ML20202E053

Updated Probabilistic Seismic Hazard Analysis for Paducah Gaseous Diffusion Plant in Paducah,Ky,Final Rept. Page 4-35 of Incoming Submittal Was Not Included

ML20202E053
Person / Time
Site:	Paducah Gaseous Diffusion Plant
Issue date:	11/24/1997
From:	RISK ENGINEERING, INC.
To:
Shared Package
ML20202E042	List: ML20202E037 ML20202E053
References
NUDOCS 9712050228
Download: ML20202E053 (188)

Text

{{#Wiki_filter:-. 1 November 24,1997 i r UPDATED PROBABILISTIC SEISMIC HAZARD ANALYSIS FOR THE PADUCAH GASEOUS DIFFUSION PLANT PADUCAH, KENTUCKY FINAL REPORT Report Prepared by Risk Engineering, Inc. 4155 Darley Avenue, Suite A Boulder, Colorado 80303 under Purchase Order No. 495153 l LOCKHEED MARTIN UTILITY SYSTEMS, INC. i for the United States Enrichment Corporation Paducah Gaseous Diffusion Plant Paducah, Kentucky under Contract Number USEC-96-C-0001 75A" 288!" B68 %, 8 ppg

TABLE OF CONTENTS Section ! INTRO DUCTION........................................ 1-1 Section 2 PROBABILISTIC SEISMIC HAZARD METIIODOLOGY............. 2-1 2.1 INTRO D UCTIO N.............................................. 2-1 2.2 BASIC SEISMIC HAZARD MODEL.............................. 2 2 2.3 TREATMENT OF UNCERTAINTY............................... 2-6 2.4 CALCULATI ONS......................................... 2.5 R E FE REN CES........................................... ..., 2-9 Section 3 SEISMIC SOURCE CIIARACTERIZATION........................ 3 1 3.1 INTRO D UCTIO N.............................................. 3 - 1 3.2 SEISMIC SOURCES IN THE NEW MADRID SEISMIC ZONE (NMSZ). 3-2 3.3 TIIE WABASH VAI, LEY SEISMIC ZONE......................... 3-5 3.4 0THER SEISMIC SOURCES.................................... 3-5 3.5 RE FE REN CES.......................................,....... 3 -6 Section 4 GROUND-MOTION AND SITE-RESPONSE MODELS................ 4-1

4. I INTRO D U CTION..............................................

4-1 4.2 A*ITENUATION EQUATIONS FOR ROCK........................ 4-1 4.3 SITE-RESPONSE CALCULATIONS AND AMPLIFICATION FA CI'O RS................................................. 4 -4 4.4 VERTICAL GROUND MOTIONS ON SOIL................,.... 4.5 REFE REN CES............................................ ...49 Section 5 RES ULTS..................................................... 5 - 1 5.1 INTRO D UCTIO N.............................................. 5-1 5.2 RES ULTS FO R ROCK......................................... 5-1 5.3 AES ULTS FO R SO IL.,....................................... 5-3 5.J RESULTS IN TABULAR FORM AND OTHER RESULTS..........,. 5-4 5.5 D I S C U S S I O N............................................ 5.6 REFEREN CES........................................... ... 5-6 Section 6

SUMMARY

AND CONCLUSIONS................................. 6-1 6.1 RE FE RE N CES................................................ 6-2 Appendir A GEOLOGICAL AND SEISMOLOGICAL FRAMEWORK Appendix B ANALYSIS OF HISTORICAL SEISMICITY Appeedix C ADDITIONAL SENSITIVITY ANALYSES

Section 1 INTRODUCTION

1. INTRODUCTION This study investigates the probabilistic hazard of earthquake-induced ground shaking at the Paducah Gaseous Diffusion Plant (PGDP), near Paducah, Kentucky. These results will be used to make decisions regarding seismic safety and levels of seismic design at the facility. An express l

purpose of this study is to use as a guideline the most recent studies of seismic hazard in the central and eastern United States, which represent uncertainty in the seismic hazard caused by multiple, altemative hypotheses on the causes and characteristics of earthquakes. Recent intensive studies of the causes and characteristics of earthquakes in the central US have been carried out by numerous investig:, tors and these studies were reviewed and evaluated by Prefs. Roy Van Arsdale and Arch Johnston, who acted as consultants to this study. Detailed references to these studies are included in Section 3. Also, recent studies of ground motions in the central US, and of site response, have been reviewed. In this respect, Dr. Walter Silva and Prof. Robert Herrmann have provided site amplification calculations and consulting on appropriate levels ofinput motions, respectively. The Nucah facility is located at latitude 37.12 nonh and longitude 88.82 west. Structures at the site are founded on deep soil, of approximately 343-foot thickness. Consistent with other recent seismic hazard analyses, we report the distribution of peak horizontal ground acceleration (PGA) and spectral accelerations (PSA) at multiple frequencies. We also show unifonn hazard spectra to demonstrste typical spectral amplitudes and shapes for canhquake ground motions ofinterest. Section 2 of this report summarizes the probabilistic seismic hazard analysis methodology used for this study. The definitions of seismic sources are described in Section 3. Section 4 contains a description of the ground motion models used here, and a description of the site amplification C:\\9120\\rept.\\ SECT 1.WPD November 25, 1991 \\*I

I calculations conducted to evaluate the levels of shaking at the surface of soil. Section 5 contains results in the fonn of seismic hazard curves and spectra, and Section 6 presents conclusions. The three appendices provide additional documentation for this study. Appendix A documents the background for choices of seismic sources and their characteristics. Appendix B describes the earthquake catalog and details of the recurrence laws derivod for seismic sources. Appendix C contains sensitivity results derived from the seismic hazard calculations. cisme sr.ptssten.wro evuxr n,2,n 12 l

Section 2 METHODOLOGY FOR PROBABILISTIC ( SEISMIC-HAZARD ANALYSIS

2.1 INTRODUCTION

This Section describes the methodclogy used to perform the probabilistic seismic hazard evaluation for the Paducah Gaseous Diffusion Plant (PGDP). Section 2.2 provides a broad overview of probabilistic seismic hazard analysis and the formulation used to address temporal clustering oflarge earthquakes in the New Madrid Seismic Zone (NMSZ), Sections 2.3 describes the treatment of uncertainty and introduces some key terms. Section 2.4 describes the steps in the calculations. State-of-the-art seismic hazard studies calculate ground-motion exceedence probailities uslag earth-science hypothews about the causes and characteristics of earthquakes in the region being studied. Scientific uncertainty about the causes of earthquakes in the vicinity of Paducah and about the physical characteristics of potentially active tectonic features and regions lead to uncertainties in the in mts to the seismic hazard calculations. Similarly, seismologists and e earthquake engineers develop equations estimating strong ground shaking in the region, expressing uncertainties in scientific understanding with alternative hypotheses on magnitude and distance scaling. These uncertainties are propagated through the entire analysis. The result is a suite of hazard curves, each with a weight that reflects the probabilities assigned to the associated hypotheses and parameters. These curves quantify the seismic hazard and its uncertainty at the site, and can be used to make decisions regarding wismic design.- In addition, this suite of hazard curves implicitly contains information about the sensitivity of the hazard results to the various assumptions or parameters and about the contributions of these assumptions and parameters to the total uncertainty in seismic hazard. C:\\9720\\rept,2\\$tCT2.WPD - November 2$,1991 21

4 2.2 BASIC SEISMIC IIAZARD MODEL The methodology for probabilistic seismic hazard analyisis (PSilA) is well established in the litenture (Cornell, 1968,1971; Der Kiureghlan and Ang,1975; McGuire, 1976,1978). Calculation of the hazard requires specification of the following three inputs: i 1. Source geometry: the geographic description of the r,eismic source. A seismic source is a portion of the earth's crust associated with a fault, with a concentration of hir, s seismicity or with a general region of the carth's emst having similar geologic characteristics, that may be capable of producing earthquakes. Source geometr. relationship between rupture size and magnitude determine the conditional pr tdig distribution of distance r from the earthquake mpture to the site (given magn.

• r infr\\m).

2. Seismicity; the rate of occurrence v and magnitude distribution funfm) of earthquakes i occurring in each source 1. This characterization includes the maximum magnitude that a source can produce. Magnitude is characterized by the moment magnitude scale M in this study. Both v and/unfm) consider only earthquakes with magnitudes greater i than a minimum magnitude m, typically taken as moment magnitude 5. Smaller carthquakes are assumed to produce no damage to enginected structures, regardless of the ground motion amplitudes they generate. 3. Attenuation function: a relationship that allows the estimation of ground motiori amplitude at the site as a function of earthquake magnitude and distance. This characterization includes both a median amplitude and a standard desiation that describes the anticipated site-to-site and event to event scatter. These inputs are illustrated in Figure 2-1, parts a through c. Figure 2-la shows the geometry of a seismic source and the distance distribution for a given value of magnitude. The distribution of magnitudefun/m) for an area source is often specified as the doubly truncated exponential distribution. Seism: city for a source with the exponential magnitude distribution is C \\9720\\tept_2\\ stet 2.WPD Wovembe r 25, 1997 22

completely specified by the m snum magnitude m, and parameters a and b. Parameter a is a measure of seismic aethity, b is a mesure of relative frequency oflarge versus small events, and log [ tifun/m)) is proportional to b m for m, < m 5 m., The distribution of magnitude fun /m) for a fault is specified by n exponential distributio or a characteristic distribution (Youngs and Coppeesmith,1985; as illustrated in Figure 2 lb). The rate information for these ticee distribution shapes mcy be specified as the rate 5; defined above, the rate orlarge earthquakes (magnitude greater than m, - %), or the slip rate (for faults only). The ground motion is modeled by an attenuation function, as illustrated in Figure 2-Ic. Attenuation functions are usudly of the form In/Af -f(Af,R) + c, where A is ground motion amplitude, Mis magnitude, R is distance, and c is a random varitble that represents scatter. The attenuation functio't is used to calculate Gw(a') = P/A > a*[m,rJ: the probability that the ground motion amplitude A is larger than a', for a given Mant R. The seismic hazard over all sources is calculated as a summation: v, (( G,,(a.)fyn(m)fyn(rire) dm dr v(A>a.)= y (21) in which the summation is performed over all seismic source: / and in which the probability is calculated per unit ti.ne. Equation 2-1 is fonnulated using the assumption that earthquake > (most particularly, successive earthquakes) are independent in size, loca+ ion, and occc4rence t' ne. In most 4 a seismic-hazard applications, primary interest is focused on computing probabilities for high (rare) ground motions (as a result, ti. probability of two exceedencer in time t is negligible). Thus, the quantity on the right side of Equation 2 which is the annut rate of earthquakes with A > a * - is a very good approximation to the probability of ucceding amplitude a* in one year. As a result of the assumption of temporal clustering oflarge earthquake in tl:e NMSZ (see Section 3 and Appendix A for details), the r ste of earthquakes with A > a* is ro; a good C:\\9728\\rept_2\\ SECT 2.WPD 14ovember 25, 1997 23 j 1 E

approximation to the probability of exceeding amplitude a*, as will be sh sw a below The annual rate of earthquakes with A > a can be expressed as l v(a ')=1 x/T1)+2x112)+3x113)+. (22) where F//J is the probability of having / exceedences of amplitude a* in one year, Similarly, the annual exceedence probability _can be expressed as ITA>a' in 1 yr.)=ITl]+P[2)+IT3)+... (2-3) Iflarge earthquakes in the NMSZ occur in clusters with durations much shoner than the inter-arrival time of the clusters, then P/21 and P/JJ are comparable to PflJ for low and moderate values ofa*, causing the exceedence rate in equation 2-2 and the exceedence probability in equation 2-3 to be significantly different. This study incorporates the assumption made in Section 3 and Apper, dix A that la.ge earthquakes on any NMSZ segment are followed by large earthquakes in the other two segments (within a time interval much shoner than the time between these clustering episodes), by treating the three NMSZ segments as one special source. These three NMSZ segments are the Blytheville Arch (BA), the Reelfoot fault (RF), and East Prairie' (EP). The probability of exceeding amplitude a* in one year due to earthquakes in this special seismic source is competed as i ' East Prairie itself will be divided into two seismic sources, namely East Prairie fault (EPF) and East Prairie extension (EPE). Large events in these two faults are trected as mutually exclusive. Therefore, their probabilities are additive. C:\\9729\\rept_2\\ SECT 2.WPD ' November 25, 1991 2-4 (a

MA>a

• in 1 yr.)

v nA >a

• or Aga
• or A,pa *)

u m fnA >a *)+nAga *)+MA pa *) v u s -nA >a *)nAga *) u (2-4) -MA#a *1MA,pa *) -MA 7a *15A >a *) r u +RA fa') Raga *)MA ya') f s wherev is the annual rate of clustering episodes and A >a e represents the event of an w u exceedence of amplitude a* given the occurrence of a large earthquake on the Blytheville arch (which is evaluated using the integral in Equation 2-1). This fonnulation assumes that occurrences oflarge events in the NMSZ are tightly clustered in time (relative to the mean time between clusters). All other assumptions ofindependence are maintained. In particuist, smaller events in the NMSZ segments (those associated with the exponential portion of the magnitude-recurrence model; see Section 3) are treated as ladependent. This modification to the standard formulation is appropriate (and necessary), given the assumption of clustering, it is useful to examine the behavior of Equation 2-4 in order to understand the effect of the clustering assumption on the calculation of the conditional probability of exceeding a* (given the occurrence of a large NMSZ event). We will compare the clustering assumption to a model in which each cluster is treated as a single event (the single-event assumption). The single-event assumption will have an exceedence probability that is only slightly higher than max {P(A >a '), P(Apa '], P(Apa *]). Consider a low value of a*, such that u P[A >a ']=P[Apa *]=P[Apa *]=0.9. The resulting probabi'ity of exceedence (i.e., u P[A >a' or Apa

• or Apa *])is approximately 1.0. Therefore, the clustering u

assumption predicts roughly the same hazard as the single-event assumption, which would predict an exceedence probability of 0.9 to 1. For a high value ofa*, such that _ P(A >a ']=P[Apa *]=P(Apa ']=0.01, the product terms in Equation 2-4 are all negligible. u - As a result, the probability of exceedence is approximately 0.03. On the other hand, the C:\\9728\\rept_2\\ SECT 2.WPD November 25, 1997 25

probability for the single-event assumption is approximately 0.01, hs, the clustering model predicts a higher hazard than the single-event assumption for high ground motion amplitudes. l l The calculation of hazard from all sources is performed for multiple values of a* in order to geneinte the hazard curve, which gives the :winual probability of exceedence as a function of a'. This calculation is performed for multiple ground motion measures (typically, peak ground acceleration, peak ground velocity, and spectral acceleration at multiple frequencies). 2.3 TREATMENT OF EPISTEMIC UNCERTAINTY The most recent seismic hazard studies distinguish between two types of uncertainty, namely epistende and aleatory. Alcatory uncertainty (sometimes called randomness) is probabilistic variability that results from natural physical processes The dze, location and time of the next earthquske on a fault and the details of the ground motion are examples of events considered aleatory. In current practice, these elements cannot be predicted even with collection of additional data,.m the aleatory component of uncertainty is irreducible. The second category of uncertainty is epistemic (sometimes called simply uncertainty), which results from lack of j knowledge about carthquakes and their effects. In principle, this uncertainty can be reduced i with the collection of additional data. I These two types of uncertainty are treated differently in advanced seismic hazard studies. Integration is carried out over aleatory uncertainties to get a single hazard curve (see Equation 2-1), whereas epistemic uncertainties are expressed by multiple assumptions, hypotheses, models, or parameter values. These multiple interpretations are propagated through the analysis, resulting in a suite of hazard curves and their associated weights. Results are typically presented as curves showing statistical summaries (e.g., mean, median, fractiles) of the accedence probability for each ground motion amplitude. The mean and median hazard curves convey the central tendency of the calculated exceedence probabilities. The separation among fractile curves conveys the effect of epistemic uncertainty on the calculated exceedence probability. C:\\9720\\rept_2\\ SECT 2.WPDL November 25, 1997 2-6 ~.

