes the I Charlevoix zone, second in ENA only to the Reelfoot rift's New Madrid seismic zone's record of large earthquakes. The principal M>6 Charlevoix earthquakes include: 1663, M-6.6; I 1860, M~6; 1870, M-6.5; and 1925, M=6.2 (Bent 1992). The intensity data of the historical I events is poor so their magnitudes have an uncertainty of about ±0.5 M units. Elsewhere in the St. Lawrence system, large earthquakes occurred in 1732, M-6.2, near Montreal; 1935, I I -K4-

M=6.1 (Bent 19955) near Timiskaming at the western extreme of the Ottawa graben; and 1988, M=5.8, in the Saguenay graber». The 1989 Ungava earthquake mentioned above is the only known M>6 earthquake in unrifted ENA crust.

The New Madrid seismic zone (NMSZ) of the Reelfoot rift system is infamous for its protracted sequence of very large earthquakes during the winter of 1811-1812. At least five of these events were felt throughout the eastern United States, and three of them were felt into Canada, one as far as Québec City. These are the largest known SCR earthquakes; magnitudes for the three largest have been recently estimated as M7.8-8.1 (Johnston 1995) with uncertainties of ±0.4-0.5 M units. The only other central U.S. M>6 earthquakes were also NMSZ events: 1843, M6.5, in northeastern Arkansas and 1895, M6.8, near the Missouri- Illinois border. All these earthquakes were in the nineteenth century; in contrast, the largest NMSZ earthquake of the twentieth century has been only in the Iow-M5 range. Because of its 1925 and 1935 M>6 earthquakes, the St. Lawrence rift has generated a greater seismic moment release in this century than the NMSZ.

The St. Lawrence and Reelfoot rifts are failed rifts or aulacogens that never reached the stage of producing oceanic crust. The Mesozoic passive margins represent one side of the successful Atlantic rift that separated North America from Europe and Africa. Thus both represent continental crust that has been 'damaged' by the rifting process. Johnston et al. (1994) have assessed the seismic activity rate for this combined category of rifted crust. This is conveniently expressed in terms of total rate of production expressed as an average time interval between occurrences of independent M>5 and M>6 events and normalized to per 105 km2 of crust. Rifted crust ENA activity is compared to that of unrifted crust in the Conclusions section. CONCLUSIONS The crust of Eastern North America was divided into the basic categories of unrifted crust and crust that experienced post-Precambrian rifting. The later category may be further subdivided into imbedded intracontinental rift systems and Mesozoic passive margins, including the COB. In terms of seismic activity the distinction between rifted and unrifted crust is significant as shown in this concluding table extracted from Johnston et al. (1994):

Independent SCR and ENA Earthquake Activity Rates -Total per105km2- Cateqory Area vrs. perM>5.0 vrs. per M>6.0 vrs. per M>5.0 vrs. per M>6.0

SCR total 132x106km2 0.20 1.9 246 2450 SCR rifted 37x106 0.33 2.5 119 916 SCR unrifted 96x106 0.49 7.6 468 7178 ENA total 24x106 1.4 7.5 331 1770 ENA rifted 8x106 3.1 17.5 254 1420 ENA unrifted 16x106 2.7 16.4 426 2630

When uncertainties are accounted for, ENA's normalized production of M>5 and M>6 earthquakes is not significantly different from the global SCR average. However, the normalized rates for ENA rifted crust are nearly twice those of ENA unrifted crust. The unnormalized rates are roughly equal because there is twice as much unrifted as rifted ENA I -E5- crust. For total SCRs, normalized rates for rifted crust exceed those for unrifted crust by a factor of 4 at M>5, increasing to a factor of 8 at M>6. If the St. Lawrence and Reelfoot failed • rifts were computed separately, their normalized rates would far exceed all other ENA crust | (except perhaps the COB whose area is unknown) because of their relatively small size and relatively high rate of M>6 earthquakes. The highest priority to further refine these activity • rates for hazard assessment is to understand the mechanism(s) that localize seismic activity I along the ENA passive margins and within the imbedded rift systems. I ACKNOWLEDGMENTS • The elements of this paper have their basis in research supported by the Electric Power - Research Institute, as reported in EPRI TR-102261. I thank Peter Basham of the GSC and I Carl Stepp for continued interest and input and Allison Bent for preprints of several important papers. _

REFERENCES - Bent, A.L. 1992. A Re-examination of the 1925 Charlevoix, Québec, Earthquake. Bull. Seism. • Soc.Am., 82, 2097-2113.

Bent, A.L. 1995a. A Complex Double Couple Source Mechanism for the Ms 7.2 1929 Grand I Banks Earthquake. Bull. Seism. Soc. Am., submitted.

Bent, A.L. 1995/b. An Improved Source Mechanism for the 1935 Timiskaming, Quebec | Earthquake from Regional Waveforms. Pure Appi Geophys., submitted.

Johnston, A.C. 1995. Seismic Moment Assessment of Stable Continental Earthquakes, Part | 3: 1811-1812 New Madrid, 1886 Charleston and 1755 Lisbon. Geophys. J. Int., submitted. •

Johnston, A.C., Coppersmith, K.J., Kanter, L.R., Cornell, C.A. 1994. The Earthquakes of Stable Continental Regions: Assessment of Large Earthquake Potential. EPRI Report TR- . 102261, Schneider, J.F. éd., Electric Power Research Inst., Palo Alto. I

Stein, S. Sleep, N.H., Geller, R.J., Wang, S.C., Kroeger, G.C. 1979. Earthquakes along the Passive Margin of Eastern Canada. Geophys. Res. Lett., 6, 537-540. I

Stein, S., Cloetingh, S., Sleep, N.H., Wortel, R. 1989. Passive Margin Earthquakes, Stresses and Rheology. in Earthquakes at North-Atlantic Passive Margins: Neotectonics and I Postglacial Rebound, Gregersen, S. and Basham, P.W, eds., Kluwer, Dordrecht, 231-259. I

Wheeler, R.L. 1995. Earthquakes and the Cratonward Limit of lapetan Faulting in Eastern I North America. Geology ,.23,105-108. I I I I -E6-

70°N 60° N

20°W

30° N

90°W 80° W 70°W 60° W

Figure 1. The three types of seismogenic continental crust of ENA. Diagonal pattern denotes the Atlantic passive margin with crust thinned and faulted to the inboard limit by the Mesozoic rift opening of the Atlantic. The Paleozoic Reelfoot (USA) and St. Lawrence (Canada) intracontinental rift complexes are shown by dark shading and tic-mark borders. Continental crust unrifted since the Precambrian is lightly shaded. Dual-dash line is the inboard limit of lapetan (proto-Atlantic) rifting of Wheeler (1995). Major (M > 5.8) ENA earthquakes are shown with dates; larger symbols denote M > 7.0. -E7- I

AECB Workshop on Seismic Hazard Assessment in Southern Ontario I Seismicitv Parameters for Eastern Canadian Seismic Source Zones I P.W. Basham | Geological Survey of Canada I ABSTRACT |

Seismic hazard estimates are critically dependent on the seismicity parameters established for each of the earthquake source zones, essentially the rates at which • earthquakes are expected to occur as a function of their magnitudes, and the upper- | bound magnitudes that may be assigned the zones. Earthquake rates are established on the basis of historical seismicity (with rare cases of some information being available • from paleo-seismic or prehistoric events), with due consideration to the completness of | earthquake reporting from written accounts, and seismograph records throughout history. In a highly active earthquake zone, such as the Charlevloix zone in the St. Lawrence m Valley the rates of both historical felt earthquakes and lower-magnitude earthquakes | detected by sensitive seismograph networks are well established, and lead to well-defined magnitude recurrence relations with small error bounds for the source zone. In zones of g much lower activity, such as the Niagara-Attica zone, the rates of historical events and | lower-magnitude recent events are very poorly defined and lead to magnitude recurrence relations with large uncertainty. For example, in the Niagara-Attica zone there has been • only one earthquake larger than magnitude 5, and it might have a return period much | longer than the historical record. The Johnston, Coppersmith, et al. study for the U.S. Electric Power Research | Institute of the seismicity of stable continental regions was established primarily to determine the maximum earthquakes that might occur in the eastern United States. The g extensive data base that they have established (see my source-zone abstract) has been | invaluable in establishing very firm grounds for upper-bound magnitudes in both the continental rift zones and the stable continental core. These have strongly influenced the • adoption of upper-bound magnitudes for the eastern Canadian earthquake source zones. | Upper-bound magnitudes as large as 7.7 are assigned to source zones representing passive margins that broke the continental crust; magnitude 6.7 for the stable continental • core; and as large as magnitude 7.5 for source zones that are transitional between these | two geological environments. The value of 6.7 for the stable continental core may come as a surprise to the uninitiated, but events of this size have occurred recently on the • stable core of the Australian Shield; the Ungava earthquake of 1989 (M6.2) is considered | a slightly smaller, rare event of this type in the Canadian stable continental core.