L There are epistemic uncensinties associated with each of the three inputs to the seismic-hazard evaluation, as follows: There is uncensinty about the seisi ogenic potential of faults and other geologic o features in the NMSZ, as a result of(1) uncertainty about the tectonic stresses and deep crustal structure in the region, and (2) incomplete knowledge of these faults geological features. There is also uncertainty about the geometry of these geologic

features, Uncensinty in seismicity is generally divided into uncertainty in maximum magnitude, o

uncertainty in the rate parameter (i.e., activity rate, rate oflarge events, or slip rate), and uncertainty in b or other shape parameters of the magnitude distribution /uM'm), Uncertainty in the attenuation functions arises from uncertainty about the dyamic o characteristics (source, path, and site effects) of earthquakes in the New Madrid region and elsewhere in the central United States. This uncertainty is large because there have been few strong motion recordings in the region. UncertJnty in seismic-source characterization is quantified in this study by considering multiple alternative geometries, multiple magnitude recurrence parameters, and multiple maximum magnitude, as documented in Section 3 and Appendix A. Uncertainty in attenuation equations is quantified by considering the four altemative sets of attenuation equations in EPRI (1993), which represent a range of models for the source and path characteristics of earthquakes in the central and eastern United States (CEUS). Considering these four attenuatior, equations with their weights is analogous to considering multiple attenuation equations developed by different experts. This study follows the SSHAC (Senior Seismic Hazard Analysis Committee; Budnitz et al., 1997) guidance on the treatment of uncertainty and the use of experts in PSHA. SSHAC = defines four levels of PSHA studies, ranging from studies based on a literature review (Level C:\\9728\\rept,2\\ SECT 2.WPD November 25, llo 2-7

1) to studies multiple experts and formal clicitation (Level 4). This study corresponds to SSilAC Level 2.

2.4 CALCULATIONS l Calculations for shaking for one Source-Characterization team proceed in two steps, as follows: 1. Calculation of seismic hazard from each individual source. This calculation is performed for ech combination of attenuation equation and source parameters, resulting in one hazard curve and one weight for each combination. 2. Calculation of total hazard (i.e., the hazard from all seismic sources) and its epistemic uncertainty. For the calculation of quantities other than the mean lutzard, this calculation takes into account the probabilistic dependence introduced by hypotheses in the logic tree that (frect more than one source and by the attenuation equations (i.e., not all combinations of hazard curves Rom two sources A and B are included, only compatible combinations are included). The result of this calculation is a set of mean and fractile hazard curves. (% e compute eleven fractiles in order to cany distribution-shape information into the integration step below.) In addition to these main results, de-aggregation results calctuate and display the contributions of various magnitude-distance-c combinations to the mean hazard. This infonnation is required for the selection of the magnitude-distance-c combinations to use in design. Furthermore, sensitivity results provide insights into the effect of various parameters and assumptions on the calcutated seismic hazard and its uncertainty.

2.5 REFERENCES

Comell, C.A. (1968). " Engineering seismic risk analysis," Bulletin of the Seismological Society of America, v.53, no. 5, pp.1583-1606. c anresr.pt_2ssecr2.m nov. e r 2s, i m 28 )

Cornell, C.A. (1971) 'Probabilhtic analysis of damage to structures under seismic loads," chapter 27 in D>ranic Waws in CivilF>rgineering, WileyInterscience. Der Klureghlan, A. (l975). A Line Sourc,Modelfor Seismic Risk Analysis, Toch. Report UILU ENG-75 2023, University ofIllinois. Electric Power Research Institute (1993). ' Guidelines for determining design basis ground motions." Palo A:to, Calif: Doctric Power Research Institute, vol.1-5, EPRI TR- -102293, vol.1: Methodology and guidelines for estimating earthquake ground motion in eastern North Ar.. erica. vol. 2: Appendices for ground motion estimation. vol. 3: Appendices for field investigations. vol. 4: Appendices for laborstory investigatious. vol. 5: Qmtificatiori of seismic source effects. McGuire R.K. (l976). FORTRAN Computer Programfor Seismic Risk Analysis. OFK 76-67, U.S. Geological Survey, 1 McGuire R.K. (l978). FRISK: Cominterprogramfor seismic risk analysis usingfaults as earthquake sources. OFR 78-1007, U.S. Geological Survey. Senior seismic Hazard Analysis Committee (1995). Recommendationsfor Probabilistic Seismic Hasard Analysis: Guidance on Uncertainty and Use ofExperts. Lawrence Livermere National Labcratory, UCRL-ID-122160. Youngs, R.R., sad K. Coppersmith (1985). ' Implications of fault slip rates and earthquake recurrence models to probabilistic seismic hazard estimates," Bulletin ofthe SeismologicalSociety ofAmerica, v.75, no. 4, pp. 939 064. C:\\9720\\rept,2\\stCT2.WPD Wovember 25, 1997 2-9

g Hupture Site Fault i A. Selsnile Source i e [ Earthquake lueellons in space feed to e distribellen of epleentrol g rlin) distences Ig (rl nil] I t Distance r II. blagnitude distilbution and role of occurrence for Source 1: f (m) gg I (m), vg gg "o '"m a s blagnitude m C. Ground motion estimstlen

• • 7 l

GAlm,r"*) I G (s') tion (lo le) t l r Distance (log scale) D. Probability analysis: P(A > s' in time t)It a I vg jj GAlm,r"*IIhl '"II (rlm)dmdr I I R I 's \\ l'(A > a* In t) /t s s (log scale) N N g N N Ground hiotion Level, s' (log scale) Figure 2-1. Seismic hazard computational model(modified from McGuire and Arabasz,1990) C:\\9728\\rept_2\\ SECT 2.WPD Wovember 24, 1997 2*10

( Section 3 SEISMIC SOL %CE CIIARACTERIZATION

3.1 INTRODUCTION

This Section documents the seismic sources used in this study to characterize seismicity in the New Madrid and Wabash regions, as well as other potential seismic sources in the region. The New Madrid and Wabash regions have been the focus of many seismological and geological studies over the past eight years, resulting in a better understanding of earthquakes in these regions. Still, many uncertainties remain and the existing data are open to alternative interpretations. As a result, the characterization of seismic sources for this study has the dual objective of representing the current state of knowledp shout earthquakes in vicinity of the PGDP and representing the limitations in that state of knowledge. This objective is accomplished by specifying alternative r,eisnde source geometries and seismicity parameters, and assigning weights to them according to th'i. s5;ility. Appendix A, prepared by Professors Roy Wn Arsdale and Arch Johnston of the University of Memphis, contains a detailed summary of geological and seismological studies on the New Madrid and Wabash regions and develops interpretations based on these studies. This information is used to define the seis de sources presented here, to quantify the rates of occurrence for large earthquakes, to specify maximum magnitudes for these sources, and to develop weights for alternative sources and parameters. The rates of occurrence for small and moderate earthquakes are calculated using a statistical analysis of the historical earthquake catalog. Appendix B documents the selection, modification, and analysis of the earthquake catalog. C:\\9728\\rept_2\\ sect 3_.wpd Novernber 25, 1997 3-1

i Traditionally, seismic-hazard studies for sites in the central and eastern Unitt d States have used Nuttli's m magnitude to characterize earthquske size because this is the iragnitude used in all a regional catalogs and in most attenuation equations for the region. Most of the recent information, coming from paleoliquefaction and other geological studies, is provided in the form of moment magnitude M (Hanks and Kanamori,1979), because this magnitude has a physical basis. Recent attenuation equations for the region are provided in terms of M (e.g., Atkinson and Boore,1997) or in terms of both M and mg (Toro et al.,1997). Because this study makes extensive use of paleoliquefaction and geological data to characterize the New Madrid and Wabash zones, it utilizes M to characterize canhquake size. 3.2 SEISMIC SOURCES IN THE NEW MADRID SEISMIC ZONE (NMSZ) Extensive geological, geophysical, and seismological work has been conducted witidn the NMSZ. As a result of these efforts, specific seismogenic faults in the NMSZ have been id stiti't e d studied in detail, particularly the Reelfoot Fault near the town ofNew Madrid. Jotuden r.nd Schweig (1996) have associated each of the three 1811-1812 earthquakes with a specific fault by using historical accounts and geological evidence (Figure 3-1; see also Figures A-18 and A-21). Their interpretation is consistent with the spatial distribution and source characteristics of contemporary NMSZ seismicity (Figure 3-2). This cudy uses those faults, augmented to the north in order to represent more diffuse patterns of seismicity, to characterize the main pattern of seismicity in the NMSZ. The December 11,1811 event is associated with a strike-s'ip rupture on the Blytheville arch - Cottonwood Grove fault or with the Blytheville arch - Pootheel lineament. Both interpretations yield identical results for the PGDP. This uses the former interpretation, with a fault length of 125 km (Figure 3-1). The January 23,1812 event is associated with a strike-slip rupture on the East Prairie fault (EPE) on the northern portion of the NMSZ. This interpretation is supported by fault-mechanics arguments and by limited historical data and is more poorly constrained than those for the other New Madrid events (see Appendix A). The northern portion of the NMSZ is also C:\\9728\\rept_2\\ sect 3.wpd November 25, 1997 3-2 l

the one with the most di&se pattern of seismicity (see Figure 3-2). This dihse seismicity is represented by the " East Prairie extension" (EPE) seismic source, which is shown in Figures 3-3 and 3-4, The remote possibility that fauhs in the East Prairie extension connect with the Fluorspar-district faults is reflected by the long version of the East Prairia extension. Thn February 7,1812 event is associated with a thrust rupture on the Reelfoot fault. This study uses a length of 72 km for the Reelfoot fauh. Datable paleoliquefaction features and displaced geologic units provide a chronology of large pre-historic earthquakes, which complement the historic seismicity catalog. One crucial assumption in this study's interpretation of the paleocarthquake chronology is that a large seismic-moment release in the region involves events on all three NMSZ faults, which occur within a time interval of the order of months or a few years (more precisely, within a time interval shorter than the temporal resolution of the paleoearthquake chronology). This - assumption is supported by the 1811-1812 events, by the observation that the history of displacement on the Reelfoot fault is consistent with the paleoliquefaction history (even though the paleoliquefaction features at the northern and southern extremes of the NMSZ could not have been caused by events on the Reelfoot fault), and by the observation that the Reelfoot and - Ridgely faults have similar displacement histories. Appendix A contains a detailed discussion of these issues. As a consequence of this assumption, each of the four NMSZ events in Figure A-8 is considered to have occurred in each one of the three NMSZ faults. Another consequence of this assumption is t.d the occurrences oflarge earthquakes in the NMSZ are not independent 1 in time. As a result, the standard PSHA asumption of temporal independence between events must be modified (see Section 2.2). Based on the paleocarthquake chronology discussed above, each of the three faults is assigned mean recurrence intervals of 500 to 1,000 years (see Table A-1). Based on strain-rate considerations, the 1000-year recurrence interval is given more weight and the 500-year I Ct\\9720\\rept_,2\\ sect 3_.wpd November 25, 199'1 3-3

interval is given a lower maximum magnitude with a probability of 0.5. The resulting logic trees for the rates oflarge earthquakes and maximum magnitudes on the three NMSZ faults are shown in Figures 3-5 through 3-8. The logic tree for East Prairie and East Prairie extension is somewhat more complicated, in order to avoid double-counting. The rates of small and moderate earthquakes in the NMSZ faults were computed using the Mueller et al. (1996) catalog (see Figure 3 9), which is based on the NCEER and other catalogs. For the purposes of seismicity calculations, the Reelfoot rift was sub-divided into areas associated with each of the NMSZ faults, as showr. in figure 3-10. Each event was assigned to the fault corresponding to the area where it falls. This approach is necessitated by the location errors of earlier historical canhquakes. Maximum-likelihood calculaticas were performed for each fault separately and for the rift as a whole. The b value obtained for the rift as a whole was assigned to each hxlividual fault, rather than the b value obtained for that fault, because the former is more stable. Half the historical seismicity in the East-Prairie Extension area was assigned to the East Prairie fault; the other half was assigned to the East-Prairie extension. Table 3-1 lists the rates and b values for the NMSZ faults and for all seismic sources considered in this study. The combined magnitude-recurrence model of the three NMSZ faults and the East Prairie extension is characteristic, with the exponential portion controlled by historical seismicity and the characteristic portion controlled by paleoseismic and geological ' formation (see Figures 3-m 11 through 3-14). These figures show the historical seismicity rates (error bars), the maximum-likelihood fit for that source (thin straight line), and the combined magnitude-recurrence model (heavy lines: solid, mean; dot-dash,15th and 85th percentiles). These figures also contain the effect of uncertain maximum magnitude; this effects sometimes produces unusual curvature changes at the higher magnitudes. It is also useful to show the magnitude-recurrence models for the combined East Prairie and East Prairie extension (Figure 3-15)and for the entire rift (Figure 3-16). C: \\ 9720\\ rept,2\\ sect 3,,.wpd November 25, 1997 3-4

l Events in the characteristic portion of the magnitude-recurrence models for the three NMSZ segments are treated as occurring in temporal clusters, so that if one event occurs in one of these segments, events will occur in the other NMSZ segments, with a time delay much shorter than the mean time between clusters (EPF and EPE are consider one segment for this discussion). Events in the exponential portion of the magnitude-recurrence models are treated as independent (in the usual way). Section 2.2 presents the mathematical formulation of the clustering model. 3.3 THE WABASH VALLEY SEISMIC ZONE Appendix A considers two alternative geometries for the Wabash Valley seismic zone. These geometries differ in their East-West extent, but both have their southeia boundary at ti e Rough Creek graben (see Figure 3-1, as well as Figures A-22 and A-22a). This study considers only the large geometry. Results for the small geometry are expected to be similar. The rate of occurrence oflarge earthquakes and the maximum magnitude are determined in Appendix A on the basis of recently obtained paleoliquefaction data. The rates of small and moderate earthquakes were computed using the Mueller et al. (1996) catalog (see Figure 3-9). The resulting magnitude-recurrence model is shown in Figure 3-17. 3A OTHER SEISMIC SOURCES The PGDP is situated in the intersection of the Reelfoot rift, which contains all the active NMSZ faults, and the rough Creek graben, which has a low rate of activity. The boundary between these two structures is not well dermed. This study considers two alternative locations (Figure 3-1S). This study considers a source zone to represent earthquakes in the Reelfoot rift outside of the NMSZ faults and the EPE. Because all historical seismicity on the rift was assigned to the NMSZ faults and the EPE, one could assume zero seismicity for the Reelfoot rift source zone. Instead, this study assumes a rate equal to 10% of the total historical seismicity on the rift. As a result, this study accounts for 110% of the historical seismicity on the Reelfoot rift. This is .C:\\9128\\rept,2\\ sect 3,,.wpd November 25, 19p 3-5 )

I l within the error bars of the historical data and allows us to account for the possibility of moderate earthquakes occurring off the main NMSZ faults. Figure 3-19 shows the magnitude-recurrence model for the Reelfoot rift source zone. The magnitude-recurrence model for the Rough Creek graben are determined from the historical seismicity on the graben (see Figure 3-20). The maximum magnitudes for the Reelfoot riR and Rough Creek graben are specified on the basis of Appendix A (see Table A-1). N We also define two background zones adjacent to the edges of the Reelfoot rift. The background zone to the west (Ozarks region) is the more active of the two. The maximum magnitude for these source zones is selected as 6.7*0.5. The magnitude-recurrence parameters for these sources are listed in Table 3-1. Appendix A also discusses the Commerce-Benton Hills and SE Flank faults. Preliminary calculations indicate that these faults have negligible contribution to the hazard at the PGDP. These faults are not considered further in this study.