I I I I I I I I Session F I

Methods to Assess Vibratory Ground Motion Hazard and Applications I I I I I I I I I I I I Uses of Probabilistic Seismic Hazard Analysis for Evaluating Nuclear Plant Core-Damage Frequency 0

Robin K. McGuire Risk Engineering, Inc. • 4155 Darley Ave, Suite A I Boulder, CO 80303, USA I Abstract

The purpose of a probabilistic seismic hazard analysis (PSHA) is to make decisions regarding ' the seismic design, retrofit, or capability of engineered facilities. A PSHA will be useful only to the extent that it characterizes and makes use of current knowledge about tectonics, • geophysics, seismology, geotechnical engineering, and strong ground motion estimation. As part ™ of this knowledge, the PSHA must capture and represent the range of interpretations in the relevant sciences and the range of models and parameters used to represent physical processes. I Doing so will allow the uncertainties in earthquake effects to be quantified, from which decisions on design, retrofit, or capability of engineered facilities can be made. _

In this context uncertainties both in seismic hazard and in structural/mechanical response of the facility need to be quantified. Each can and does contribute to uncertainty in the effects of _ earthquakes on the facility, and ignoring uncertainties will lead to non-optimal decisions. I

A common criterion used to judge acceptable seismic risk for nuclear power plants is the mean « core damage frequency. This is the expected frequency with which plant core damage will occur, I expected in the sense of a mean value over epistemic uncertainties in seismic hazard and in plant fragility. More integrative measures could be used such as the distribution of loss for a wide m range of plant damage states, or the potential short- and long-term health effects on the affected | population, but these would require progressively more studies and incorporation of additional uncertainties in models and parameters. The advantages of using the criterion of mean core • damage frequency are that it constitutes a true (if simple) seismic risk analysis (SRA), and the fl procedure for calculating it is well established.

To conduct a seismic risk analysis, we start with a PSHA that represents both aleatory and I epistemic uncertainty. The fundamental calculations integrate over earthquake magnitude and location to evaluate the frequency or probability per unit time that certain ground motion • amplitudes are exceeded by any earthquake. This hazard calculation represents the aleatory I uncertainty and requires specification of earthquake sources, a magnitude distribution for each, and an attenuation equation for the relevant ground motion parameter. In addition there are I epistemic uncertainties recognized in the models and parameters used to calculate the hazard, and • these epistemic uncertainties are captured by using alternative models and ranges of input parameters. As a result of the epistemic uncertainties in input models and parameters, the I analysis represents epistemic uncertainties in the hazard with multiple hazard curves or fractiles • I RKM/REPORT/AECL.695 I -F2-

of hazard (fractiles of annual probabilities or frequencies). A typical PSHA representation of epistemic uncertainties is shown in Figure 1.

To evaluate the mean core damage frequency we also need a plant-level fragility curve showing both the aleatory and epistemic uncertainty in core damage for a range of ground motion levels. Figure 2 illustrates a generic plant-level fragility curve, taken from McGuire et al. (1989). This curve shows the fraction of nuclear plants that will experience core damage as a function of peak ground acceleration (PGA). The "median-median" fragility A is 0.61 g in this example, meaning that for the median fragility (the central curve horizontally) there is a 50% damage fraction at 0.61g. This seismic fragility curve was developed by determining fragility curves for each critical component in the plant, putting together the logic describing combinations of component failures that result in core damage, and (using Boolean expressions that account for system redundancies) determining the plant-level fragility as a summary of all the combinations of component failures that could lead to core damage. As Figure 2 shows, the analysis represents both aleatory uncertainty through PR and epistemic uncertainty Pu, which controls the horizontal range of the curves. Thus the fragility representation is consistent with a PSHA. Although the example of Figure 2 is in terms of PGA, the most recent fragility curves have been developed as a function of response spectrum acceleration, often averaged over a range of frequencies.

Once a PSHA and seismic fragility analysis have been conducted for a facility, an SRA for mean core damage frequency is straightforward. We discretized both the seismic hazard and seismic fragility representations into a finite set of hazard and fragility curves. Then for each combination of hazard curve X-, and fragility curve If we perform a numerical integration over ground motion amplitude to estimate a seismic core-damage frequency (SCDFy) for a particular combination ij as follows: MNT fa SCDF(j = £ {Xt[ak] - \[ak + LatfFfa + _*] (1) K= 1 2

where

Aak = ak+i - ak (2)

and NINT is the number of intervals used to discretize the range of ground motion amplitudes.

By performing the integration of hazard and fragility for all possible combinations of curves from the seismic hazard curve and fragility curve families, a distribution of episteir.ic uncertainty in seismic core-damage frequency can be obtained. Designating the weights of hazard curve i and j fragility curve j with wh' and wf respectively, the weight of any combination is calculated as: <4 = <*Wf (3)

RKM/REPORT/AECL.695 -F3-

The mean seismic core-damage frequency can then be obtained as:

NHAZ NFRAG SCDF = Ê £ of SCDF, (4)

where NHAZ and NFRAG are the number of hazard curves and fragility curves, respectively, in each family. The entire distribution of seismic core-damage frequency is determined from:

P[SCDF < x] = Y; £ *4 for W UJ such that SCDF < x) (S)

Figure 3 shows a typical plot of a seismic core-damage distribution function, including its median and mean value. This is a basic SRA result; it shows the epistemic uncertainty (the ordinate) for . core damage frequency (the abscissa). I

If we assume that the hazard is linear on a log-log plot, and that plant-level fragility curves can i be represented by the double-lognormal distribution typically used for component fragilities, then | an analytical expression can be derived between the mean and variance of SCDF and characteristics of the hazard and plant-level fragilities: . • -•BrKA'M'to («) _ I 2 [P + )1] OICDF = SCDF {e " ™" - 1} (7) ,

where ÎT(Â) denotes the mean of the hazard curve at À, Kn is the negative of the slope of the I hazard curve in log-log space, PH measures the uncertainty of the hazard curve (it equals i/[ln(COVjj + l)] of the hazard curve at Â, where COV is the coefficient of variation), and py and j pR are the epistemic and aleatory uncertainty measures for plant-level fragilities, as before. I Under the same assumptions we can also relate the HCLPF to À as follows:

HCLPF = ^'^TW (8) |

which allows the SCDF to be determined as a function of hazard at the HCLPF: I

SCDF = H[HCLPF]e 2 *?> j

RKM/REPORTMECL.695 -F4-

These equations can be used to determine design ground motions for new plants or review level ground motions for existing plants, at several levels of sophistication and detail:

Most detailed: One can develop a plant level fragility curve for an existing plant based on its original design and construction, or for a proposed plant based on its proposed design. A site-specific PSHA will also quantify the seismic hazard for the plant site. One can then perform the analyses summarized in Equations (4) and (5) to calculate the SCDF distribution and its mean, and compare these results to a quantitative criterion. If SCDF exceeds the criterion, the existing plant can be seismically retrofit or the new plant can be redesigned to increase the plant level fragility curve so that the core damage criterion is met.

Less detailed, One can conduct a study of seismic margins to determine HCLPF values for critical plant structures, components, and equipment, identify critical core-damage sequences, and derive a plant-level HCLPF from these. A site-specific PSHA will quantify the seismic hazard for the plant site, and Equation (10) can be used to estimate SCDF. For this application, generic values of PR and ftj are appropriate. If SCDF exceeds the criterion for acceptable core-damage frequency, one can determine what HCLPF is necessary so SCDF is below the criterion, and can retrofit or redesign the facility to that new HCLPF.

Generic. Without any plant-specific information, one can conduct a site-specific PSHA for a plant site, and adopt distributions for plant fragility based on studies conducted at similar nuclear plants. In this application it is appropriate to incorporate an additional uncertainty Pu' to reflect uncertainty in À for the specific plant being considered. It would also be appropriate and conservative to assume that the À value for the subject plant is less than the average A value for similar plants.

All of these methods will give estimates of the seismic core-damage frequency and allow estimated frequencies to be compared to a core-damage criterion. The less-detailed and generic procedures use a generic plant-level fragility curve, which is justified because values of PR and Pu appear to be stable from plant to plant (values of 0.22 and 024 are typical). It may be appropriate to adopt a more stringent acceptance criterion than if a plant-specific fragility analysis were available, or to assume a lower A value in the generic study than is the mean of similar plants studied in detail.