3.5 REFERENCES

Atkinson, G.M. and D.M. Boore. (1997). "Some Comparisons Between Recent Ground-Motion Relations." SeismologicalResearch Letters, v.68, no.1, pp. 24-30. Hanks, T., and H. Kanamori (1979) "A Moment Magnitude Scale." J. Geoph. Res., 89, B5, May. Johnston, A.C., and Schweig, E.S.,1996. The enigma cf the New Madrid earthquakes of 1811-1812. An. Rev. Earth Planet. Sci., 24: 339-384. Mueller, C., M. Hopper, and A. Frankel (1996). Preparation ofEarthquake Catalogsfor the Interim NationalSeismic Ha:ardMaps: Documentation,1anuary 19. C:\\ 9728\\ rept,2\\ sect 3,,,.wpd November 25, 1997 3-6

Toro, G.R., N.A. Abrahamson and J.F. Schneider (1997). A Model of Strong Ground Motions from Earthquakes in Central and Eastem North America: Best Estimates and Uncertainties. SelsmologicalResearch letters, v.68, no.1, pp. 41-57. C:\\9728\\r pt,2\\ sect 3_.wpd November 25, 1997 3-7

Table 3-1 Seismicity Parameters (Exponential Portion) Seismic Source (i) v, a(In[v,]} b o(b) P E. Prairie Extension 4.5E-03 0.30 0.86 0.05 0.83 E. Prairie fault 4.5E-03 0.30 0.86 0.05 0.83 Reelfoot fault 9.3E-03 0.28 0.86 0.05 0.90 Blytheville arch-CGF 7.9E-03 0.29 0.86 0.05 0.89 Wabash 2.4E-02 0.35 0.80 0.09 0.93 Reelfoot riR 2.7E-03 0.83 0.86 0.16 0.96 Rough Creek graben 5.3E-03 0.77 0.65 0.15 0.94 Western background 8.9E-03 0.50 0.79 0.10 0.% Eastern Background 9.3E-04 1.09 0.80 0.21 0.9 e Notes:

1. v, is the annual rate of earthquakes with M>5 in source i 2 p is the correlation coefficient between in[v,) and b.

C:\\9728\\rept_2\\ sect 3_.wpd November 25, 1997 3-8

l l + + + + + + + + WabashValleyZone Large I { + + + + + .g% d. GDP alrie Fault Y y~ \\ + + + + + + t S + + + + + + + + h Legend

• PGDP Site KM

- Fault C o 50 100 OSSZ g Figure 3-1. Map showing NMSZ and other seismic sources considered in this study. C:\\9728\\rept_2\\ sect 3_.wpd November 25, 1997 3-9 I

+ + + + + + + l e e 9 + + %g + + + + I 9 f .. g**4 + ' } k.. 6

• e

+ + + , 4 r. A 4

8.. y.

• 9P g-

.,e,- =

g.,.

g +.,* + + + + + + g* .y. ., i '. . 'g', s* s .4g t' ,k +. + + + + g lf f 4 s f e + + + e+ + + + l P Legend Magnitude (Mw)

• 3.0 > M > 2.0 KM G M>5.0
• PODP Site m

M < 2.0 o so too 9 5.0 > M > 4.0 e Netwak EQ 1974-94 o 4.0 > M > 3.0 Figure 3-2. Map showing events in the New Madrid earthquake catalog (1974-1997). C:\\9728\\rept_,2\\ sect 3_.wpd November 25, 1991 3-10

+ + + + s + t t t ast Prairie E on - La PGD!' Eas airie Extention - Small + + + Legend lir PGDP Site y O ^reasource Figure 3-3. Map showing the two alternative geometries of the East Prairie fault extension (EPE). C:\\9728\\rept_2\\ sect 3_.wpd November 25, 1997 3-11

l l 4 + + + + + t + o'

• • • o o o
    • oooo
• o,o o' *
• oo p

,0o,o,o, o oo s o PGDP," + + + io ed ) + + + Legend - E. Prairie Exension - Shon KM = = E. Prairie Extension - Long 2c e

• PGDP Site Figure 3-4. Map showing the faults used to represent the East Prairie fault extension (EPE)in the seismic hazard calculations.

1 C:\\972B\\rept_2\\ sect 3_.wpd November 25, 1997 3-12 l

1 l 1 Rate of Large Maximum Earthquakes Magnitude 1/1,000yr 8.1 +/- 0.3 0.7 1.0 0 7.7 +/- 0.2 0.5 1/ 500 yr a6 0.3 8.1 +/- 0.3 0.5 Figure 3-5. Logic tree for the rate oflarge events and maximum magnitude on the Blytheville Arch-CGF fault. The logic tree for the Reelfoot fault is identical. C:\\9728\\rept_2\\ sect 3_.wpd November 25, 1907 3-13

Rate of Large Rate of Large l Earthquakes, Earthquakes, EP + EPE Each Source EPE: 1/1,000yr EP: 0 (-1/10,000yr) 0.171 1/1,000yr 0.7 EPE: 1/10,000yr EP: 9E-4 (~ 1/1,000yr) 0.829 EPE: 1/1,000yr EP: 1/1,000yr 0.6 1/ 500 yr 0.3 EPE: 1/10,000yr EP: 1.9E-3 (- 1/ 500 yr) 0.4 Figure 3-6. Logic tree for the rates oflarge events on the East Prairie fault and East Prairie extension. The trees for the corresponding maximum magnitudes and source geometries are shown in Figures 3-7 and 3-8. C:\\9728\\rept,2\\ sect 3_.wpd November 25, 1997 3-14

Maximum Magnitude 8.1 +/- 0.3 O@@= 1.0 (Rates 1/1000yr) 7.7 +/- 0.2 0.5 (Rate-1/500yr) 8.1 +/- 0.3 0.5 Figure 3-7. Logic trees for the maximum magnitude of the East Prairie fault. C:\\9728\\rept_2\\ sect 3_.wpd November 25, 1997 3-15

Maximum Geometry Magnitude Short (100 km) 8.0 OO-1.0 (Rata = 1/1,000yr) Short (100 km) 8.0 .w 0.857 1.0 @@c (Rate = 1/10,000 yr) Long (160 km) 8.2 0.143 1.0 Figure 3-8. Logic tree for the maximum magnitude and geometry of the East Prairie extension (EPE). C:\\9728\\rept_2\\ sect 3_.wpd November 25, 1997 3-16

+ + + +. + e + + l o o + + + +. o +. + + .g. % O,

• l

.G s g. ,o.. o d o e + + + oo 0+ o + +

• O. 6 e

s.. O o c o o o a

DP OO

+ + +' G. d ce, e + pg o.o se o 9 e f S. o + + + o, + + + oo o a .o o. O e o. d' e e + o + g o o 4 +o + + + + + i Legend Magnitude (Mw)

• PGDP Site 6.5 > M > 7.5 o 3.5 > M > 4.5 o

so 100 O CEUS Historical Eart. quakes O 5.5 > M > 6.5 M < 3.5 Figure 3-9. Map showing events in the historical earthquake catalog. C:\\9728\\rept 2\\ sect 3.wpd November 25, 1997 3-17

I i + t v t t + + + + + + + + + + + + + + + + P P + + + + + + + + East airie gg e*Mt Rft # + + + + + + + lythevide r + + + ) t. + + Legend

• PGDP Site KM Polygons used to Compute Seismicity 0

00 g,8 Rift Boundary Figure 3-10. Map showing the areas used to assign historical seismicity to the NMSZ faults. I l C:\\9720\\rept_2\\ sect 3,.wpd November 25, 1991 3-18

1 E. Prairie Flt ~o d ^ E ; e

1. 4 IL1 m

o O ~ E 4 N M 3 \\. 5 E E l 2 3 4 5 6 7 8 MAGNITUDE (in) Figure 3 11. Magnitude. recurrence model for the East Prairie fault. The error bars and maxunum-likelihood line correspond to half the historical seismicity in the East Prairie extension, s C:\\97?8\\tept_2\\ sect 3_.wpd November 25 1991 3-19 _ _a

E. Prairie Extension R. E A h r v2 bBs j m m o O Q g~ E 2a !. q~ : 6.. g. ~, L ,N. N E_ h 2 3 4 5 6 7 8 MAGNITUDE (m) Figure 3-12. Magnitude-recurrence model for the East Prairie extension. The error bars arJ maximum-likelihood line correspond to half the historical seismicity in the East Prairie rxtension. C:\\9728\\rept_2\\ sect 3.wpd November 25, 1997 3-20 = _.

Reelfoot fault ~,o, E ^ h : e p1 i = m 8: ; N:. @- =~ s.%,,%.~.%- s Ig d- :5

• .~.s.s

\\ S i i a I I: 2 3 4 5 6 7 8 MAGNITUDE (rn) Figure 3-13. Magnitude-recurrence model for the Reelfoot fault. C:\\ 9728 \\ rept_2\\ sect 3_.wpd November 25, 1997 3-2l

Blytheville ARCH + CGF i-E E A k m =. ~ w m O E s. N. ?;-- = u. 's ~ ~s x d i \\' N 3 1-i I sum .t. t. I T T 2 3 4 5 6 7 8 MAGNITUDE (r2) Figure 3-14. Magnitude-recurrence model for the Blytheville Arch - Cottonwood Grove fault. C:\\9728\\rept_2\\ sect 3_.wpd November 25, 1991 3-22

i E. Prairie Fit + E. Prairie Extension R; l 5 5 A v2 b#s i N o g O ~ p E ' s o o' p%. E s x i 1 n 1 1 l eue .t f, t. I 9. 2 3 4 5 6 7 8 MAGNITUDE (m) Figure 3 15. Magnitude-recurrence model for the combined East Prairie fault and East Prairie extension. C:\\t?28\\rept_2\\ sect 3_.wpd November 25, 1997 3 23

REELFOOT RIFT (incl. all NM faults and E. Prairie ex h; i-E ^E i ra

1. s !

m m O E o m 3 H

• 7 N.

o 2a ; \\ 's N; q~ j g L \\ 9 2 2 3 4 5 6 7 8 MAGNITUDE (ni) Figure 3 16. Magnitude recurrence for all seismic wurces in the Reelfoot rift (i.e., East Prairie fault, East Prairie extension, Reelfoot fault, Blytheville arch. CGF, and Reelfoot rift source zone). C:\\9720\\rept_2\\ sect 3.wpd November 25, 1997 3 24 l

WABASH i-i E ^ E i m k

1. s U1

~ N.% %O E N* mF 3 \\, N o g1 1 s s s s 5 g .y. h h Ys' .i. 2 3 4 5 6 7 8 MAGNITUDE (m) Figure 3-17. Magnitude-recurrence model for the Wabash Valley seismic zone. C:\\9728\\rept,2\\ sect 3_.wpd November 25, 1997 3-25

l l { + + + + + + + + 5 + + + + + + + + + + + + + 's h a Rough Creek ( raben + + 4 + PGD R foot Ri + + + + + + + + + + l + + + + + + + Legend KM PGDP Site 0 50 100 3 Figure 3-18. Map.ehowing the two alternative locations of the boundary between the Reelfoot rift and the Roug'i Creek graben. Solid, boundary based on seismicity (60% weight); dashed, boundary based on geology (40% weight). I C:\\9728\\rept_2\\ sect 3_.wpd November 25, 1997 3-26 l

l' REELFOOT RIFT "o E ^E i = ra .;\\ s,e p.1 g O g c %~ [ N.% N. \\ i \\. \\. To ~ \\. E N M .f f, t. .t t 2 3 4 5 6 7 8 MAGNITUDE (m) Figure 3-19. Magnitude-recurrence model fc,r the Reclfoot rift seismic source zone. The historical seismicity corresponds to 10% of the seismicity of the Reelfoot rift. C.,3728\\rept_2\\ sect 3_.wpd November 25, 1997 3-27

Rough Creek graben i-i- E E A 1 = V3 b Z l p.1 m ~ l O E .*%,N g% s N, E N. \\ L. , = N a I W =- I - I 1. t ,t, 2 3 4 5 6 7 g MAGNITUDE (m) Figure 3-20. Magnitude-recurrence model for the Rough Creek graben seismic source zone. i s 4 4 C:\\9728\\rept_2\\ sect 3_.wpd November 25, 1997 3-28 4

l l 1 Section 4 GROUND-MOTION AND SITE-RESPONSE MODELS

4.1 INTRODUCTION

Cuculation of ground motion for given magnitude and distance is performed in two steps; namely, calculation of ground motions on rock using a rock attenuation equation, and multiplication by a site amplification factor. This Section documents the rock attenuation equations used in this study and the development of site specific soil amplification factors. 4.2 ATTENUATION EQUATIONS FOR ROCK Ground motior* for rock are calculated using the attenuation equations recently developed by EPRI (EPRI, H v3), as s. lightly revisd and extended by Toro et al. (1997). These attenuation equat!ons are of the form inWC +C (Af-6)+C (Af-6)? i 2 3 - C In R - (C -C ) max in( 100), 0 -C R (41) y 3 y + c, + c, u=[+C (4-2) R with coeflicierts given by Table 2 of Toro et al. (1997). In the above equation, M is moment magnitude, R is horizontal distance, c,is epistemic uncertainty, and c, is aleatory uncertainty. The EPRI attenuation equations include the effect of crustal structure and contain a thorough treatment of epistemic and aleatory uncertainty in source characteristic:, path effects, and near - site anclastic attenuation (kappa). Aleatory uncertainty is treated as magnitude-and distance-g dependent (see Toro et al.,1997 for details). Epistemic uncertainty (i.e., e, in Equation 4-l) is treated as magnitude-dependent and is modeled in the seismic-hazard calculations by using C:\\972B\\rept,2\\St.CT4.WPD November 25, 1997 4*l

four separate attenuation equations and their associated weights. These four attenuation equations differ from the corresponding median attenuation equation in the values of coefficients C and C (see Table 2 ofToro et al.,1997). Figures 4-1 and 44 show these four i alternative attenuation equations, as well as the median attenuation equation, for peak ground acceleration (PGA) and 1 Hz spectral acceleration (PSA) at 1 Hz. Figures 4-1 and 4 2 a:so kw the ground motion amplitudes predicted by lierrmann (personal communication,1997) for M 5. He uses the Atkinson Boore (1997) source spectrum and models of geometric and anelastic attenuation based on t'.e analysis of nearly 2200 vertial ground motion records from small earthquakes in the New Madrid region (Samiezade-Yadz, et al.,1996). Herrmann's predictions for PGA are consistent with the EPRI attenustion equations. His predictions for 1 Hz are significantly lower, as a result of the Atkinson Boore two-corner source spectrum, which predicts much lower amplitudes near 1 Hz. Figure 4 3 compares the EPRI (1993) and Atkinson Boore (1997) predictions for 1 Hz spectral acceleration and multiple magnitudes, and shows that the difference between the two sets of predictions becomes larger at higher magnituA s. The issue of one-corner (or Brune) vs. two-corner spectra has not been settled. This study uses the Brune model for the sake of conservatism and consistency with recent practice. L The EPRI attenuation equations consider the effects of rupture size for small and moderate earthquakes, but not for large earthquakes with extended ruptures. For calculations involving the faults in the NMSZ faults and the East Prairie extension, especially for close distances, it is - necessary to utilize attenuation equations that consider the potentially large dimension of the earthquake rupture. These attenuation equations must use closest distance to the rupture (or some similar measure) to charactedze distance. They must also include the effects of elongated ruptures (i.e., only a portion of the energy release occurs near the site) and the effect of possible variations in source scaling (i.e., large earthquakes are postulated to have lower stress drop). These effects are usually denoted as magnitude-saturation or extended source effects and are used in nearly all attenuation equations for California. l C:\\9720\\rept,2\\ SECT 4.WPD November 2$, 1997 4-2

Two approaches are followed here to introduce extended source efects. The first approach (which we denote as the Empirical Approach) uses a result by Atkinson and Silva (1997), who utilized the extensive strong motion database from California. In order to fit the short. distance California data with a point source model (like that used to derive the EPRI attenuation equations), Atkinson and Silva found that it was necessary to reduce the stress drop by a factor of two between magnitudes 5.5 and 7.5. This reduction does not imply a factor of two reduction in the actual stress drop; it simply implies that the combined efect of geometric efects and source scaling is equivalent to a factor of two reduction in stress drop. This efect may be represented by substituting Equation 4-2 above with the Equation R =/R +C,2[exp(a+bM)]2 (4-3) 2 y where R is the closest horizontal distance to the rupture. The second approach (which we denote the Modeling Approach) uses extended source ground motion modeling to determine the shape of the am.olitudc vs. rupture-distance curve for large magnitudes. Thus, this approach considers only the geometric efects of extended ruptures, Details on this modeling approach are contained in Risk Engineering (1993). The resulting estet may be represented by substituting Equation 4 2 with the equation R =R 0.006 exp(mm) (44) y f for attenuation equations in terms of m, or u R =R 0.089 exp(0.6M) (45) f u for attenuation equations in terms of moment magnitude, in the above two equations, R is the f shortest (slant) distance to the fault rupture. . The inedian ground motion amplitudes predicted by the attenuation equations modified according to the Empirical and Modeling approached are shown in Figures 4-4 and 4 5, where Ct\\9728\\rept,2\\ SECT 4.WPD November 25, 1991 4-3

they are compared to the amplitudes predicted by the model with no saturation. This study uses the following weights for the saturation approrches for the fault sources: Empirical,0.4; Modeling,0.4; and no saturation,0.2. Calculations for the area sources use no saturation. 4.3 SITE RESPONSE CALCULATIONS AND AMPLIFICATION ' ACTORS The mechanical properties of the t. oil deposits underlying the site have a profound effect on the amplitude and frequency content of the earthquake ground motions at or near the roil surface (e.g., Seed et al.,1976; Mohraz,1976; Joyner and Fumal,1984; Schneider et al.,1993). Ideally, one would quantify these effects based entirely on strong motion recordings obg\\ ined at the site, but this is rarely possible in practice. Thus, analytical methods or empirical methods based on recordings from other sites must be used to quantify site response. Equivalent Linear Computational Scheme The standard computational scheme to evaluate one-dimensional site response assumes vertically propagating plane shear-waves. Departures of soil response from a linear l constitutive relation are treated in an approximate manner through the use of the equivalent-linear approach. The equivalent linear approach, in its present form, was introduced by Seed and Idriss (1970). This scheme is a particular application of the general equivalent-linear theory developed by Iwan (1967). Basically, the approach is to approximate a second order nonlinear equation, over a limited range ofits variables, by a linear equation. Formally this is done in such a way that the average of the difference between the two systems is minimized. This was done in an ad hoe manner for ground response modeling by defining an effective strain which is assumed to exist for the duration of the excitation. This value is usually taken as 65% of the peck time-domain strain calculated at the midpoint of each layer, using a linear analysis. Modulus redmion and hysteretic damping curves are then used to define new parameters for each layer based on the effective strain computations. The linear response calculation is repeated, new effective strains evaluated, and iterations performed until the changes in parameters are below C:\\9728\\rept_2\\ SECT 4.WPD Novemt>e r 25, 1997 4*4

some tolerance level. Generally a few iterations are sufficient to achieve a strain compatible linear solut es t This step wise analysis procedure was formalized into a one-dimensional, vertically propagating shear wave code called SHAKE (Schnabel et al.,1972). Subsequently, this code has easily become the most widely used analysis package for one-dimensional site response calculations. The advantages of the equivalent linear approach are that parameterization of complex nonlinear soil models is avoided and the tuathematical simplicity of a linear analysis is preserved. A truly nonlinear approach requires the specification of the shapes of hysteresis curves and their cyclic dependencies through an increased number of material parameters. In the equivalent linear methodology the soil data are utilized directly and, because at each iteration the problem is linear and the material properties are frequency independent, the ' damping is rate independent and hysteresis loops close. Careful validation exercises between equivalent linear and fully nonlinear formulations using recorded motions from 0.05 to 0.50g showed little difference in results (EPRI,1993). Both formulations compared very favorably to recorded motions suggesting both the adequacy of the vertically propagating shear wave model and the approximate equivalent linear formulation. While the assumptions of vertically propagating shear-waves and equivalent-linear soil response certainly represent approximations to actual conditions, their combination has achieved demonstrated success in modeling observations of site effects and represent a stable, mature, and reliable means of estimating the effects of site conditions on strong ground motions (Schnabel et al.,1972; Silva et al.,1988; Schneider et al.,1993; EPRI,1993). RVT Based Computational Scheme The computational scheme employed to compute the site response for this project uses an alternative approach employing random vibration theory (RVT). In this approach the control motion power spectrum is propagated through the one-dimensional soil profile using the plane-C:\\912t\\rept.2\\SCCT4.WPD Wovember 25, 1991 4*5 1/*

wave propagators of Silva (1976). In this formulation only horizontally polarized shear waves (Sil waves) are considered. Arbitrary angles ofincidence may be specified but normal incidence is used throughout the present analyses. In order to treat possible material nonlinearities, an RVT based equivalent linear formulation is employed. Random process theory is used to predict peak time domain values of shear strain based upon the shear strain power spec;.A m, Mhis sense the procedure is analogous to the program SIIAKE except that peak shear.ws MilAKE are measured in the time domain. The purely frequency domain approach obviates a time domain control motion and, perhaps just as significant, eliminates the need for a suite of analyses based on different input motions. This arises because each time domain analysis may be viewed as one realization of a random proens. Different control motion time histories reflecting different time domain characteiistics but with nearly identical response spectra can result in different nonlinear and equivalent linear response. In this case, several realizations of the random process must be sampled to have a statistically stable estimate of site response. The realizations are usually performed by employing different control motions with approximately the same level of peak accelerations and response spectra. In the case of the frequency domain approach, the estimates of peak shear strain as well as oscillator response are, as a result of the random process theory, fundamentally probabilistic in nature. For fixed material properties, stable estimates of site response can then be obtained with a single run. In the context of the RVT equivalent linear approach, a more robust method ofincorporating uncertainty and randomness of dynamic material properties into the computed response has been developed. Because analyses with multiple time histories are not required, parametric uncertainty can be accurately assessed through a Monte Carlo approach by randomly varying dynamic material properties. This results in mean as well as percentiles (e.g.16*, median, 84*) of smooth response spectra or amplification factors at the surface of the site. I C:\\9728\\rept,2\\ SECT 4.WPD November 25, 1997 4*b

In order to randomly vary the shear wave velocity profile, a profile randomization schema has been developed, which varies both layer velocity and thickness. The randomization is based on a correlation model developed from a statistical analysis on about 500 measured shear wave velocity proAles (EPRI,1993; Silva et al.,1997). Profile depth (depth to competeat material) is also varied on a site specific basis using a uniform distribution. The depth range is selected to reflect expected variability over the structural foundation as well as uncertainty in the estimation of depth to competent material. To accommodate uncertainty in modulus reduction and hysteretic damping curves on a generic basis, the curves are independently randomized about the base case values. A log-normal distribution is assumed with a o of 0.35 at a cyclic shear strain of 3 x 10*4. These values are m based on a statistical analysis oflaboratory test results. - An upper-and lower-bound truncation cf 20 is used to prevent modulus reduction or damping models that are not physically possible. The random curves are generated by sampling the transformed normal distribution with a o, of 0.35, computing the change in normalized modulus reduction or percent damping at 3 x 10*4 shear strain, and applying this factor at all strains. The random perturbation factor is reduced or tapered near the ends of the strain range to preserve the general shape of the median curves (Silva,1992). Site Specific Parameters for PGDP - The shear wave velocity profile for the PGDP is shown in Table 4-1 and Figure 4-6. This best-estimate profile was based on the four measured profiles at the plant (Staub and Wang, 1991; see Figure 4-7). The depth to competent material is taken at 343 ft-the average depth to Limestone basement material at the only boreholes (3 and 4) to reach bedrock (Staub and Wang,1991). Profile depth was then randomized between 393 and 293 ft along with the velocities and layer thickness to reflect uncertainty and variability over the facility area. To accommodate nonlinear soil response, the generic cohesionless soil G/G and hysteretic damping curves developed by EPRI (1993) were used. These curves accommodate the effects of confining pressure and were developed based on laboratory tests as well as a careful ce \\9120\\ropt_2\\ SECT 4.WPD November 2b 1991 4*7 1

literature review of recent test results. The curves are appropriate for profiles comprised of low PI clays, sands, and gravels as reflected in the site material descriptions of Sykora and Davis (1993). Damping ratios were kept below 15% for all strain levels. Site response calculations were performed by Pacific Engineering and Analysis, Inc., for ground-motion amplitudes associated with peak ground acceleration.,(on rock) of 0.05,0.10, 0.20,0.30,0.40,0.50,0.75, and 1.00 g. Figures 4 8 through 4-15 show the calculated amplification factors as a function of frequency for the various ground motion amplitudes. Figures 4-16 through 4 23 show amplification factors as a ftmetion of the corresponding rock spectral acceleration, for the frequencies considered in this study. This latter set of results will be used in Section 5 to convert the spectral accelerations on rock to spectral accelerations on soil. Following the guidance in EPRI (1993), the conversion from rock to soil motions is performed by using the median amplification factors. 4.4 VERTICAL GROUND MOTIONS ON SOIL EPRI (1993) contains recommendations for the calculation of vertical motions (including PGA), which are derived on the basis of modeling results and empirical observations (the latter use mostly data from the western United States). These recommendations specify a VM ratio that depends on frequency and on outcrop PGA (see Figure 4-25). At low frequencies, these recommendations specify a VM ratio of 0.7. At high frequencies, these recommendations specify VM ratios of 0.71.0 for low amplitudes and ratios as high as 1.5 for high amplitudes. The VM ratio of 1.5 arises because soil nonlinearity is more pronounced for horizontal than for vertical motions (i.e., the soil VM ratio of 1.5 for 10 Hz occurs when the soil / rock amplification factor is approximately 0.2 [ Figure 4-15), so that the V(soil)/H(rock) ratio is approximately 0.3). The crude shape of the VM ratios in Figure 4-25 leads to vertical spectra with unrealistic shapes. A smoother set of VM ratios is shown in Figure 4-26. These ratios are constructed so that they are consistent with the main features of the EPRI recommendations and are compatible with the horizontal soil / rock amplification factors for the Paducah site. These VM C:\\9728\\rept_2\\$ECT4.WPD November 25, 1991 4*b

ratios are calculated as the soil / rock amplification factor raised to a power of-1.29 for frequencies > 2.5 th. The soil V/H ratios in Figure 4 26, together with the soil / rock amplification factors in Figures 16 through 4 23, will be used in Section 5 to convert the horizontal spectral accelerations on - rock to venical spectral accelerations on soil.

4.5 REFERENCES

Atkinson, G.M. and D.M. Boore. (1997). "Some Comparisons Between Recent Ground. Motion Relations." SeismologicalResearch 12tters, v.68, no.1, pp. 24-30. Atkinson, G.M., and W. Silva (1997). "An empirical study of earthquake source spectra for California earthquakes." Bull. Seism. Soc. Am.,87,97-113, Electric Power Research Institute (1993). " Guidelines for determining design basis ground motions." Palo Alto, Calif: Electric Power Research Institute, vol.1-5, EPRI TR-102293. vol.1: Methodology and guidelines for estimating canhquake ground rr.otion in eastern . North America, vol. 2: Appendices for ground motion estir.ution. vol. 3: Appendices for field investigations. vol. 4: Appendices for laboratory investigations, vol. 5: Quantification of seismic source effects. Iwan, W.D. (1%7). "On a class of models for the yielding behavior of continuous and composite systems." J. Appl. Mech., 34, 612-617. Johnson, L.R., and Silva, W. (1981). "The effects of unconsolidated sediments upon the ground motion during local earthquakes." Bull. Seism. Soc. Am., 71,127 142. Risk Engineering, Inc. (l993). Seismic Hazard Evaluationfor the Paducah Gaseous Dgusion Plant, Paducah, Kentucky, K/GDP/SAR/SUB 1, September 30. Samiezade-Yazd, M., R.B. Herrmann, L. Malagnini, and W. Liu (1997). "A Regional Comparison of Vertical Ground Motion in North America", March 28. Available at web address http://www.eas.slu.edu/ People /RBHerrmann/GroundMotion/ Schnabel, P.B., Lysmer, J., and Seed, H.B. (1972). SHAKE: a Computer Program for FArthquake Response Analysis ofHorizontallylayeredSites. Earthq. Engin. Res. Center, Univ. of Calif. at Berkeley, EERC 72-12. C:\\9728\\rept_2\\ SECT 4.WPD November 25, 1997 4-9

i Schneider, J.F., W.J. Silva, and C.L. Stark (1993). Ground motion model for the 1989 M 6.9 Loma Prieta earthquake including effects of source, pt.th and site. Esthquake Spectra, 9(2),251 287. 4-Seed, H.B., and Idriss, l.M. (l970). *SollModull and Damping Factorsfor Dynamic Response Analyses," Eanhq. Eng. Res. Center, Univ. of Calif. at Berkeley, Report No. UCB/EERC-70/10. Silva, W.J., N. Abrahamson, G. Toro, C. Costantino (1997). " Description and validation of the stochastic ground motion model." Submitted to Brookhaven National Laboratory, Associated Universities, Inc. Upton, New York. Silva, W. J.; Turcotte, T.; Moriwaki, Y. (1988). " Soil Response to Earthquake Ground Motion," Electric Power Research Institute, Walnut Creek, California, Repon No. NP-5747. Silva, W.J. (1976). " Body Waves in a Layered Anelastic sollid." Bull. Seis. Soc. Am., vol. 66(5), 1539-1554. Staub, W. P. and J. C. Wang (1991). " Analysis of bore-hole seismic velocity surveys at the Paducah Gaseous Diffusion Plant." Third DOE Natural Phenomena Hazards Mitigation Conference. Sykorn, D. W. and J. J. Davis (l993). Site-Specifc Earthquake Response Analysisfor Paducah Gaseous Dipission Plant, Paducah, Kentucky. Engineers Waterways Experiment Station, Misc. Paper GL-93-14. Toro, G.R., N.A. Abrahamson and J.F. Schneider (1997). A Model of Strong Ground Motions from Eanhquakes in Central and Eastern North America: Best Estimates and Uncenainties. Seismological Research Istters, v.68, no.1, pp. 41-57. Wald, D.J., Heaton, T.H. (1994). "A multidisciplinary source analysis of the 1994 (h( 6.7) Nonhridge canhquake using strong motion, teleseismic, and geodetic data." Proc. The 89th AnnualMeeting of the SeismologicalSociety ofAmerica, Programfor Northridge Abstracts, Pasadena, California. C:\\9728\\rept_2\\ SECT 4.WPD Nevernbe r 25, 1997 4-10