The above equations allow plant core-melt frequencies to be estimated for a range of information available about the plant. One should also understand the influence of uncertainties on SCDF. As mentioned above, PR and Pu are observed to be quite stable among detailed Probabilistic Risk Assessments performed for nuclear plants in the US. The flat nature of seismic hazard curves in ENA means further that epistemic uncertainty in hazard will dominate uncertainty in seismic core damage, i.e. reducing Pu will not substantially reduce uncertainty in core damage frequency. Note however that pH does not appear in equations (6) and (9); this means only that, once the mean hazard curve is known, derivation of the mean core damage frequency does not depend on uncertainty in the hazard. This follows from the linear nature of the mathematical expressions

RKM/REPORT/AECL.695 4 -F5-

involved. To obtain the proper mean hazard curve it is necessary to consider the range of hypotheses involved; best estimate hypotheses alone will not lead to the correct mean hazard I curve nor the correct mean core damage frequency.

In summary, the derivation of SCDF for nuclear plants is straightforward, but requires a site- I specific seismic hazard curve. Decisions on the level of resources appropriate to devote to study the plant margins and sequences must be made, but estimates of SCDF can be obtained for a range of studies, from detailed, plant-specific analyses to generic reviews of similar studies at I other plants. This allows a sequential decision process to be used, starting from more generic, less costly studies, and proceeding to more detailed, expensive studies only if warranted by results from initial studies that indicate core damage frequencies near or exceeding accepted j guidelines.

References I

Sewell, R.T., R.K. McGuire and G.R. Toro (1993). Use of probabilistic seismic hazard results: j general decision-making, the Char.^ston earthquake, and severe-accident evaluations, Risk Engineering, Inc. Rept. to EPRi, Projects 2356-50 and 2356-52, July. •

McGuire, R.K., G.R. Toro, J.P. Jacobson, T.F. O'Hara and W.J. Silva (1989). Probabilistic seismic hazard evaluations at nuclear plant sites in the central and eastern United States: I resolution of the Charleston earthquake issue, EPRI Special Rept. NP-6395-D, April. \ I 1 I I I I I I RKMTUIPORT/AECL.695 I -F6-

-î 10 1 1 1 1 1 r 1 1 : 05lh fraclile • median 15Lh fraclile- •— - mean

0. 200. 400. 600. 800. 1000. ACCELERATION (cm/sec2 )

Figure 1. Typical representation of epistemic uncertainty in seismic hazard using fractile hazard curves.

RKM/RErORT/AIiCL.695 -F7-

CORE-MELT FRAGILITY CURVES AND PARAMETERS

1.0 1 ! 1 1 ]

0.9 - F7f^ 1 / JVariabilily in 9 O.'i E- U 2 0.7

O.G -

0.5 (in 0.4 Y : t A JUncertainLy in O / ^Median - fiu \ 0.3 r Q 0.2 "- O 0.1 A = Median-Median Capacity : 0.0 tffliifilifii 0.0 HCLPF 0.5 1.0 1.5 2.0

PGA (G-UNITS) Î I

Figure 2. Illustration of core-melt fragility curve family showing parameters À, PR, and Pu (from McGuire et al., 1989). I I RKM/REI'ORT/AECL.695 I -F8-

SEISMIC CORE DAMAGE FREQUENCY DISTRIBUTION

1.0 "I I l~l~mTl 1 1 1—t-lTTT

0.9 o.n

0.7 cO O 0.6 P-. 0.5 CD t> 0.4 iat r 8 0.3 u 0.2

0.1

0.0

Plant-Damage-Stale (Core-Melt) Frequency

Figure 3. Illustration of a seismic core-damage frequency distribution and its median and mean value (from Sewell et al., 1993).

RKM/REl'ORT/AHC1..695 -F9- "

Sources of Uncertainty and its Treatment in Seismic Hazard j Analysis in Eastern Canada I John Adams National Earthquake Hazards Program, I Geological Survey of Canada, • 1 Observatory Cres, OTTAWA, K1A 0Y3 Canada

There have been three previous generations of seismic hazard maps: 1953, 1970, and 1985. It is clear that it is time for new set of maps, partly because of our better earthquake catalog, new I understanding of seismotectonics, and new strong ground motion relations, but also because for the first time improved hazard computation codes can quantify the uncertainty in our estimates. . ] For the fourth generation maps we have retained the "Cornell-McGuire" methodology but use proprietary code "FRISK88", which has been designed to incorporate uncertainty and provide the m percentiles of the hazard distribution. For each input we establish best estimates for each | parameter together with upper and lower estimates to quantify the uncertainty. The chief parameters are: « 1. Seismicity parameters These are: a, the seismicity rate; P, the relative rate of big and small earthquakes; earthquake depth; and the Upper Bound Magnitude (UBM), the 1 largest earthquake thought plausible. I

2. Eastern Canada Strong Ground Motion (SGM) parameters. We use the Atkinson I and Boore (1995) "hard rock" equations, together with the aleatory and epistemic * uncertainty suggested by Atkinson (1995). Reference ground condition factors are used convert "hard rock" to "firm ground" for Building Code purposes. I

3. Seismic source zones. Two sets of source models are used to capture different _ opinions about the seismotectonic causes of the earthquakes and also to capture a range 1 .of geographic scales. For Building Code purposes we suggest computing the probabilistic hazard for each source model using input parameters and their uncertainty for the same m grid of points; for our "robust" maps we then chose the larger of the two values. |

All uncertainties are fed into the analysis, but the median is dominated by best estimates for each • parameter. For a computed hazard value at a given probability level, the chances of the worst | case (i.e. highest SGM relation, upper UBM, flattest beta, highest alpha, shallowest depth) are extremely low. In terms of their relative importance to the final hazard the uncertainties can • generally be ranked thus: (High importance) SGM relations, beta, seismotectonic model, alpha, I UBM, depth (Low), though the exact order is site dependent.

The intended method estimates the uncertainty about a given hazard value. For example, at the • 0.0021 p.a. probability level we can use the 50th percentile (median), or use a percentile higher

I -F10- than the 50th to incorporate appropriate conservatism; greater where the uncertainty is larger (the 84th percentile is often chosen).

Uncertainty can be expressed by tabulating or mapping percentiles such as the 84th or 95th together with the median for each hazard value (each percentile higher than the 50th expressing an increased confidence that the mapped value will not be exceeded). Alternatively, for a given site and a particular probability, the percentiles of the uniform hazard spectra can be shown.

Prepared on 950530 for AECB Workshop on Seismic Hazard Assessment in southern Ontario, Ottawa 19-21 June, 1995. Cf\ci(o0ù30(o 1 -Fil-

GROUND MOTION RELATIONS IN EASTERN NORTH AMERICA I GAIL M. ATKINSON | For AECB Workshop, June 1995 I ABSTRACT Within the last five years significant effort has been j expended on the analysis of ground motion from earthquakes in eastern North American (ENA) (Ou and Herrmann, 1990; Atkinson and 1 Mereu, 1992; Boatwright and Choy, 1993; Atkinson, 1993a,b; EPRI, 1993; Boatwright, 1994, Atkinson and Somerville, 1994; Atkinson f and Boore, 1995; Toro et al., 1995; Herrmann and Ou, 1995). The effort was partly motivated by the occurrence of the 1988 M 5.8 I Saguenay, Quebec earthquake. The Saguenay earthquake occurred, * perversely enough, just following an earlier and more limited . period of research into ENA ground motion relations (Atkinson, 1 1984; Herrmann, 1985; Boore and Atkinson, 1987; Toro and McGuire, 1987). This earlier research produced ground motion relations j (Boore and Atkinson, 1987; Toro and McGuire, 1987) which underpredicted the high-frequency ground motions experienced I during the Saguenay earthquake by factors of two to five (Boore and Atkinson, 1992). Since the Saguenay earthquake was the j largest event to have occurred in the east in the last 50 years, and is the only well recorded large ENA event, this i underprediction was a source of concern. It highlighted how • little was actually known about ENA source and propagation processes. I There have been two major directions to ENA ground motion research in the 1990s. One of these has been to improve the empirical basis of our knowledge of ENA source and propagation processes, and provide better empirical validation of ENA ground motion relations. These efforts culminated with new empirically- based stochastic ground motion relations for ENA (Atkinson and Boore, 1995). A second approach has been to use improved modeling techniques to gain insight into ENA ground motion -F 12- processes and their variability (Ou and Herrmann, 1990, 1995; EPRI, 1993; Atkinson and Somerville, 1994). This effort culminated with the new engineering ground motion relations of EPRI (Toro et al., 1995). The approach taken by Atkinson and Boore (1995) to improving ground motion relations in light of recent developments was largely empirical. We analyzed data to define each of the seismological input parameters to the stochastic model. It has been shown that if the input parameters can be accurately defined, then the stochastic model will provide accurate ground motion predictions (Atkinson and Somerville, 1994). The input parameters to our 1995 ground motion relations were constructed from the following elements: (i) the analysis of over 1500 ENA seismograms from small-to- moderate events to determine regional attenuation and source characteristics (Atkinson and Mereu, 1992; Atkinson, 1993a, b; Boatwright, 1994), and the duration of motion (Atkinson, 1993b); (ii) analysis of telese.ismic spectra from intra-plate events (Somerville et al., 1987; Boatwright and Choy, 1992); and (iii) estimation of source parameters for large historic ENA events, based on regional seismographic data and calibration of felt areas to spectral parameters (Atkinson, 1993a). The ground motion model follows the stochastic approach first proposed and validated for California by Hanks and McGuire (1981) arid Boore (1983). In this model, the ground motion is treated as finite-duration bandlimited Gaussian noise, whose amplitude spectrum is given by a seismological model of source and propagation processes. The seismological model specifies the earthquake spectrum as a function of moment magnitude (or seismic moment Mo), hypocentral distance (R) and frequency (f), according to:

A(Mo,R,f) = E(Mo,f) D(R,f) P(f) I(f) . (1)

E(Mo,f) is the earthquake source spectrum for a specified seismic moment (ie. Fourier spectrum of the ground acceleration at a distance of 1 km). Using the data described above, the empirical -F13-

source spectral model shown on Figure 1 (solid lines) was derived (Atkinson, 1993a). It differs from the previous Brune (1970) model in that it uses two corner frequencies to describe a spectral shape similar to that indicated by the Saguenay data; significantly, the high-frequency amplitudes are larger than those of the 100-bar Brune model used in our 1987 and 1990 computations, while the low-frequency amplitudes are lower. I D(R,f) is a diminution function that models the geometric and anelastic attenuation of the spectrum as a function of hypocentral distance (R) , as shown on Figure 2. It is also defined empirically, from the regression analyses described by j Atkinson and Mereu (1992); it includes the effect of the 'Moho ' bounce1 (Burger et al., 1987) on the shape of the attenuation i curve. P(f) is a high-cut filter that rapidly reduces amplitudes j at very high frequencies (f»10 Hz) ; we use the fmax model

(Hanks, 1982), with fmax = 50 Hz. I(f) is a filter used to shape | the spectrum to correspond to the particular ground motion measure of interest. For example, for the computation of I response spectra I is the response of an oscillator to ground acceleration. I The final input element of the stochastic predictions is the ' duration of motion. The duration model generally has two terms: •

T = To + b R (2) • where To is the source duration and b R represents a distance- ~ dependent term which accounts for dispersion. For the source | duration, we assume that To = l/(2fA) (Boatwright and Choy,

1992), where fA is the lowest corner frequency in the source I spectrum. The empirical basis for the distance-duration term is the collection of 1500 ECTN seismographs used to define the 1 attenuation function. As shown in Figure 3, the duration increases with distance, steeply at first, then more gradually at J| larger distances. ™ Using the inputs described above, response spectra (5% _ damped pseudo-acceleration, PSA) for frequencies of 0.5 Hz to 20 I Hz, and peak ground acceleration (PGA) and velocity (PGV) were simulated for 4.0 < M < 7.25, in 0.25 magnitude-unit increments, • I -F14- from R = 10 km to R = 500 km, in increments of 0.1 log units. Fifty trials were used for each magnitude-distance combination. The median ground motions for the random horizontal component at hard rock sites are summarized in the Appendix table. Figure 4 plots the estimated PSA at four frequencies, and PGA and PGV, for a range of magnitudes and distances. The figure also shows simple quadratic equations that approximate the estimates for the purposes of seismic hazard calculations. The coefficients of the plotted quadratic prediction equations are listed in Table 1. In comparing hazard estimates made with our new relations to those made with our 1987 and 1990 relations, we have found that high-frequency (f>5 Hz) ground motion estimates have increased significantly. This reflects new knowledge of the potential for high-stress events like the 1988 Saguenay (500 bars) and 1990 Mont Laurier (500 bars) earthquakes. Intermediate-frequency (f^l Hz) motions have decreased, in some cases dramatically, as a consequence of the new source model shape. The relative shift in expected ground motions towards higher frequencies has important implications for seismic hazard evaluations throughout ENA. It may be that the eastern earthquake hazard is mostly restricted to high-frequency structures. The ground motion estimates given in the appendix table, and by the quadratic equations, apply to bedrock sites. For typical ENA deep soil sites (ie. dense or stiff soils more than 60 m in depth), linear analyses indicate that the bedrock values would be amplified by a factor of 1.4 to 2 over most of the frequency range from 0.5 to 10 Hz, as shown in Table 2. Table 2 was produced by comparing the ground motion equations derived by Boore and Joyner (1991) for deep soil sites to the equivalent relations derived by Boore and Atkinson (1987) for rock sites. The amplification factor tends towards unity at high frequencies; the frequency dependence of the amplification is attributable to the depth of the soil column. As a general statement for firm soil sites of unknown depth, the bedrock values should be multiplied by a factor of about two. This does not account for -F15- any decreases in amplification that may be observed at large amplitudes due to nonlinear effects. The extent to which predictive relations can be validated by data is documented in Figures 5 and 6. The data used in these comparisons is listed in Table 3. Figure 5 shows the differences between the observations and the AB95 predictions as a function of M, for several frequencies; each point represents the average log residual for all stations recording an event. There is no systematic dependence on M. Average event residuals are generally within 0.15 log units (40%) of zero, with the exception of the Saguenay (M=5.8) earthquake, which has average residuals of about 0.35 log units (factor of 2.2). Averaged over all eight ' events (equal-event weighting), the mean (log) residuals are , 0.03, 0.04, -0.03 and -0.01 for frequencies of 1, 2, 5 and 10 Hz, ' respectively, with standard deviations of 0.13, 0.14, 0.17 and 0.18. Note these standard deviations represent only the inter- j event component of variability. Figure 6 shows the differences between individual j observations and predictions as a function of distance. The only apparent trend is a region of positive residuals at about 100 km, I which includes most of the Saguenay strong motion records. • Overall, the agreement between data and predictions is a satisfactory. I The standard deviation of the (log) residuals (a), expressing the random variability of ground motions (sometimes referred to | as aleatory uncertainty), is an important parameter for hazard analyses. In western North America, the observed value of a lies I within the range of 0.2 to 0.3 (Boore et al., 1993). In ENA, the random variability depends partly on the magnitude scale used for 1 ground motion predictions (Atkinson, 1995b); it is generally in the range of 0.25 to 0.35. • Epistemic uncertainty in the true level of the median ground • motion relations is also important to seismic hazard analysis. « This uncertainty is distinct from the issue of random scatter 8 about the median (aleatory uncertainty), although there can be significant interplay between these two types of uncertainty. J I -F16-

The AB95 ground motion relations do not explicitly address the issue of epistemic uncertainty in the median. There are several approaches to defining epistemic uncertainty in the median relations. One approach would be to attempt to define the range in professional opinion regarding the true level of the median. This is the approach recently taken by SSHAC (Senior Seismic Hazard Analysis Committee), an organization set up by EPRI, Lawrence Livermore Labs and the Department of Energy (see SSHAC, 1995). Atkinson (1995a) made use of the SSHAC results to estimate upper and lower bounds on the AB95 relations; the AB95 median is near the top of the range of opinion for high- frequency motions, and near the bottom of the range of opinion for low-frequency motions. The disadvantage of this approach is that it involves a subjective interpretation of the meaning of various expert opinions. Another approach is that taken by Toro et al. (1995), namely to model the variability that results from estimated variability in input parameters to the model. This approach is objective, but excludes any uncertainties that are not modeled (source spectral shape, nature of duration, etc.). Further work in defining the uncertainty in median relations appears warranted. As a broadbrush statement, the epistemic uncertainty in the true median is probably about a factor of two (for one standard deviation from the median). This subjective estimate takes into account the following factors: (i) the Toro et al. (1995) characterization of epistemic ' uncertainty suggests a factor of 2.3 for 90% confidence limits on their median; (ii) the SSHAC studies indicate a range of professional opinion corresponding to about a factor of 2 on the median; (iii) the AB95 and Toro et al. (1995) relations differ by factors that exceed 2 in some magnitude-distance ranges, for frequencies less than 3 Hz. In summary, the new ground motion relations developed for ENA over the last five years are a significant improvement on previous models. They are founded on a much larger database, including the data from the 1988 Saguenay, Quebec earthquake. -F17-

These new data suggest lower amplitudes for low-frequency ground motion parameters, and higher amplitudes for high-frequency ground motion parameters, than were predicted by previous models.