1 Table 4-1 Best Estimate Soil Profile Thickness Shear-Wave Density Niaterial Description (fi) Velocity (lb/cu ft) (ft/sec) 13 600 124 Silty Clay 7 800 124 10 950 121 27 1100 121 18 1150 120 Clayey Sand 49 1250 120 Sand & Gravel 62 1400 124 Sand, Silt, Clay 68 1575 126 Clayey Silt 67 1400 165 Silty Sand 50 1400 165 C \\9728\\tept_2\\5ECT4.WPD November 25, 1997 4*il l

Toro et al. (1997) PGA Multiple Equations and Weights 1 E+01 1E+00 TM5. g g1E-01 ~ s M 7.5 1 E-02 Thick +-0.74 sigma (e) (0.454) Thin +-2.33 sigma (e)(0.046) Dashed median Squares Henmann (NM) 1 E-03 ' ' c' ' i 1E0 1E1 1E2 1E3 Horizontal Distance (km) Figure 4 1. EPRI attenuation equations for PGA on rock (point-source assumption). l l 1 l C:\\9728\\rept.,2\\ SECT 4.WPD November 2$,1997 k l2

i i Toro et al. (1997) 1-Hz Multiple Equations and Weights 1 E+01 Thick +-0.74 sigma (e) (0.454) Thin +-2.33 sigma (e) (0.046) Dashed median - 1 E+00 _en e O 8 _e 8 1 E-01 u .h ~~~~ M 5.... M 7.5 o e 1 m 1 E-02 1 E-03 1EO 1E1 1E2 1E3 Horizontal Distance (km) Figure 4-2. EPRI attenuation equations for 1-Hz spectral acceleration on rock (point-source assumptior!). C:\\9728\\rept_2\\ SECT 4.WPD November 25, 1997 4*13

i AB95 versus EPRl93 freq = 1 Hz 1-~....................,,,,,,.. ... ~........ .' ~ ~ ~...................., ' ' '.. 4 ......, ~.. '....., ' '.. 0.1 -g N. m 0.01;................... " " " " *.....,,, - .. ' "....,M=7.25 M-6.0 M=4.5 0.001 1 10 100 1000 Horizontal Distance (km) AB95 "~~ EPRl93 Figure 4 3.. Comparison of EPRI (1993) and Boore-Atkinson (1997) attenuation equations for 1-Hz spectrr3 acceleration. Source: Atkinson and Boore (P97). C:\\9129\\rept,2\\$ECT4.WPD Wovember 25, 1997 4*I4

Attenuation Equations for Faults Peak Ground Acceleration 1 E+01 no saturation ernpiricai 1E+00 -......',.',s. Modeling -a '. \\ ..' N \\y g c ' ~ j N s s g1E-01 M 7.5

[;

N x ^ oo< M 6.5 1 E-02 M S.5^ ' ' ' 1 E-03 1E0 1E1 1E2 1E3 Horizontal Distance (km) Figure 4-4. Median extended source attenuation equations for PGA. Note: the no-saturation and empirical predictions for M 5.5 are identical. C:\\9728\\rept,2\\ SECT 4.WPD Novernber 25, 1997 4-15

l Attenuation Equations for Faults 1-Hz Spectral Acceleration 1 E+01 no saturation empiricai ^ 1E+00 3 Modeling ,,,,,7 - -. _,.. .a m x. s s g - - - - - - - ~,, s, N Y 1 E-01 ' s N u N s M 7.5 g N g s M 1 E-02 M 6.5 M 5.5 1 E-03 1E0 1E1 1E2 1E3 Horizontal Distance (km) Figure 4-5. Median extended-source attenuation equations for 1-Hz spectral acceleration.. Note: the no-saturation and empirical predictions for M 5.5 are identical. C:\\9728\\rept_2\\$ECT4.WPD Wyember 25, 1997 4*l6

d j l N ~ ~ ~ g a O c~ L. v 6h S" o 0 O 8 v O. 1000. 2000.

3000, SHEAR WAVE VELOCITY (FT/SEC)

PADUCAH GENERIC PROFILE LEGEND l Figure 4-6. Bar,e-case shear-wave velocity profile for PGDP. l c \\9728\\REPT_2\\SE0"T4.WPD November 24, 1997 4* } 7 I

i l_J,;,,j,, O t4 r l I -i l e o M l '-l

e... q.,.

l s. a 8 IK = d i...... I e 6....... l. F-i L. l 1 x8 l 1 CL N Luo t ........y i 1 1 i 8 m i e t................., i O t 8, - O.

1000, 2000.

3000, i

SHEAR WAVE VELOCITY (FT/SEC) PADUCAH. 1.EGEND 511E 1 SITE 2 =*** 511E 3 511E 4 L.- 1 i Figure 4-7. All measured shear wave velocity proses for PGDP. c \\9729\\REPT,,2\\ SECT 4.WPD~ November 24, 1997 4*II i- ._,......, _,_.._...-__ ~.,.-,,

n-g C . o. T , /' 'N.,.%.s /..'..,..' +..... s u s .. ~~'~-- o t ~. ,; e ; tn T a 10 ~3 10 0 10 1 10 2 Frequency (hz) PADUCAH/MIDCONTINENT ROCK INPUT MOTION 0.05 G, MAX DAMP =15% LEGEND 04TH PERCEtCILE -+- T1EDim 16TH PERCEtCILE r1Em Figure 4-8. Paducah sit & specific soil amplification factors as a function of frequency. Input-motion PGA: 0.05g. = _. = c i \\97 2 e\\REPT.,2\\ ster 4.wPD November 24, 1997 4=I9

M I I y I I I 4 I I I 5 I I I I I I I 4 1 I I I E l g C o .a .~. e N,s .e /,... ........,,..,s. a j.- ~ x / a. -......~--s, s g - ~........ ~, - a; o ee ~, m l

• t E

to -3 10 0 10 1 10 2 Freauency (hz ) PADUCAH/MIDCONTINENT ROCK INPUT-M0 TION 0.10 G, MAX DAMP =15% LEGEND 84TH PERCEtGILE -+- 11EDIM 16TH PERCENTILE NEm i l Figure 4 9. Paducah site-specific soil-ampli6 cation factors as a function of frequency. Input-motion PGA: 0.10g. 'c \\9728\\REPT,2\\ SECT 4.WPD November 24 1997 4*20 l -,u_-.___

o ,,, crv, ., i i_ i.. ....i, o C O p

• ,e s.

C .3 /,..***..,*N.w.N

• O

/. * ' ,N'N + , _, s o- _d g_ _ _,/_ *.,,,- s,,s x. x. / $ cm s g : b' N, 'a., U) - \\ .,.. j,- r 's, ,e ~ ,o ~.* 'I R 10 -3 10 0 10 1 10 2 g Frequency (hz) PADUCRH/MIDCONTINENT ROCK INPUT MOTION 0.20 G, MAX DAMP =15% LECEND 84TH PERCEICILE l1EDIfti 16TH PERCEtGILE MErn Figure 410. Paducah site-specific soil-amplification factors as a function of frequency. Input-motion PGA: 0.20g. c \\9738\\REPT,2\\ SECT 4.WPD November 24 1997 _4-2 I l b

g C O o e w. .N p \\. .'..... *.... N'%, //'.... '~ ....-(,, N. [ ~ .,.N ./. o w ~.. N -a s, 's.j' 6 's, ~. 's, ,i / ,,_,,i i' a t0 '3 10 0 10 1 10 2 Frequency (hz)- PADUCAH/MIDCONTINENT ROCK INPUT MOTION 0.30 G, MAX DAMP =15% LEGEND 04TH PERCEtiTILE T1EDIAN 16TH PERCEtalLE liErri ( Figure 411. Paducah site-specific soil amplification factors as a function of frequency. Input-rnotion PGA: 0.30g. c e \\9728\\REPT,2\\ SECT 4.WPD Noverber 24 1997 4*22

e C O ,.. m, .2 r' N. /.. s. d a.

• x C

O / ,5 E : s\\ .\\. .M. s 'N /(. 7 m s y ./ 's ~ ~... -

• t R

10 -3 10 0 10 1 10 2 Frequency (hz) PADUCAH/MIDCONTINENT ROCK INPUT MOTION 0.40 G, MAX DAMP =15% LEGEND 84TH PERCEffilLE NEDIAN 16TH PERCEffilLE Em Figure 412. Paducah site-specific soil amplification factors as a function of frequency. Input. motion PGA: 0.40g. essmosarrr_asstere.wpo nov d.: 24, 2,,, 4 23

o -C .o o .e ,.. s .U N. ./.*.........'. ~. c. 7 , - ~,~ h - y,,a 's,' ,'y .,'N. /....c eR s, ~ N. /..- u 's w. n a' '~, t ' ~ ~..,, ' l i' R 10 ~3 10 0 10 1 10 2 Frequeneg_(hz) PADUCAH/MIDCONTINENT ROCK INPUT MOTION 0.50 G, MAX DAMP =15% LEGEtiD 84TH PERCEtCILE f1EDIAN ) 16TH PERCEtCILE ntas Figure 413. Paducah site specific soil amplification factors as a function of frequency.- Input-motion PGA: 0.50g. l ei\\o72esaret assect4.wro Nov.md.: 2., 1997 4-24 1

M I I I i 1 5 3 I I I I I 4 1 3 I I I I I I I I I 1 I g C .O ae ~. .N / \\. ,1 / N 's g---' N g s, 's.,, N o g ve ~, N. m / 2 n ~ ~, ~ ~ T e 10 -3 10 0 10 1 10 2 Frequent 3 (hz) PADUCAH/MIDCONTINENT ROCK INPUT MOTION 0.75 G, MAX DAMP =15% LEGEliD 84TH PERCEfUlLE T1ED144 161H PERCENTILE tiEg4 Figure 4-14. Paducah site-specific soil-amplification factors as a function of frequency. Input motion PGA: 0.75g. es\\9728\\REPT_2\\ SECT 4.WPD Noverrber 24, 1997 4-25

M 3 4 4 I I q v I I E 4 q g i e i 1 i a s a i I E q T o C O o eu ,. - s. C ~ N. / s. ~_. O x. o, .., e w ~ m ~, \\. /... s = ~ s s s. s. i s ~.. ~ , ~., 1 i a 10 ~3 10 0 10 1 10 2 h equency (hz ) PADUCAH/MIDCONTINENT ROCK INPUT MOTION'1.00 G, MAX DAMP =15% LEGEND 84TH PERCEH11LE - NED1m 16TH PERCENTILE l ntm Figure 4-15. Paducah site-specific soil-amplification factors as a function of frequency. Input motion PGA: 100g. i i cs\\9724\\REPT_2\\ SECT 4.WPD Noverbet 24, 1997 4-26 1 w----- o-.ww~--.- ..v~. .+-.....--.,c.,-- --.w,n .----,-wmr ,r-m-+...o,%-.-..~... - ..-,---,.m--mc~,cww.-.-%.-..w-e-,,.-,,w-,%-: si,

Amplification Factor,0.5 Hz 10.00 g Ua 1 eyz 1.00 uoc= (l1. E< 0.10 O.01 0.10 1.00 Rock PSA'(g) Figure 4-16. Amplification factors for 0.5 Hz spectral acceleration. Solid line, median; dashed' 0.15 and 0.85 fractiles. c \\972t\\ REP

• 2\\fEU4.WPD Never.ber 24 1997 4*27

Amplification Factor,1.0 Hz 10.00 i 1 i b.o j m ~~,,'s'~. eoz 1.00 co u C =o. E 0.10 O.01 0.10 1.00 Rock PSA (g) t l Figure 4-17. Amplification factors for 1.0 Hz spectral acceleration. Solid line, median; dashed, 0.15 and 0.85 fractiles. l c \\9728\\REPT_2\\ SECT 4.WPD Novemtet 24, 1997 4*28 l

Amplification Factor,2.5 Hz 10.00 ke __________~,, u s O s * ~s ti { ~ Ns c ~~ O s *s = 1.00 m s s .9 's

e:

o E e 0.10 O.01 0.10 1.00 Rock PSA (g) t Figure 4-18. Amplification factors for 2.5 Hz spectral acceleration. Solid line, median; dashed,0.15 and 0.85 fractiles. c i \\9728 \\REPT_2 \\ SECT 4.WPD November 24, 1997 4-29

Amplification Factor,5 Hz 10.00 4 9 - - - ~ .._,'s,s~s .ooa s a s ~ s _,'s 's c s s 'N \\. 0 1.00 ~ ' s. 8 s is:: s s s -a s E N 's, 4 0.10 O.10 1.00 10.00 Rock PSA (g) Figure 4-19. Amplification factors for 5 Pz spectral acceleration. Solid line, median; dashed, j 0.15 and 0.85 fractiles. a c \\9728\\REPT_2\\ SECT 4.WPD November 24, 1997 4*30

l Amplification Factor,10 Hz 10.00 9 6. ~~_' ~~,~~, L)u L3 E %'N c ~~N 's (;j N y 1.00 's s 'y s o s s s h \\, \\ s 1 N N. E< \\ s s s N N % 0.10 O.10 1.00 10.00 Rock PSA (g) Figure 4-20. Amplification factors for 10 Hz spectral acceleration. Solid line, median; dashed,0.15 and 0.85 fractiles. t c \\9728\\REPT_2\\ SECT 4.WPD November 24 1997 4-3 I 9

) Amplification Factor, 25 Hz 10.00 4 m Som ~,'~g E 8 N s 1.00 m 's .o

's's N

E 'N 's s - c2. N 's E 's 's s s 's s s s,, s s 0.10 O.10 1.00 10.00 Rock PSA (g) Figure 4-21. Amplification factors for 25 Hz spectral acceleration. Solid line, median; dashed,0.15 and 0.85 fractiles. c \\9728\\REPT_2\\ SECT 4.WPD November 24, 1997 4-32

Amplification Factor, 35 Hz 10.00 m 2o ~ '~~ s N ' s.- 1.00 s m s s o

s'~

s N .t s, 's s s N o. s's 's s E s' s N s s s' N s N 's 's 0.10 O.10 1.00 10.00 Rock PSA (g) Figure 4-22. Amplification factors for 35 Hz spectral acceleration. Solid line, median; dashed,0.15 and 0.85 fractiles. c : \\9728 \\REPT_2 \\ SECT 4.WPD November 24 1997 4-33

Amplification Factor, PGA 10.00 9 t u 2 ~~~~~ o ~ c ~~

s. ' s s

f 1.00 '\\ N o .c s,s s Q. s's E pua 0.10 O.01 0.10 1.00 Rock PSA (g) Figure 4-23. Amplification factors for peak ground acceleration. Solid line, median; dashed, 0.15 and 0.85 fractiles. c \\9726\\REPT_2\\5ECT4.WPD November 24 1997 4-34 f