REFERENCES

Atkinson, G. (1984). Attenuation of strong ground motion in Canada from a random vibrations approach. Bull. Seism. Soc. Am., 74, 2629-2653.

Atkinson, G. (1989). Attenuation of the Lg phase and site response for the Eastern Canada Telemetred Network. Seism. Res. Let., 60, 2, 59-69.

Atkinson, G. (1993a). Source spectra for earthquakes in eastern i North America. Bull. Seism. Soc. Am., 83, 1778-1798. J

Atkinson, G. (1993b). Notes on ground motion parameters for I eastern North America: Duration and H/V ratio. Bull. Seism. Soc. | Am., 83, 587-596.

Atkinson, G. (1995a). Ground motion relations for use in eastern I hazard analyses. Proc. 7th Can. Conf. Earthq. Eng., Montreal, June 1995. a

Atkinson, G. (1995b). Optimal choice of magnitude scales for seismic hazard analysis. Seism. Res. L., 66, 51-55. I

Atkinson, G., and D. Boore (1990). Recent trends in ground motion and spectral response relations for North America. Earthquake | Spectra, 6, 15-36. I

Atkinson, G., and D. Boore (1995). New ground motion relations I for eastern North America. Bull. Seism. Soc. Am., 85, 17-30. I

Atkinson, G. and R. Mereu (1992). The shape of ground motion I attenuation curves in southeastern Canada. Bull. Seism. Soc. Am., ™ 82, 2014-2031. | Atkinson, G. and P. Somerville (1994). Calibration of time " history simulation methods. Bull. Seism. Soc. Am., 84, 400-414.

Boatwright, J. (1994). Regional propagation characteristics and source parameters of earthquakes in eastern North America. Bull. m Seism. Soc. Am., 84, 1-15. • I -F18-

Boatwright, J. and G. Choy (1992) . Acceleration source spectra anticipated for large earthquakes in Northeastern North America. Bull. Seisin. Soc. Am., 82, 660-682.

Boore, D. (1983). Stochastic simulation of high-frequency ground motions based on seismological models of the radiated spectra. Bull. Seism. Soc. Am., 73, 1865-1894.

Boore, D. and G. Atkinson (1987). Stochastic prediction of ground motion and spectral response parameters at hard-rock sites in eastern North America. Bull. Seism. Soc. Am., 77, 440-467.

Boore, D. and G. Atkinson (1992). Source spectra for the 1988 Saguenay, Quebec earthquakes. Bull. Seism. Soc. Am., 82, 683-719.

Boore, D. and W. Joyner (1991). Estimation of ground motion at deep-soil sites in eastern North America. Bull. Seism. Soc. Am., 81, 2167-2185.

Boore, D. M., W. B. Joyner, and T. E. Fumal (1993). Estimation of response spectra and peak accelerations from western North American earthquakes: An interim report, U.S. Geol. Surv. Open- File Rept. 93-509. 72 pp.

Brune, J. (1970). Tectonic stress and the spectra of seismic shear waves from earthquakes. J. Geophys. Res., 75, 4997-5009.

Burger, R., P. Somerville, J. Barker, R. Herrmann, and D. Helmberger (1987). The effect of crustal structure on strong ground motion attenuation relations in eastern North America. Bull. Seism. Soc. Am., 77, 420-439.

EPRI, 1988. Engineering model of earthquake ground motion for eastern North America. EPRI NP-6074, Electric Power Research Institute, Palo Alto, Calif.

EPRI, 1993. Guidelines for determining design basis ground motions. Early site permit demonstration program, Vol. 1, RP3302, Electric Power Research Institute, Palo Alto, Calif.

Hanks, T. (1982). fmax. Bull. Seism. Soc. Am., 72, 1867-1879.

Hanks, T. and R. McGuire (1981). The character of high-frequency strong ground motion. Bull. Seism. Soc. Am., 7*1, 2071-2095.

Herrmann, R. (1985) . An extension of random vibration theory estimates of strong ground motion to large distances. Bull. Seism. Soc. Am., 75, 1447-1453. -F19- 9

Ou, G. and R. Herrmann (1990). A statistical model for peak ground motion from local to regional distances. Bull. Seism. Soc. Am., 80, 1397-1417.

Somerville, P., J. McLaren, L. Lefevre, R. Burger and D. Helmberger (1987). Comparison of source scaling relations of eastern and western North American earthquakes. Bull. Seism. Soc. Am., 77, 322-346.

SSHAC (1995). Probabilistic seismic hazard analysis: a \ consensus methodology. Senior Seismic Hazard Analysis Committee (R. Budnitz, G. Apostolakis, D. Boore, L. Cluff, K. Coppersmith, A. Cornell, P. Morris). U.S. Dept. Energy, U.S. Nuclear Reg. Comm., Elec. Power Res. Inst.

Toro, G. and R. McGuire (1987). An investigation into earthquake ground motion characteristics in eastern North America. Bull. Seism. Soc. Am., 77, 468-489. .

Toro, G., N. Abrahamson and J. Schneider (1995). Engineering model of strong ground motions from earthquakes in the central • and eastern United States. Earthquake Spectra, submitted. I I 1 I I I I I I I I -F20-

TABLE l - Regression Coefficients for Quadratic Equation

C freq(Hz) 3 c4

0.5 2.27 0.634 -0.0170 0.0000 0.8 2.60 0.635 -0.0308 0.0000 1.0 2.77 0.620 -0.0409 0.0000 1.3 2.95 0.604 -0.0511 0.0000 2.0 3.26 0.550 -0.0640 0.0000 3.2 3.54 0.475 -0.0717 0.000106 5.0 3.75 0.418 -0.0644 0.000457 7.9 3.92 0.375 -0.0562 0.000898 10. 3.99 0.360 -0.0527 0.00121 13. 4.06 0.346 -0.0492 0.00153 20. 4.19 0.328 -0.0477 0.00226 PGA 3.79 0.298 -0.0536 0.00135 PGV 2.04 0.422 -0.0373 0.0000

Notes: Equation gives PSA, PGA in cm/s2, PGV in cm/s, where PSA is the pseudo-acceleration (5% damped), for the random horizontal component on rock. 2 log PSA = cx + c2(M-6) + c3(M-6) - log R - c4 R -F21-

TABLE 2 - Soil amplification factor to be applied to ground motion j relations for rock, to obtain relations for deep soil sites I I Frequency (Hz) log factor* Multiplicative factor I 0.5 0.27 1.9

1.0 0.27 1.9 I 2.0 0.29 2.0 I 5.0 0.24 1.7 I 10. 0.15 1.4

20. -0.03 0.93 1 I Amplification factor is given in log units. -

log PSAsoil = log PSArock + log factor I i I S I I I I -F22-

TABLE 3 - Summary of data for comparison with ground motion predictions

Event M MN m stress(bars) No. obs. dist.(km)

Gaza 82/01/19 4.0 4.8 3.7 86 5 200 -1000 Goodnow 83/10/07 5.0 5.6 4.8 113 13 200 - 800 Nahanni 85/12/23 6.8 6.1 6.2 53 6 8-23 Painesville 86/01/31 4.8 5.3 4.8 149 9 20 -1000 Ohio 86/07/12 4.5 4.9 4.5 154 5 700 -1000 Saguenay FS 88/11/23 4.1 4.6 4.2 190 10 100 - 500 Saguenay 88/11/25 5.8 6.5 6.5 517 29 50 - 700 4.7 5.1 5.4 517 14 30 - 500

Notes: Only mainshocks are included. All records were obtained from the Geophysics Division of the Geological Survey of Canada, m = 2 log Ahf + 3, where Ahf is the high-frequency level of the Fourier spectrum of acceleration (horizontal component), in cm/s, at a distance of 10 km from the earthquake source (Atkinson and Hanks, 1995) . -F23-

ENA Model vs. Brune 100 bars M = 5, 6, 7

C/3 E

CD O

0.01 0.1 1 10 frequency (Hz)

Proposed model Brune 100 bars 1 FIGURE 1 - Comparison of horizontal-component source spectra (R = l km) for the ENA empirical model with I those of the 100-bar Brune model, for M 5, 6 and 7. (from Atkinson, 1993a) I 1 I I I -F24-

Trilinear3 FI (c=0.5) Trir,near3 FI (c=0. 5) 1.0 Hi 2.0 Hz

! -- r : o N

r\Q' 5 5 » s s i ai jo* 5 5 «si» ii'io1 in' i i « s i >«»'irf i 3 « 5 17 11 R Ik ml R lien) (a) (b) Trilinear3 FI Ic=0.5) TnlinearS FI lc=0.5) 7.9 Hz 4.0 Hz -h

I -- s X b

10'. i ' « s • iii'iQi I V 10' R de»)

FIGURE 2 - Observed decay of the mean of normalized spectral amplitudes, for shear waves recorded on the ECTN, for frequencies of 1, 2, 4 and 8 Hz. The height of the vertical bars shows the 90% confidence limits on the means. The smcoth lines are those determined by regression of the data, (from Atkinson, 1993a). -F25-

(Duration - source duration)

I o CO I I I I 0 50 100 150 200 250 300 350 400 450 500 I Hypocentral Distance (km) I

FIGURE 3 - Mean of the total duration minus the source duration fl for ENA data, averaged by 15-km distance bins. Vertical bars 1 show 90% confidence limits on the mean. Trilinear line is that used in the 1995 ground motion simulations; simple straight line is 0.05 R, the distance-duration term used in previous (1987 and 1990) simulations, (from Atkinson and Boore, 1995) i I I -F26-

AB94 Simulations vs. Quad. Eqn. AB94 Simulations vs. Quad. Eqn. f=1.3Hz f = 0.5 Hz 3T

I o

100 1000 1000 R(km) R(km)

f = 3.2 Hz f = 7.9 Hz

100 1000 100 1000 R(km) R(km)

PGA PGV

i O) o

1000 100 1000 R(km) R(km)

a M 4.0 + M 5.5 « M 7.0 M 4.0 + M 5.5 « M 7.0

FIGURE 4 - Predicted response spectral values (PSA for 5% damping) for four frequencies and peak ground acceleration (PGA) and velocity (PGV), for M 4.0 (a), 5.5 (+), and 7.0 (*). Symbols show ground motion predictions. Lines show quadratic equations of Table 1.

(from Atkinson and Boore, 1995) -F27- I Average Event Residuals I I 1 I i -0.4 3.5 4.5 5 5.5 6 6.5 Moment Magnitude I + 1 Hz D 2 Hz * 5 Hz s 10 Hz 1

FIGURE 5 - Differences (residuals in log units) between observed and predicted ground motions as a function of M, for I oscillator frequencies of 1, 2, 5 and 10 Hz (mainshocks only). (from Atkinson and Boore, 1995) 1 I f=1 Hz f=10Hz I I 3 CO 5 1 5 I CD CD o o I I 10 100 1000 10 100 1000 Distance (km) Distance (km) I FIGURE 6 - Differences (residuals in log units) between observed m and predicted ground motions as a function of distance, for • oscillator frequencies of 1 and 10 Hz. (from Atkinson and Boore, 1995) 1 28 -F28- G. M. Atkinson and D. M. Boore

Appendix Table of 5% Damped Pseudo-Acceleration Values (cm/sec2), Peak Ground Acceleration (cm/sec2), and Peak Ground Velocity (cm/sec) (ENA Median Horizontal Component: Rock Sites) (Atkinson and Boore, 1995)