1 F I Soil VerticallHorizontal Ratios 10 0.,g - -0.2g ~ - ~ 0.3 g ,Q ~ ~ = 0.4 g ~~ '8 1 "sg f._c - = Z ..... 1,0 g j;; 0.1 O.1 1 10 100 Frequency (Hz) Figure 4-25. Vertical / Horizontal (V/H) ratios for ground motions on soil. C:\\9728\\rept_2\\ SECT 4.WPD November 25, 1997 4-36

Section 5 1 RESULTS i

5.1 INTRODUCTION

This Section presents the results from the probabilistic seismic-hazard analysis at the PGDP site. PSHA results convey three types ofinformation about earthquake hazard at the site, as follows: (1) a best-estimate or central measure of the earthquake hazard at the site (typically conveyed by the mean or median hazard), (2) information about the epistemic uncertainty in the calculated hazard as a result ofincomplete information, and (3) sensitivity results that indicate the major contributors to seismic hazards and its uncertainty. 4 i Section 5.2 presents results for rock, including the de-aggregation by source and by magnitude-distance-c. Section 5-3 presents results for soil, including uniform-hazard spectra ] for damping ratios other than 5 percent and for both horizontal and vertical motior.s. Section 5-4 discusses the hazard results. Additional sensitivity results are contained in Appendix C. s 51 RESULTS FOR ROCK The seismic hazard calculations consider multiple alternative interpretations on maximum 1 magnitudes, activity rates, and attenuation equations, using a logic-tree approach described

l carlier. Each combination of assumptions and parameters produces a hazard curve (i.e., a j

curve of annual exceedence probability versus ground-motion amplitude). The result of the logic-tree analysis is a family of rock-site hazard curves, with associated weights. The spread among these curves represents uncertainty in the seismic hazard, which is the result of competing hypotheses and limited data. The most important characteristics of this family of hazard curves are represented by summary hazard curves. Typically, the mean hazard curve (sometimes the median curve') represents the central tendency of the lazard. The spread 3 Arguments based on classical decision theory suggest the use the mean hazard. The median hazard has been used in wme situations because it is a more stable quantity. C:\\9128\\rept,2\\SectSfwpd November 25, 1997 5-1

between the 0.15-and 0.85-fractile hazard curves represents epistemic uncertainy in seismic hazard. Figures 5-1 through 5-8 show t. summary hazard curves for peak acceleration and for spectral acceleration at 0.5,1.0. 2.5, 5,10,25, and 35 Hz (5% damping). Figures 5-9 and 5-10 present the median and mean uniform-hazard spectra obtained from these hazard curves, for annual exceedence probabilities of 4x 10-3,2x10-3,1 x 10-3, and 2x10 (corresponding to 4 average retum periods of 250,500,1,000, and 5,000 years). The shape of these spectra are 9 typical of eastern-U.S. ground motions, with significant energy at frequencies in excess of 10 Hz. In order to provide better definition of the spectral shape at high frequewies, the spectral acceleration on rock at 55 Hz was calculated as 1.17xPGA. The value of 1.17 was derived from a typical spectral shape for ground motions on hard rock in the central and eastern United States (CEUS). Figures 5-11 and 5-12 show the contributions of tie various seismic sources to the mean total hazard, for PGA and for 1-Hz spectral acceleration. At return periods of 250 years or longer, more than half the hazard comes from the East Prairie extension and the East prairie fault. Next in importance are other NMSZ faults (Blytheville arch-CGF, Reelfoot fauh). The Wabash Valley seismic zone has a small contribution to the hazard. Figures 5-13 through 5-16 show the contributions of various magnitude, distance, and e combinations to the exceedence probability. Separate figures are shown for PGA and for I-2 Hz PSA, and for two exceedence probabilities. For the 250-yr return period and PGA, small and moderate earthquakes on the EPE contribute approximately 50% of the hazard Large earthquakes in EPE, EP, and the other NMSZ faults contribute the remaining ~50%. For the 250-yr return period and 1-Hz PSA, large earthquakes dominate seismic hazard. Earthquakes on the Reelfoot fault and Blytheville Arch-CGF make a signi6 cant part of that contribution. For the 5,000-year return period, nearly all hazard comes from large events on EPE and EP. 2The contributions shown for 90-105 km represent distances greater than 90 km. Similarly, the contributions shown for magnitudes 7.5-8.0 represent magnitudes greater than 7.5. C:\\9728\\rept,2\\ Sect 5_.wpd Noverber 25, 1997 5-2

5.3 RESULTS FOR SOIL Multiplication of the rock uniform-hazard spectra in Figures 5-9 and 5-10 by the frequency-and amplitude-dependent site-specific ampli6 cation factors yields the uniform-hazard spectra for ground motions at the PGDP ground surface. Results are shown in Figures 5-17 and 5-18. Results for vertical motions are obtained by applying the amplitude-and frequency-dependent V(soil)/H(soil) factors obtained in Section 4. Results are shown in Figures 5-19 and 5-20. All spectral accelerations and uniform-hazard spectra computed above correspond to a damping ratio of 5%. Response spectra for other damping ratios are computed from the corresponding 5% PSA sad PGA by means of the following procedure. For frequency 1< f < 5 Hz, we apply the procedure by Rcsenbueth (1980); i.e.,

• d' 1+4.9Qf T PM(f,p)=PM(f,0.05) 1 +4.9x0.05xfT, (5-1) where p indicates damping ratio (as a fraction of critical, not as a percentage) and T represents the duration (sec). The duration is taken as 18 sec, based on the magnitude and distance of the dominant event for the 250-year return period. (Sensitivity to duration is fairly low.)

For higher frequencies, the procedure is modified to provide a transition to the PGA-controlled portion of the spectrum. The rewlting equation takes the form: M PMgp)= PGA p y[7,p)2_pgj2 1 +4.9pfT -a4i 2 2 (5-2) 1 +4.9x0.05 xf T, Some of the calculated spectra exhibited slight roughness, as a result of-the approximate nature of Equations 5-1 and 5-2 and of slight inconsistencies among the attenuation equations, the site amplification factors, the V/H ratios. The most common of these problems was a C:\\9728\\rept_2\\ Sect 5_,.wpd November 25, 1997 5-3 ~

+ l PSA (SSIIz) slightly higher than PSA (35Hz); the former had to be reduced by less than 15%. The resulting uniform-hazard spectra, for both horizontal and vertical components, are shown in Figures 5-21 through 5-32. 5.4 RESULTS IN TABULAR FORM kND OTHER RESULTS Tables 5-1 through 5-9 contain the values, in tabular form, for all uniform-hazard spectra shown in this Section. In addition, we calculate the mean probability of exceeding a horizontal PGA of 0.15 g on soil, obtaining a value of 4.5x10'8 (corresponding to an average return period of 220 years). 5.5 DISCUSSION The results presented here reflect the current state of knowledge on the recurrence-frequency and characteristics of earthquakes in the New Madrid Seisnec Zone (NMSZ). Though this state of knowledge has advanced significantly over the past ten years, many questions remain. In terms of seismic hazard at the PGDP, the northern portion of the NMSZ (East Prairie fault and East Prairie extension) are the most important. This northern portion is, unfortunately, the most poorly understood. There are several assumptions in this study that may be considered conservative. The first of these is the choice of attenuation equations. The EPRI (1993) attenuation equations used in this study (which assume a Brune source spectrum) predict significantly higher ground motions at frequencies near 1 Hz than the two-corner model of Atkinson and Boore (1997), particularly for large magnitudes (see Figures 4-2 and 4-3). There is significant evidence supporting the two-corner model(Atkinson,1993; Atkinson and Chen,1997; Atkinson and Silva,1997), but the issue of the spectral shape for CEUS earthquake remains unresolved. In spite of the evidence cited above, there has been some reluctance to assign significant weight to these equations (in large part because they predict much lower amplitudes near 1 Hz than other models). If one sets aside considerations of conservatism and consistency with recent practice, one should assign some weight to the Atkinson-Boore attenuation equations. C:\\9728\\rept_2\\ Sect 5,.wpd November 25, 1997 5-4

i Another source of conservatism relates to the maximum-magnitude assignments for the NMSZ faults.. This study uses Johnston's (1996) estimates (see Appendix A), which are somewhat higlier than Nuttli's (1973) estimates. Both Johnston's and Nuttli's estimates ;.re based largely I on the intensity observations from the 1811-1812 events and involve significant extrapolation cf the data on the relationship between magnitude and intensity (or isoseismal area). Although j the estimates used here are the most authoritative ones given the current state of knowledge, j they are not universally accepted. There is the perception among some seismologists and - 1 geologists that these magnitudes are too high given the California experience and the tectonic setting of the NMSZ Lower values of maximum magnitude would reduce the hazard i i estimates obtained here, especially for long reurn periods. i l_ A third source of conservatism relates to the rate of occurrence oflarge earthquakes in the l NMSZ. There are two elements to this issue. The first one relates to whether alllarge events in the paleoseismic record are comparable in size to the 1811-1812 events. The second one l relates to whether a large event in one of the NMSZ segments always triggers events in the l other two segments, as happened in 1811-1812. The effect of the this clustering assumption - is discussed in Section 2.2. The number and spatial distribution of paleoliquefaction sites t l investigated to date are not sufficient to resolve these questions, leaving a wide margin for scientific interpretation. This study used assumptions that some scientists may consider j conservative, particularly on the second issue. l l - Most of these potential sources of conservatism have a larger effect for long return periods. Thus, while we consider that the results presented here are appropriate and well supported for i quantifying ground motions for return periods of 500 years or less, they may be somewhat i { conservative for longer return periods. Given the current state of knowledge about the NMSZ, the evaluation of seismic hazard for longer retum periods should include a broader l characterization of uncertainty than was possible given the resources and scope of this project. Ideally, this effort should include multiple source-characterization experts (and possibly multiple ground-motion experts as well). 4 i C \\9728\\rept_2\\ Sect 5,.wpd November 25, 1997 55 l i

5.6 REFERENCES

Atkinson, G.M. (1993). " Source spectra for earthquakes in eastern North America." Bull. Seism. Soc. Am., v. 83, pp.1,778-1,798. Atkinson, G.M. and D. M. Boore. (1997). "",ome Comparisons Between Recent Ground-Motion Relations." Seismological Researci,I2tters, v.68, no.1, pp. 24-30. Atkinson, G.M., and W. Silva (1997). "An empirical study of earthquake source spectra for California earthquakes." Bull Seism. Soc. Am.,87,97-113. Electric Power Research Institute (1993). " Guidelines for determining design basis ground motions." Palo Alto, Calif: Electric Power Research Institute, vol.1-5, EPPl TR-102293. vol.1: Methodology and guidelines for estimating earthquake ground motion in eastern North America. vol. 2: Appendices for ground motion estimation. vol. 3: Appendices for field investigations. vol. 4: Appendices for laboratory investigations. vol.5: Quantification of seismic source effects. Johnston, A.C.,1996. Seismic moment assessment of earthquakes in stable continental regions -III. New Madrid 1811-1812, Charleston 1886 and Lisbon 1755. Geophys. Jour. International, 126:314-344. Rosenblueth, E. (1980). " Characteristics of Earthquakes," chapter 1 in Rosenblueth, ed., Design ofEsthquake Resistant Structures, Halsted Press, Wiley, New York. Nuttli, O. W. (N73). "The Mississippi Valley Earthquakes of 1811 and 1812: intensities, grour 1 motion, and magntudes." Bull. Seism. Soc. Am., v. 63, pp. 227-248. C:\\9728\\rept_2\\ Sects _.wpd November 25, 1997 5-6

Table 5-1 Uniform-Hazard Spectra for Rock,5% damping Spectral Accelerations (g) Median Annual Exceedence Prob. freq(Hz) 4E-03 2E-03 1E-03 2E-04 0.5 0.026 0.060 0.137 0.363 1.0 0.056 0.113 0.229 0.613 2.5 0.116 0.226 0.418 1.070 5.0 0.168 0.326 0.574 1.430 10 0.199 0.403 0.707 1.710 25 0.215 0.464 0.841 2.240 35 0.199 0.433 0.794 2.210 55 0.109 0.233 0.410 1.030 100 (=PGA) 0.093 0.199 0.350 0.880 Mean Annual Exceedence Prob. freq(Hz) 4E-03 2E-03 IE-03 2E-04 0.5 0.026 0.060 0.143 0.494 1.0 0.055 0.115 0.238 0.801 2.5 0.117 0.226 0.437 1.340-5.0 0.174 0.335 0.625 1.930 10 0.204 0.403 0.752 2.280 25 0.224 0.471 0.899 2.800 35 0.211 0.450 0.862 2.740 55 0.114 0.231 0.443 1.350 100 (=PGA) 0.098 0.198 0.378 1.150 c \\9728\\rept_2\\ SECT 5.WPD November 24, 1997 5-7

Table 5-2 Uniform-Harard Spectra for Soil,5% damping Spectral Accelerations (g) Median Annual Exceedence Prob. freq(Hz) 4E-03 2E-03 IE-03 2E-04 0.5 0.051 0.126 0.293 0.774 1.0 0.160 0.309 0.574 1.170 2.5 0.282 0.506 0.730 1.120 5.0 0.374 0.589 0.757 0.861 10 0.356 0.541 0.622 0.685 25 0.233 0.335 0.406 0.612 35 0.201 0.298 0.377 0.559 55 0.172 0.285 0.3 % 0.550 100 (=PGA) 0.159 0.264 0.a.52 0.537 Mean Annual Exceedence Prob. freq(Hz) 4E-03 2E-03 IE-03 2E-04 0.5 0.051 0.126 0.305 1.050 1.0 0.158 0.314 0.591 1.530 2.5 0.283 0.506 0.750 1.410 5.0 0.384 0.599 0.778 1.080 10 0.362 0.541 0.619 0.913 25 0.239 0.337 0.410 0.689 35 0.208 0.304 0.387 0.674 55 0.178 0.284 0.380 0.654 100(=PGA) 0.165 0.263 0.365 0.652 c \\9728\\rept_2\\ SECT 5.WPD November 24, 1997 5-8

I 1 Table 5 3 Uniform-Hazard Spectra for Soil, Vertical,5% damping l Spectral Accelerations (g) Median Annual Exceedence Prob. freq(Hz) 4E-03 2E-03 IE-03 2E-04 2 0.5 0.036 0.0P8 0.204 0.541 1.0 0.112 0.217 0.402 0.821 2.5 0.218 0.401 0.621 1.100 5.0 0.2 % 0.495 0.700 0.997 10 0.301 0.495 0.644 0.852 25 0.228 0.368 0.501 0.770 35 0.200 0.332 0.468 0.741 55 0.151 0.270 0.395 0.681 100 (=PGA) 0.136 0.243 0.352 0.620 4 Mean AnnualExceedenceProb. freq(Hz) 4E-03 2E-03 IE-03 2E-04 0.5 0.036 0.088 0.213 0.737 1.0 0.111 0.220 0.414 1.070 2.5 0.219 0.401 0 641 1.380 5.0 0.304 0.505 0.730 1.280 10 0.307 0.495 0.654 0.995 25 0.234 0.371 0.515 0.930 35 0.209 0.341 0.488 0.877 i 55 0.157 0.268 0.414 0.807 100 (=PGA) 0.1.41 0.242 0.369 0.769 c \\9728\\rept_2\\ SECT 5.WPD November 24, 1991 5-9