Moment M = 4.50 Log Values for Frequency (Hz) log A 0.5 0.8 1.3 2.0 3.2 5.0 7.9 13.0 20.0 PGA K}V 1.00 0.48 0.75 1.04 1.40 1.74 2.08 2.34 2.56 2.70 2.41 0.56 1.10 0.28 0.58 0.89 1.26 1.63 .94 2.18 2.38 2.53 2.19 0.32 1.20 0.08 0.38 0.72 1.11 1.48 .76 2.04 2.23 2.34 2.00 0.11 1.30 -0.10 0.20 0.58 0.98 1.34 .67 .89 2.08 2.18 .81 -0.09 1.40 -0.31 0.04 0.48 0.83 .20 .53 1.71 1.90 2.00 .64 -0.26 L.50 -0.48 -0.14 0.28 0.67 1.04 .32 1.54 1.75 .84 .45 -0.44 1.60 -0.62 -0.22 0.18 0.57 0.88 1.18 .41 1.56 .64 .26 -0.62 1.70 -0.80 -0.38 0.04 0.40 0.77 1.04 1.26 1.40 .46 .08 -0.80 1.80 -0.92 -0.52 -0.10 0.26 0.59 0.89 .08 .20 .30 0.87 -0.96 1.90 -0.96 -0.57 -0.16 0.20 0.54 0.79 .00 .11 .18 0.75 — .05 2.00 .00 -0.62 -0.19 0.18 0.48 0.79 0.95 1.08 .11 0.68 _ .08 2.10 .00 -0.60 -0.19 0.18 0.51 0.76 0.91 1.00 .04 0.62 — .11 2.20 .06 -0.68 -0.25 0.08 0.41 0.64 0.79 0.88 0.87 0.45 — .22 2.30 1.20 -0.77 -0.36 -0.02 0.28 0.52 0.64 0.68 0.64 0.25 _ .37 2.40 .30 -0.85 -0.47 -0.11 0.15 0.36 0.45 0.41 0.26 -0.01 - .52 2.50 — .37 -0.96 -0.60 -0.28 -0.01 0.18 0.23 0.18 -0.01 -0.22 — .68 2.60 — .52 -1.08 -0.72 -0.42 -0.18 -0.04 -0.01 -0.11 -0.31 -0.46 — .87 2.70 - .64 -1.21 -0.89 -0.59 -0.35 -0.26 -0.28 -0.42 -0.62 -0.72 -2.06 Moment M = 5.00 Log Values for Frequency (Hz) si log* 0.5 0.8 1.3 2.0 3.2 5.0 7.9 13.0 20.0 PGA rev .00 0.73 1.00 1.32 .68 2.08 2.34 2.57 2.74 2.90 2.57 0.75 .10 0.58 0.88 1.26 .63 .92 2.18 2.45 2.60 2.72' 2.38 0.57 .20 0.43 0.74 1.11 .45 .79 2.04 2.28 2.45 2.58 2.21 0.4O .30 0.20 0.60 0.94 .32 .66 .95 2.15 2.28 2.40 2.05 0.21 .40 0.11 0.52 0.86 .20 .53 .79 .98 2.15 2.26 .88 0.04 .50 0.00 0.36 0.71 .08 .38 .65 1.83 1.98 2.08 .70 -0.13 .60 -0.18 0.23 0.58 0.98 .28 .48 1.67 1.82 .90 1.52 -0.30 .70 -0.25 0.08 0.48 0.80 .08 .34 .52 1.65 .72 .34 -0.47 .80 -0.40 -0.03 0.28 0.64 0.96 .20 .36 .46 .54 .13 -0.64 .90 -0.49 -0.09 0.26 0.59 0.90 .11 .26 .38 .41 .01 -0.72 2.00 -0.49 -0.09 0.26 0.59 0.87 .11 .23 .32 .36 0.95 -0.73 2.10 -0.46 -0.10 0.26 0.57 0.85 .08 .20 1.28 .30 0.88 -0.76 2.20 -0.54 -0.17 0.15 0.48 0.76 0.95 .08 1.15 .11 0.73 -0.88 2.30 -0.62 -0.28 0.08 0.38 0.63 0.83 0.92 0.95 0.91 0.54 _ .00 2.40 -0.72 -0.38 -0.04 0.23 0.51 0.64 0.72 0.69 0.53 0.28 - .16 2.50 -0.85 -0.48 -0.15 0.11 0.34 0.49 0.52 0.45 0.26 0.07 — .31 2.60 -0.89 -0.55 -0.29 -0.04 0.18 0.30 0.28 0.18 -0.01 -0.15 — .46 2.70 -1.05 -0.72 -0.43 -0.18 -0.01 0.08 0.00 -0.14 -0.32 -0.40 - .65 Moment M = 5.50 Log Values for Frequency (Hz) = log« 0.5 0.8 1.3 2.0 3.2 5.0 7.9 13.0 20.0 PGA rev .00 0.94 1.28 1.65 2.00 2.32 2.58 2.80 2.93 3.08 2.72 0.93 .10 0.90 1.20 1.54 .90 2.20 2.45 2.65 2.81 2.93 2.56 0.78 .20 0.76 1.08 1.45 .78 2.11 2.34 2.52 2.65 2.79 2.41 0.63 .30 0.64 1.00 1.34 .62 1.95 2.18 2.38 2.52 2.61 2.26 0.47 .40 0.51 0.86 1.20 .51 1.83 2.04 2.26 2.36 2.46 2.10 0.32 .50 0.40 0.75 1.04 .38 .68 .89 2.08 2.23 2.32 .92 0.16 .60 0.26 0.60 0.95 .28 .56 .76 1.94 2.04 2.15 1.75 0.00 .70 0.15 0.49 0.80 .15 .40 .64 .77 .88 .97 .57 -0.15 .80 0.00 0.34 0.67 0.99 .26 .48 .60 .72 .78 .37 -0.33 .90 -0.08 0.26 0.63 0.91 .20 .40 .53 1.61 .66 .26 -0.41 2.00 -0.05 0.28 0.58 0.90 1.18 .36 .49 1.58 .60 .19 -0.43 2.10 -0.06 0.26 0.60 0.88 1.15 .34 1.45 1.52 .52 .14 -0.44 2.20 -0.11 0.20 0.52 0.81 1.08 .23 1.34 1.38 .34 0.97 -0.54 2.30 -0.21 0.11 0.43 0.72 0.93 .08 1.18 1.18 .15 0.78 -0.68 2.40 -0.30 0.04 0.34 0.58 0.81 0.93 0.97 0.93 0.77 0.56 -0.81 2.50 -0.39 -0.07 0.20 0.48 0.66 0.79 0.78 0.71 0.53 0.36 -0.94 2.60 -0.48 -0.21 0.04 0.30 0.49 0.57 0.54 0.43 0.26 0.13 - .12 2.70 -0.60 -0.33 -0.08 0.15 0.30 0.34 0.28 0.15 -0.04 -0.11 -1.26 Ground-Motion Relations for Eastern North America _-pog- 29 I Appendix—Continued I Moment M = 6.00 Log Value* for Frequency (Hz) = logR 0.5 0.8 1.3 2.0 3.2 5.0 7.9 13.0 20.0 PGA rov .00 1.28 .61 2.00 2.32 2.58 \'..81 2.98 3.11 3.23 2.85 1.11 I .10 1.18 .53 .86 2.18 2.45 2..66 2.83 2.98 3.08 2.71 0.97 1.20 1.04 .38 .73 2.08 2.32 2.56 2.70 2.84 2.95 2.58 0.84 1.30 0.95 .28 .61 .93 2.20 '*!.43 ' 2.58 2.72 2.81 2.43 0.69 1.40 0.85 .18 1.49 .86 2.11 2.30 2.45 2.56 2.66 2.30 0.57 I .50 0.71 .04 .38 .67 .94 :!.15 2.32 2.41 2.51 2.12 0.41 1.60 0.61 0.92 .26 .57 1 .83 :-.00 2.15 2.28 2.36 .96 0.29 1.70 0.45 0.80 .15 .43 .70 .86 2.00 2.11 2.20 .79 0.11 1.80 0.36 0.66 0.97 .28 .53 .73 .86 .93 2.00 .61 -0.03 1 1.90 0.28 0.61 0.92 .23 .45 .64 .77 1.83 .89 .50 -0.11 2.00 0.30 0.61 0.94 .23 .45 .62 .73 .81 .83 .43 -0.12 2.10 0.30 0.60 0.92 .20 .43 .59 .68 .75 .76 .36 -0.14 2.20 0.20 0.51 0.85 .15 .34 .49 .57 1.61 .59 .22 -0.23 i 2.30 0.11 0.43 0.73 .00 .23 .34 .43 1 43 .40 .03 -0.36 2.40 0.04 0.32 0.64 0.90 .08 .20 .23 1.18 .04 0.80 -0.48 2.50 -0.04 0.26 0.53 0.78 0.94 .04 .04 0.94 0.76 0.62 -0.61 2.60 -0.12 0.15 0.40 0.62 0.79 0.85 0.81 0.68 0.52 0.42 -0.75 I 2.70 -0.23 0.04 0.30 0.48 0.61 0.60 0.53 0.38 0.23 0.17 -0.89 Moment M = 6.50 Log Values for Frequency (Hz) SI log A 0.5 0.8 1.3 2.0 3.2 5.0 7.9 13.0 20.0 FCA rov l .00 .56 .88 2.23 :2.53 2.77 2.94 3.15 3.26 3.36 2.98 1.27 1.10 .45 .79 2.11 :2.45 2.66 2.85 3.00 3.11 3.26 2.87 1.14 1.20 .32 1.69 :2.00 2.30 2.56 2.72 2.91 3.00 3.11 2.74 1.05 .30 .26 .56 .89 2.18 2.45 2.61 2.76 2.88 2.97 2.60 0.92 i .40 .15 .48 .78 2.08 2.32 2.51 2.65 2.74 2.85 2.46 0.81 .50 .04 .34 .66 .95 2.20 2.36 2.49 2.61 2.71 ' 2.30 0.66 .60 0.89 .23 .53 .86 2.08 2.23 2.36 2.46 2.54 2.16 0.52 .70 0.80 .08 .45 .70 .92 2.08 2.23 2.32 2.38 1.99 0.41 i .80 0.66 0.97 .32 1.58 .80 .95 2.04 2.15 2.23 1.82 0.23 .90 0.60 0.92 .23 .52 .72 .87 .98 2.08 2.08 .71 0.17 2.00 0.60 0.95 .23 .49 .70 .85 .97 2.00 2.04 1.64 0.15 2.10 0.59 0.90 .23 .48 .68 .83 .91 1.95 .97 1.58 0.13 i 2.20 0.53 0.83 .15 .40 .59 .71 .79 1.81 1.81 1.44 0.05 2.30 0.45 0.74 .08 1.32 .46 .58 .64 1.64 1.59 1.26 -0.06 2.40 0.36 0.66 0.97 .20 .36 .45 .45 1.40 .23 1.05 -0.19 2.50 0.28 0.57 0.85 .04 .20 .28 .26 1.18 1.00 0.86 -0.29 i 2.60 0.18 0.46 0.71 0.92 .04 .08 .04 0.91 0.76 0.66 -0 42 2.70 0.08 0.34 0.59 0.79 0.87 0.88 0.79 0.63 0.49 0.45 -0.55 Womenl Si = 7.00 Log Values for Frequency (Hz) = i log« 0.5 0.8 1.3 2.0 3.2 5.0 7.9 13.0 20.0 PGA rov .00 .81 2.18 :Ï.48 2.74 2.98 :Î.15 3.28 3.40 3.51 3.13 1.45 .10 .72 2.08 :2.38 2.61 2.87 :5.04 3.15 3.28 3.38 3.00 1.36 I .20 .61 .96 2.28 :2.56 2.75 2.90 3.04 3.15 3.26 2.87 1.23 .30 .51 .85 2.15 2.41 2.62 2.79 2.92 3.04 3.11 2.74 1.13 .40 .41 .74 :2.04 2.32 2.51 :2.65 2.81 2.92 3.00 2.62 0.99 1.50 .30 .62 .92 2.20 :2.40 :2.56 2.68 2.79 2.86 2.48 0.88 i .60 1.20 .53 .84 2.08 2.28 2.43 2.54 2.64 2.73 2.33 0.74 .70 1.08 .38 .68 .94 2.18 2.30 2.41 2.51 2.58 2.17 0.60 .80 0.95 .28 .58 .83 2.04 2.15 2.28 2.34 2.38 1.99 0.48 .90 0.89 .20 .52 .78 .94 2.08 2.18 2.23 2.28 1.89 0.41 i 2.00 0.87 .18 .51 1.76 .91 2.04 2.15 2.20 2.20 1.83 0.39 2.10 0.90 .18 .48 1.73 .91 :2.00 2.11 2.15 2.11 1.76 0.38 2.20 0.81 .11 .40 .64 .83 .92 1.99 2.00 1.98 1.62 0.30 2.30 0.76 .00 .34 1.54 .72 .81 .84 1.84 1.77 1.46 0.19 i 2.40 0.64 0.97 .26 1.49 .59 1.68 .67 1.60 1.46 1.27 0.10 2.50 0.57 0.88 .15 1.34 .46 1.49 1.48 1.38 1.23 1.09 0.01 2.60 0.49 0.76 .00 1.20 .30 1.32 1.26 1.11 0.98 0.89 -0.12 2.70 0.36 0.64 0.89 .04 .11 .08 .00 0.86 0.73 0.69 -0.24 l i 30 G. M. Atkinson and D. M. Boore -F30-

Appendix—Continued

Moment M » 7.25 jog Values for Frequency (Hz) - log/? 0.5 0.8 1.3 2.0 3.2 5.0 7.9 13.0 20.0 rev .00 .93 2.32 2.63 2.86 :1.04 3.20 3.36 3.45 3.57 3.18 1.55 .10 .83 2.18 :2.51 2.74 2.96 :J.ll 3.23 3.34 3.45 3.08 1.43 .20 .76 2.08 2.38 2.64 2.83 2.99 3.11 3.23 3.32 2.94 1.32 .30 .62 1.98 2.28 2.54 2.74 :2.89 2.99 3.11 3.20 2.82 1.21 .40 .56 1.86 2.18 2.41 2.63 :2.77 2.89 2.«9 3.08 2.70 1.09 .50 .45 .79 :2.08 2.30 2.51 2.64 2.78 2.86 2.93 2.55 0.98 .60 .30 .66 1.93 2.20 2.38 :2.51 2.63 2.72 2.77 2.39 0.85 .70 .20 1.54 1.84 2.08 2.26 2.38 2.49 2.58 2.62 2.25 0.74 .80 1.08 .40 .71 .94 2.11 2.26 2.36 2.41 2.46 2.08 0.59 .90 1.04 1.34 .66 .89 2.04 2.18 2.28 2.34 2.38 .97 0.53 2.00 1.04 .36 .63 .88 2.04 2.15 2.23 2.28 2.30 .91 0.52 2.10 0.99 .32 .61 .86 2.04 :2.11 2.20 2.23 2.23 .85 0.51 2.20 0.95 1.28 .54 .76 .93 :2.04 2.08 2.08 2.08 .73 0.43 2.30 0.86 .18 1.46 .66 .81 1.91 1.94 1.94 1.86 .57 0.32 2.40 0.80 1.08 1.36 .57 .70 1.76 1.79 1.72 1.54 1.37 0.24 2.50 0.71 .04 .28 .46 .56 1.60 1.58 1.48 1.32 .19 0.14 2.60 0.62 0.92 .18 .34 .40 .43 1.36 1.23 1.08 1.01 0.04 2.70 0.52 0.80 .04 .18 .20 .20 1.11 0.98 0.85 0.81 -0.09

324 McLean Avenue U.S. Geological Survey Arnprior, Ontario Menlo Park, California 94025 Canada K7S 3T2 (D.M.B.) (G.M.A.) Manuscript received 12 August 1993. -F31-

DETERMINISTIC SEISMIC HAZARD ANALYSIS 1

Kevin J. Coppersmith I Geomatrix Consultants 100 Pine Street _ San Francisco, CA I

Deterministic seismic hazard analysis (DSHA) is distinguished from probabilistic seismic hazard | analysis (PSHA) more by what it does not contain than by what it does. Deterministic analysis provides an estimate of ground motion (or other hazard descriptor) devoid of an expression of | the probability of exceeding the value; DSHA does not consider explicitly the frequency of | occurrence of earthquakes; nor are uncertainties incorporated explicitly into a DSHA. Despite these absences, DSHA has a long history of application in the development of design ground M motions for critical facilities in the United States. The purpose of this talk to outline the I procedures used in DSHA. The procedures discussed are those that had common application for nuclear power plants in the United States, particularly following Appendix A to 10 CFR 100 and 1 associated Standard Review Plans. It should be remembered, however, that Appendix A applies ' to new power plants and that there has not been an application for a new license since the late 1970's. Hence, the deterministic methods used in support of Appendix A, as described below, I are dated. In the meantime, Appendix A is being revised to emphasize probabilistic methods and • all of the existing power plant sites have been evaluating using PSHA.

The basic elements of a DSHA are the following: 1) identification of seismic sources of potential significance to ground motions at the site, 2) assessment of the maximum magnitude associated with each seismic source, 3) assessment of the ground motions associated with each seismic I source assuming the closest approach of each seismic source to the site, and 4) selection of the source and maximum magnitude that leads to the largest ground motions at the site (usually . called the 'controlling source'). Each of these elements are discussed briefly below. ]

Methods for identifying seismic sources for deterministic analysis differ between the eastern and i western United States. In the West, the focus is on evaluations of the 'capability' of faults. J Capability assessments are based primarily on the recency of fault displacement (e.g., past 500,000 years), structural relationships among faults, as well as the association of seismicity with | faults. In the East seismic sources for DSHA are usually identified as seismotectonic provinces | having similar tectonic characteristics and, presumably, relatively uniform maximum earthquake potential. A site of interest will lie within such a source (the 'host' source), which is usually ] surrounded by other sources. In subsequent ground motion analyses, the closest distance of each I source to the site is used in the analysis. This sensitivity to source boundary location often leads to significant contention regarding the configuration of seismic sources. In the case of the host 1 source, the earthquake is assumed to occur "at the site" in the case where the maximum ' earthquake is defined by intensity. Where the maximum earthquake is defined by magnitude, the location of the maximum earthquake is assumed to occur randomly within 25 km of the site. I -F32-

Maximum earthquakes must be assessed for each identified seismic source. The variety of methods for assessing maximum magnitudes for fault sources (a common problem in the West) will not be discussed here (see Coppersmith, 1991, for such a discussion). In the East, the maximum earthquake is usually assessed based on the largest earthquakes that have occurred within the source zone during the historical period. For example, the largest historical earthquake within a certain source might have been an intensity MM VII. Commonly, the largest historical earthquake is increased by one intensity unit to ensure that a 'reasonably conservative' earthquake is being selected (i.e., an earthquake whose recurrence interval may be longer than the period of historical observation). In more recent assessments, maximum magnitudes—tdihct than intensities-are assessed for each seismic source to take advantage of better ground motion estimators using recorded ground motion data. Because the maximum earthquake estimated for seismic sources is a key determinant of the DSHA results, licensing contention has often centered around the values assessed and their technical basis.

The final step in the DSHA is the assessment of ground motions for the set of maximum earthquakes and closest source-to-site distances surrounding a site. In the case where intensities have been used to characterize the maximum earthquake, the maximum intensity for the host source is usually converted directly to peak acceleration using intensity/acceleration relationships. The maximum intensities for surrounding sources are either attenuated as intensities with distance to the site-and then converted to accelerations, or are converted to magnitudes and then attenuated with distance from the closest source boundary to the site.

It has long been recognized that empirical attenuation relationships are based on data that show considerable scatter or variability. To account for this variability, and to ensure a 'conservative' ground motion prediction, it has been common practice for the deterministic ground motion to be selected as the 84th percentile of the distribution relating magnitude to distance and acceleration amplitude. In more recent DSHAs, considerable effort has been devoted to quantifying the statistical variability of ground motion relationships as a function of magnitude and spectral frequency. This effort is directly related to the importance of the scatter in ground motion predictions for seismic hazard analysis.

In common practice, the ground motions (peak acceleration as well as spectral accelerations at engineering frequencies of interest) resulting from all seismic sources in a region are compared to arrive at the 'controlling' ground motions and source(s). In some cases, different sources will control at different parts of the spectrum (e.g., a nearby source controls at the high frequencies and more distant sources at low frequencies). The design spectrum is usually deemed to be the envelope of the individual spectra resulting from the individual sources.

Perhaps the principal motivation for the deterministic approach and the oft-cited reason for its continued application is the belief that it provides a result that is: 1) easy to understand (i.e., a ground motion value that results from a single seismic source and magnitude) and 2) conservative (i.e., a maximum magnitude is used; at the closest distance; at the 84th percentile). Problems with the approach stem from the fact that uncertainties are not dealt with explicitly and there is no formal inclusion of the frequency of earthquake occurrence. A direct outcome of the former -F33- I is a history of contention in licensing, as applicants, regulators, and intervenors argue about 1 which seismic sources and maximum magnitudes are 'reasonably conservative.' A result of the latter is the fact that the deterministic ground motion values arrived at for the various plant sites | in the United States do not have the same probability of being exceeded. Now that formal ' PSHAs have been conducted, it is seen that the annual probabilities of exceedance of design values (Safe Shutdown Earthquake ground motions) at sites in the central and eastern United I States varies by over an order of magnitude across the sites. *

REFERENCES |

Coppersmith, K. J., 1991, Seismic source characterization for engineering seismic hazard analysis: • Proceedings of the Fourth International Seismic Zonation Conference, Stanford University, CA. £ v.l, p.1-60. 1

I I I 1 I 1 I I