Table 5-4 Uniform Hazard Spectra for Soit,2% damping Spectral Accelerations (g) Median Annual Exceedence Prob. ) freq(Hz) 4E-03 2E-03 IE-03 2E-04 0.5 0.064 0.157 0.364 0.%2 1.0 0.211 0.407 0."55 1.540 2.5 0.391 0.702 1.010 1.560 5.0 0.531 0.836 1.070 1.220 10 0.485 0.727 0.816 0.950 25 0.294 0.399 0.457 0.600 35 0.239 0.331 0.403 0.582 55 0.186 0.300 0.380 0.560 100(=PGA) 0.159 0.264 0.352 0.537 Mean Annual Exceedence Prob. freq(Hz) 4E-03 2E-03 IE-03 2E-04 0.5 0.064 0.156 0.379 1.310 1.0 0.208 0.413 0.779 2,020 2.5 0.392 0.702 1.040 1.950 5.0 0.545 0.850 1.100 1.530 10 0.492 0.729 0.805 1.130 25 0.300 0.403 0.454 0.728 35 0.248 0.344 0.409 0.698 55 0.193 0.307 0.370 0.680 100 (=PGA) 0.165 0.263 0.365 0.652 c \\9728\\rept_2\\ SECT 5.WPD November 24, 199, 5-10 { l

=. . - _ - ~. i Table 5 5 Uniform-Hazard Spectra for Soil,7% damping 4 Spectral Accelerations (g) Median Annual Exceedence Prob. freq(Hz) 4E-03 2E-03 IE-03 2E-04 0.5 0.046 0.114 0.265 0.700 1.0 0.143 0.276 0.511 1.040 2.5 0.248 0.445 0.643 0.987 5.0 0.328 0.516 0.663 0.754 10 0.321 0.489 0.569 0.653 25 0.218 0.319 0.394 0.595 i 35 0.192 0.290 0.372 0.554 55 0.180 0.280 0.360 0.550 -100 (=PGA) 0.159 0.264 0.352 0.537 Mean Annual Exceedence Prob. freq(Hz) 4E-03 2E-03 IE-03 2E-04 0.5 0.046 0.114 0.276 0.953 1.0 0.141 0.280 0.527 1.360 2.5 0.249 0.445 0.660 1.240 5.0 0.336 0.524 0.681 0.943 10 0.326 0.490 0.569 0.858 25 0.223 0.321 0.400 0.680 35 0.199 0.295 0.382 0.669 55 0.180 0.280 0.370 0.660 100 (=PGA) 0.165 0.263 0.365 0.652 s A c \\9728\\rept_2\\ SECT 5.WPD November 24, 1997 5-11 i

F' Table 5-6 Uniform-Hazard Spectra for Soil,10% damping Spectral Accelerations (g) Median Annual Exceedence Prob. freq(Hz) 4E-03 2E-03 IE-03 2E-04 0.5 0.041 0.102 0.236 0.624 1.0 0.126 0.242 0.449 0.916 2.5 0.216 0.387 0.559 0.859 5.0 0.284 0.448 0.575 0.730 10 0.289 0.444 0.524 0.626 25 0.205 0.306 0.384 0.581 35 0.184 0.284 0.367 0.550 55 0.170 0.270 0.360 0.540 100(=PGA) 0.159 0.264 0.352 0.537 Mean Annual Exceedence Prob. freq(Hz) 4E-03 2E-03 lE-03 2E-04 0.5 0.041 0.101 0.246 0.849 1.0 0.124 0.246 0.463 1.200 2.5 0.217 0.388 0.575 1.080 5.0 0.292 0.455 0.591 0.940 10 0.294 0.444 0.525 0.812 25 0.210 0.307 0.391 0.673 35 0.191 0.287 0.378 0.665 55 0.180 0.280 0.370 0.660 100(=PGA) 0.165 0.263 0.365 0.652 c \\9728\\rept_2\\ SECT 5.WPD November 24, 1991 5-12

Table 5-7 Uniform-Hazard Spectra for Soil, Vertical,2% damping Spectral Accelerations (g) Median Annual Exceedence Prob. freq(Hz) 4E-03 2E-03 IE-03 2E-04 0.5 0.045 0.110 0.254 0.673 1.0 0.148 0.285 0.530 1.080 2.5 0.302 0.556 0.861 1.530 5.0 0.420 0.703 0.993 1.410 10 0.409 0.665 0.851 1.200 25 0.298 0.469 0.625 0.971 35 0.253 0.408 0.569 0.855 l 55 0.159 0.284 0.412 0.680 100 (=PGA) 0.136 0.243 0.352 0.620 Mean Annual Exceedence Prob. freq(Hz) 4E-03 2E-03 1E-03 2E-04 0.5 0.045 0.109 0.265 0.916 1.0 0.146 0.490 0.546 1.410 2.5 0.304 0.557 0.889 1.920 5.0 0.432 0.716 1.040 1.810 10 r 116 0.666 0.859 1.450 25 0.305 0.474 0.637 1.080 35 0.264 0.424 0.592 0.982 55 0.165 0.283 0.432 0.860 100 (=PGA) 0.141 0.242 0.369 0.769 j c:\\9728\\rept_2\\ SECTS.WPD Wovember 24, 1997 5-13

l l Table 5-8 Uniform-Hazan! Spectra for Soil, Vertical,7% damping Spectral Accelerations (g) Median Annual Exceedence Prob. freq(Hz) 4E-03 2E-03 IE-03 2E-04 0.5 0.032 0.080 0.185 0.490 1.0 0.100 0.193 0.358 0.731 2.5 0.192 0.353 0.546 0.%8 5.0 0.259 0.434 0.613 0.873 10 0.271 0.448 0.588 0.803 25 0.210 0.343 0.470 0.740 35 0.187 0.313 0.443 0.714 55 0.159 0.284 0.412 0.670 100 (=PGA) 0.136 0.243 0.352 0.620 Mean Annual Exceedence Prob. freq(Hz) 4E-03 2E-03 IE-03 2E-04 0.5 0.032 0.079 0.193 0.667 1.0 0.099 0.1 % 0.369 = 0.957 2.5 0.193 0.353 0.564 1.220 5.0 0.267 0.442 0.639 1.120 10 0.277 0.448 0.599 0.947 25 0.216 0.345 0.484 0.894 35 0.195 0.320 0.462 0.852 55 0.165 0.283 0.432 0.820 100 (=PGA) 0.141 0.242 0.369 0.769 c \\9128\\rept_2\\5ECT5.WPD November 24, 1997 5-14 l

Table 5-9 Uniform Hazarti Spectra for Soit, Vertical,10% damping Spectral Accelerations (g) Median Annual Exceedence Prob. freq(Hz) 4E-03 2E-03 IE-03 2E-04 0.5 0.029 0.071 0.165 0.437 1.0 0.088 0.170 0.315 0.643 2.5 0.167 0.307 0.475 0.843 5.0 0.225 0.376 0.531 0.757 10 0.244 .406 0.539 0.761 0 25 0.194 0.320 0.443 0.732 35 0.175 0.297 0.422 0.691 55 0.159 0.270 0.390 0.660 100(=PGA) 0.136 0.243 0.352 0.620 Mean Annual Exceedence Prob. freq(Hz) 4E-03 2E-03 1E-03 2E-04 0.5 0.029 0.071 0.172 0.594 1.0 0.087 0.172 0.324 0.841 2.5 0.168 0.307 0.491 1.060 5.0 0.231 0.383 0.554 0.970 10 0.250 0.407 0.550 0.%5 25 0.200 0.322 0.458 0.864 35 0.183 0.302 0.441 0.832 55 0.165 0.283 0.410 0.800 100(=PGA) 0.141 0.242 0.369 0.769 c \\9728\\rept_2\\ SECT 5.WPD November 24, 1997 5-15

r I Total Hazard, PGA (Horizontal) 1E-01 i i i i i g5th...... : Mean Median --- - 15th 1E42 o e.*.* o + c o e O o u 1E

• g 4

+ g c c c %g% %g, %..,, ~~ 1E-04 1E-05 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Peak Ground Acceleration (g) Figure 1. Summary hazard curves for PGA on rock. I c \\9728\\rept,2\\ SECT 5.WPD November 24, 1997 5-16 l

d i Total Hazard, 0.5-Hz PSA (Horizontal) i 1E-01 i ^ 85th Mean t Median --- - 15th 1E-02 s [' i 7 U c 1 O 2 o O d IE-03 4 w A g s

s g

c s c s s s s s s s ......**. N s + 's 1E-04 s s s~,% + s,~, ~,g~, ..e j s 1 1E-05 0 0.2 0.4 0.6 0.8 1 Spectral Acceleration (g) Figure 2. Summary hazard curves for 0.5-Hz spectral acceleration on rock (5% damping ratio). 3 c:\\9728\\rept_2\\5ECT5.WPD November 24, 1997 5-17

Total Hazard,1-Hz PSA (Horizontal) 1E-01 i i i 85th Mean Median --- 15th s @N

4., '.

7'. - o, 49 v ~- n " " ~ " " ~ ~ &Y ~ 1E-05 0 0.2 0.4 0.6 0.8 1 Spectral Acceleration (g) Figure 3. Summary hazard curves for 1-Hz spectral acceleration on rock (5% damping ratio). c:\\9728\\rept,2\\ SECT 5.WPD November 24, 1997 5-18 D

Total Hazard,2.5-Hz PSA (Horizontal) 1E-01 85th Mean ,y.,..s* 4 A 1

v.,.

&g,

• ., ~.,

f g o 46, 4 \\ + \\ .....,,,,, ~.* ~ ' " ".. 1E-05 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Spectral Acceleration (g) Figure 4. Summary hazard curves for 2.5-Hz spectral acceleration on rock (5% damping ratio). ct\\9728\\rept_2\\ SECT 5.WPD November 24, 1997 5-19

Total Hazard,5-Hz PSA (Horizontal) 1E-01 i i i i i i i i 85th """ : Mean [ Median --- - 15th &g

• {

S g$ a V + ~ &Y 1E-05 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 Spectral Acceleration (g) Figure 5. Summary hazard curves for 5-Hz spectral acceleration on rock (5% damping ratio). I ct\\9728\\rept,2\\ SECT 5.WPD November 24, 1997 5-20 1

1 4 Total Hazard,10-Hz PS A (Horizontal) 1E-01 85th Mean Median --- - 15th N,,., + Q p :-.;., f o 4 s 9 s D o 0 + h h ~~, + 4 s hg

• a 1

s I 4 1E-05 0 0.5 1 1.5 2 2.5 Spectral Acceleration (g) Figure 6. Sumrnary hazard curves for 10-Hz spectral acceleration on rock (5% damping ratio). i c:\\9128\\rept_2\\ SECT 5.WPD November 24, 1997 5-21 l

Total Hazard,25-Hz PSA (Horizontal) 1E-01 i i i i i 85th """ : Mean Median --- - 15th """ - \\

1. g), ;,

7.,. e \\@^ " ~ ~ ~.."." " "*" "... c, + 4.& ' 1E-05 0 0.5 1 1.5 2 2.5 3 Spectral Acceleration (g) Figure 7. Sumnwy hazard curves for 25-Hz spectrai acceleration on rock (5% damping ratio). 5/2 l et\\l12t\\tept,2\\ stet 5.wPD Novembe r 24, 1997 2

Total Hazard,35-Hz PSA (Horizontal) lE-01 i i i i i P 85th Mean Med!an --- - 15th ' + 46,.., \\ 4-@. x x .a " "....... 4.# ' 1E-05 0 0.5 1 1.5 2 2.5 3 Spectral Acceleration (g) Figure 8. Summary hazard curves for 33.Hz spectral acceleration on rock (5% damping ratio). I ci\\9728\\rept,2\\ SECTS.WPD November 24, 1997 5 23 __J

Rock, 5% damp. median spectra 1 10 l .l nU v co 100 s .2.0E-04 e u v e 1.0E-03 u o< ~ 2.0E-C3 -ew 10-1 a -4.0E-03 o e Q,, Cr) 1010-2 10" 101 10e Frequency (Hz) Figure 9. Median uniform hazard spectra for rock ($% damping ratio). c \\9720\\rept_2\\$ECT$.WPD Novembe r 24, 1997 5 24

Rock 5% damp. mean spectra 1 10 -M v E 10' 2.0E-04

3e w

c T 1.0E-03 o u< 2.0E-03 -e .b 10-1 .4.0E-03 o o a en 10~'10-1 10' 108 10' Frequency (Hz) l Figure 10. Mean uniform hazard spectra for rock (5% damping). c \\9128\\rept.,,2\\ SECTS.WPD November 24, 1997 5-25

Source Contributions to Mean Hazard PGA (Horizontal) 1E-01 ~ E. Prairie Ext. E. Prairie fit -- Reelfoot fit - - - Blytheville Arch -CGF " " Wabash - - Rough Creek graben - - I Reelfoot rift (excl. fits.) -- -- IE-02 r W. Background --- j E. Background ---- l "o' \\ E 1 o 3 p$ 8 IE-03 h(', o o e ,t. 'v

g

,e a E

),8.

s j 1,',1 's s 1 1 '.'t, N 1 i 's) s's s d IE4M N ..{ ( g%s~, v t. s, 3 I \\ 1\\ s\\ 3 \\ s, % \\ 3 s. 3 s \\ 's s i '\\ IE-05 0 0.2 0.4 0.6 0.8 1 1.2 1.4 i Peak Ground Acceleration (g) Figure 11. Seismic-source contributions to mean hazard: PGA on rock, i l l ct\\972B\\rept,2\\ SECTS.WPD Novender 24, 1997 5-26

Source Contributions to Mean Hazard - 1 Hz (Horizontal) IE-01 E. Prairie Ext. E. Prairie fit --- Reelfoot fit - - - Blytheville Arch -CGF t Wabash - - I Rough Creek graben - - j Reelfoot rift (excl. fits.) -- -- IE-02 ! W. Background --- E. Background ---- : \\ 8 8 o 8 u s 5 's's y lE-03 m i B >' s w s i s's E

\\,,,'

s~'s a i,t .','..\\. \\ s

'.. s.

N s s, s 1E-04 -; '. '. \\..\\ \\ s,, ' '.

', ',\\'s N

i ', 's\\,'\\ \\ sd '\\*s, 's g s s 's \\ s t ' *s %g% N, 's ,N,- lE-05 0 0.2 0.4 0.6 0.8 1 Spectral Acceleration (g) Figure 12. Seismic-source contributions to mean hazard: 1-Hz spectral acceleration on rock (5% damping). c \\972B\\rept_2\\ SECT 5.WPD November 24. 1997 5 27

I Contributions to Hazard,250-yr PGA E e: 2+ h E e: l to 2 $ N-D1e: O to 1 E e: -1 to 0 E a' E e: -2 to -1 c:W. 8Do%- d W- [ =& rs a;, ~ 7q,3 Figure 13. Magnitude-distance-c de aggregation of seismic hazard for PGA: 250 year return period. c \\9128\\rept.,2\\$ECT5.WPD November 24, 1997 5 28

Contributions to Hazard,250-yr PSA(1Hz) E e: 2+ h M e: Ito 2

• N-0 e: O to 1 0'

M e: -I to 0 0 B e: -2 to -1 C D-o q,. R m-DQ

• P x

$d ~ is D 60 Qy rg,. # Figure 14. Magnitude-distance-e de-aggregation of seismic hazard for 1-Hz PSA: 250-year 3 return period. c \\C728\\rept, %.0?$.WPD Wovember 24, 1991 5-29

1 Contributions to Hazard,1000-yr PGA E e: 2+ h E e: Ito 2 @ N-Ei e: O to 1 E e: -1 to 0 E' 3 E e: -2 to -1 C S- &+s j y e g*"&qp g g +, C %- gM gM.* 8 4 y Mg 5 s@ ro Figure 15. Magnitude-distance-c de-aggregation of seismic hazard for PGA: 1,000-year return period. { ct\\9728\\rept_2\\stCTS.tfPD Wovember 24, 1997 5-30

i Contributions to Hazard,1000-yr PSA(1Hz) E e: 2+ D R e: I to 2

• N-El e: O to 1 0'

B e: -I to 0 9 E e: -2 to -1 c: n. a t e 8 *' 4 o e, is #% ~ eu. @ r Figure 16. Magnitude-distance-c de-aggregation of seismic hazard for 1-Hz PSA: 1,000-year return period. l es\\9728\\rept,2\\ SECT 5.WPD November 24 1997 5 31

i i i Soil, 5% damp. median spec'ct. 101 9 v c .g 10' / a 2 ~

2.0E-04

] 1.0E-03 8 2.0E-03 4.0E-03 _e .b 10-1 o e Q. in 10-8 10-8 100 10 10' Frequency (Hz) Figure 17. Median uniform hazard spectra for soil (5% damping). c \\9729\\rept_2\\ SECTS.WPD November 24, 1997 5-32

Soil, 5% damp. mean spect.ra 101 Q v C /__. ..o 10o 1$ !2.0E-04 LWg ~ 1.0E-03 8 7 ,2.0E-03 f 4.0E-03 _e .b10-1 oW Q. in e 10-a ...~i ... i 10-1 10 102 0 10: Frequency (Hz) Figure 18. Mean uniform hazard spectra for soil (5% damping). c \\9728\\rept_2\\ SECT 5.WPD November 24, 1991 5-33

Soil, V 5% damp. median spectra 1 10 g v c: 0 ,o 10 Y ^ N2.0E-04 ' 1.0E-03 u 2.0E-03 4.0E-03 .b 10-2 u W Q. t/J 10-a .....i ....mi 10-2 10' 10 108 Frequency (Hz) Figure 19. Median vertical uniform-hazard spectra for soil (5% damplag). I c \\9728\\rept,2\\st. CTS.WPD November 24, 1997 5-34 i

Soil, V 5% damp. mean spectra lot v 4 c 8 o 10 3 2.0E-04 e L.y ' 1.0E-03 8 N ~2.0E-03 -e 4.0E-03 .b 10-8 ov-04 v2 10-8 10-1 108 101 10' Frequency (Hz) Figure 20. Mean vertical uniform-hazard spectra for soil (5% damping). c \\9728\\rept,2\\ SECTS.WPD November 24, 1991 5-35

l Soll, 2% damp. median spectra 1 10 ~ Q v c / 0 o 10 B 8 ~

2.0E-04 v

'E ' 1.0E-03 8 '2.0E-03 2 4.0E-03 m310-8 o U CL. y}. 10-a ....i ......i 10-8 10' 10 10: Frequency (Hz) Figure 21. Median uniform-hazard spectra for soil (2% damping). ct\\9720\\rept_2\\ SECTS.WPD November 24, 1997 5-36

Soil, 2% damp. mean spectra 101 n / c: 0 .o 10 1 ~ e I2.0E-04 uW;; ' 1.0E u -2.0E-03 4.0E-03 _e .b10-o W Che v3 10-a ..i ...i 10-1 10 10 108 8 Frequency (Hz) Figure 22. Mean uniform-hazard spectra for soil (2% damping). r et\\972t\\rept 2\\stcT5.WPD Wovember 24, 1997 5-37 _____________J

4 Soil, 7% damp. median spectra 101 ^ te v c:o 10 3 / \\ 0

2.0E-04

~ v Tu ' 1.0E-03 22.0E-03 8< 4.0E-03 ,_,e .5 10-u v Q tn 10-8 ' ' ' ' e 0 10-! 10 10 10 Frequency (Hz) Figure 23. Median uniform hazard spectra for soil (7% damping). ) ct\\9720\\rept_2\\$ECT5.WPD November 24, 1997 5-38

Soll, 77. damp. mean spectra 101 To v / 0 10 e

2.0E-04 6

1.0E-03 ,8 2.0E-03 4.0E-03 -e .b 10-1 o U t/J 10-a ..i 10-1 10 102 0 10 Frequency (Hz) Figure 24. Mean uniform-hazard spectra for soil (7% damping). c \\9120\\rept_2\\St. CTS.WPD November 24, 1991 5-39

Soil,10% damp. median spectra 101 g v c 8 o 10 j ~ / '2.0E-04 ~ u m '.1.0E-03 2.0E-03 e .b 10-O o Q. en 10-e ...mi ......i 10-1 10 10 10 0 1 8 Frequency (Hz) Figure 25. Medir aniform-hazard spectra for soil (10% damping). c \\9728\\rept_2\\ SECTS.WPD November 24. 1797 5-40

Section 6

SUMMARY

AND CONCLUSIONS The probabilistic seismic hazard results obtained here reflect the current state of knowledge on the recurrence-frequency and characteristics of earthquakes in the New Madrid Seismic Zone (NMSZ) and the Webash Valley Seismic Zone. These results also include site-specific information on soil conditions and amplification factors based on that information. The major contributor to seismic hazard at the Paducah site is the northern portion of the NMSZ, which is characterized by means of the East Prairie Fault (the fault where the January 23,1812 earthquake is believed to have occurred) and by the East Prairie Extension (which represents the seismicity that extends from the Reelfoot fault to the Ohio river The Wabash -Valley seismic zone makes a negligible contribution to seismic hazard at Paducah because ofits distance to the site. Although the state of knowledge about the New Madrid Seismic Zone has advanced significantly over the past ten years, many questions remain. The same is true for ground-motion models and attenuation equatiens for the central and eastem United States. This study characterizes the effect of this epistemic uncertainty by specifying multiple alternative assumptions, propagating uncertainty through the calculations, and representing seismic hazard in tenns of mean, median, and fractile curves. The results for longer retum periods are very sensitive to assumptions about attenuation i equations and about the rate, maximum magnitude, and clustering oflarge events in the NMSZ Because these issues involve large uncertainties, the results obtained here for return period of 1,000 years or more should be usM with caution. If seismic-hazard estimates for those retum periods are required, it may be advisable to perform a study with multiple experts, corresponding to Levels 3 or 5 as defined by tne Senior Seismic Hazard Analysis Committee (SSHAC,1995). C \\9728\\rept.,2\\ SECT 6.WPD November 25, 1991 6-1

An additional note of caution is necessary for the amplification factors used to compute the 5,000-year spectra on soil. The 5,000-year motions produce very high strains in the soil and are near the limits of applicability of the equivalent linear methods used in the calculation of site response. A better approach for site-response calculations at these strain levels would be to use a nonlinear soil model to represent the softer naterials near the surface.

6.1 REFERENCES

Senior seismic Hazard Analysis Committee (1995). Recommem.'ationsfor Probabl//stic Selmic Hazard Analysis: Guidance on Uncertainty med Use ofExperts. Lawrence Livermore National Laboratory, UCRL ID-122160. 1 l ] C:\\g728\\rept 2\\ SECT 6.WPD November 2$,1997 6-2

n i T Appendix A Geological and Seismological Setting of the Ncw Madrid Seismic Zone and the Wabash Valley Seismic Zone by Roy Van Arsdale and Arch Johnston ? l

The purpose of this report is to assess the seismic hazards of the Paducah, Kentucky area (Fig. A 1). We divide the report into two parts. In Part 1 (RVA) the active faults of the study region are identified and their known properties discussed. In Part 2 (ACJ), the seismic potential of these faults is assessed, along with other more generalized seismic source zones that could affect the Paducah site. Part 1: Introduction 'Io the south of Paducah is the New hfadrid seismic zone (NhtSZ), a zone of dense microseismic activity (Figs. A-1 and A 2). It is within this zone of current seismicity, that the great New hfadrid earthquakes of 1811-1812 are believed to have occurred (Fig. A 2)(Johnston,1996; Johnston and Schweig,1996). To the north of Paducah is the Wabash Valley seismic zone; an area of diffuse seismicity with a palcoscismic history of earthquakes greater than magnitude 7 (Obermeier et al.,1993; hiunson and hiunson,1997). Structure of the htississippl Embayment The NhtSZ lies within the hiississippi embayment, a broad and gentle south-southwest-plunging syncline of Cretaceous and Tertiary age (Stearns,1957; Stearns and hfarcher,1962)(Fig. A-3). Late Cretaceous and Tertiary sediments sit with angular unconformity on top of Paleozoic strata. North of the Ouachita Front, the syncline plunges 0.21' southwest and is slightly asymmetric with western limb dips of 0.59' and eastern limb dips of 0.34*. Thus, although the hiississippi embayment is commonly portrayed in stmetural figures as a conic section, the Paleozoic /Creta-ceous unconformity is essentially a flat surface. I,ying beneath the northern hiississippi embayment is the Reelfoot ria (Fig. A-2). Seismogenic bsement faults of the NhlSZ are believed to have originated in Late Precambrian and Early Cambrian time during formation of the Reelfoot ria. The ria is largely defined by geophysical l maps (Ervin and hicGinness,1975; lilldenbrand,1985; Langenheim and Hildenbrand,1997; Braile et el,1997). Clear definition of faults is achieved through seismic reflection profiling (see Dart,1995, for the status of reflection lines in the northern hiississippi embayment). Nelson and I Zhang (1991) collected COCORP lines across the embayment that imaged the geologic section to depths of 20 seconds of two-way-travel time. The COCORP lines reveal that the Reelfoot rin contains steeply-dipping faults and is bounded by inward-dipping listric normal faults. hiost of l the faults imaged in the COCORP data do not appear to have Tertiary displacement. Ilowever, this is probably a function of the resolution of the data; COCORP is designed for deep uustal imaging. Oil exploration seismic reflection lines (Ilowe,1985; Crone et al.,1985), vibroseis lines (Zoback,1979; llamilton and Zoback,1982), a hiississippi River reflection survey (Shedlock and Harding,1982), hiini Sosie lines (Sexton and Jones, 1986; 1988; Luzietti et al., 1992;1995; Schweig et al.,1992; Stephenson et al.,1995; Van Arsdale et al.,1995a; Purser,1996), and S-wave studies (Woolery et al.,1996) reveal Tertiary faulting on numerous faults in the upper hiississippi embayment. hiini Sosie lines image the uppermost crust between depths of 1,200 m and 70 m, rnd therefore have been obtained to determine shallow deformation. hiini Sosie surveys have revealed Tertiary and Quaternary faulting beneath the margins of Crowley's Ridge (Van Arsdale et al., 1994; 1995a), Sikestons Ridge (Sexton,1992), Blytheville arch (Van Arsdale l et al.,1996), Benton Hills (Palmer et al.,1997a; 1997b), along the Rectfoot fault (Sexton and l A1

Jones,1986; 1988; Purser,1996), Cottonwood Grove and Ridgely faults (Stephenson et al., 1995; Purser,1996), Dootheel lineament (Schweig et al.,1992; Sexton et al.,1992), Crittenden County fault (Luzietti et al., 1992; 1995), and one of the west-bounding faults of the Reelfoot riA (Jonesboro Junction fault)(Van Arsdale et al.,1995a)(Fig. A 4). Western Marnin of ti.e Ra.Iroot RIA - Various studies demonstrate that faults within the westem margin of the Reelfoot riR have Quaternary displacement and thus may be seismic source zones (Schweig and Van Arsdale,1996) (Fig. A-4). Geophysical studies along the Commerce geophysical lineament suggest that this is an active riR margin fault (Langenheim and Hild:abrand,1997). The Commerce fault segment of the lineament, at Thebes Gap, has Quaternary movement that displaces 10 to 24 thousand year old Peoria loen (Harrison and Schultz,1994; Palmer el al.,1997a; 1997b; Harrison, personal communication). W Jonesboro Junction fault appears to have Quaternary reactivation (Fig. A-4)(Van Arsdale et al.,1995a). Although not well tied to a particular structure, the northem arm of the NMSZ appears to follow a fault that we call the East Prairie fault (Figs. A-2 and A-4). Strain measurements indicate that the western half of the Reelfoot riA is undergoing right lateral shear at a rate of 0.1 microstrain per year; 1/3 that of the San Andreas fault (Liu et al.,1992). Thus, the current streu field appears to be loading strain energy within the western margin of the Reelfoot riR. Crowlev's RIAge Crowley's Ridge is a topographic ridge that trends diagonally acrou the underlying westem Reelfoot riA bour.dary (Fig. A-4). Mini Sosie profiles have revealed faulting beneath the ridge margins (Van Arsdale et al.,1995a) and geomorphic analyses d he ridge suggest that it may be a t horst, which is subdivided into different segments underlain b, s.screte fault blocks (Cox,1988a; Boyd and Schumm,1995; Spitz and Schumm,1997) Most of the upliA is Tertiary in age but there is geomorphic evidence that arching occurred in Wisconsin time (Van Arsdale et al.,1995m). Fa=*ern Marain of the Reelfoot Rift h eastern margin of the Reelfoot riA has been tectonically active in Quaternary time along the Crittenden County fault zone (Fig. A-4)(Crone,1992; Luzietti et al., 1992; 1995). The Mississippi River flood plain sediments are faulted to within 7 m of the ground surface (Williams et al.,1995). Since the flood plain is Holocene in age at this location, then the faulting reflects Htlocene movement. However, a trench excavated over the Crittenden County fault found no clear evidence of surface faulting (Crone et al.,1995). As more microseisms are plotted, there also appears to be an alignment of seismicity along the eastern margin north of Memphis (Fig. A-

2) (Braile et al.,1997; Chiu et al., in press).

New Madrid Seinmic Zone The principal seismic activity within the upper Mississippi embayment is interior to the Reelfoot riA along the NMSZ (Fig. A 2).' The NMSZ consists of three principal trends of seismicity; two northeast-trending arms with a connecting northwest-trending arm. This seismicity pattern has been interpreted as a northeast trending, right lateral strike-slip fault system with a compressional lea stepover zone (Russ,1982; Schweig and Ellis,1994; Van Arsdale,1997). Earthquake

• are occurring at depths between 4 and 12 km in Precambrian granites and Lower Paleozoic sedimentary rocks.- The southern arm is coincident with the subcrop Blytheville arch, the central A2

l arm is coincident with the subcrop Pascola arch and surface Lake County uplin, and the northern arm trends at a low ang's to the western margin of the Reelfoot riA. The southern seismicity arm extends from Marked Tree, Arkansas, to Reelfoot Lake, Tennessee, and it is believed that the December 16,1811, M 8.1 New Madrid earthquake occurred on this segment and/or on the Bootheel lineament discussed below (Fig. A-4) (Johnsten,1996; Johnston and Schweig,1996). Earthquake focal mechanisms primarily reflect right-lateral strike-slip motion along this segment (Crone et al.,1985). The Blytheville arch has I km of structural relief on the Paleozoic section and formed before Late Cretaceous time (probably Mid-C.etaceous) because the Upper Cretaceous sec. tion has only minor structural displacement. Vibroseis (Hamilton and Zoback,1982) and Mini-Sosie (Van i Arsdale et al.,1996) reflection data reveal faulted Tertiary and Quaternary strata along the Blytheville arch. Hamil;on and Zoback (1982) cite 80 m of down-to-the-southeast cumulative displacement of Cretaceous and Tertiary reflectors acrou the arch's southeastern boundary near Osceola, Arkansas. Shedlock and Harding (1982) document faults with vertical displacements of 10 to 15 m associated with a broad arch with 55 m of structural relief on the Cretaceous section near Caruthersville, Arkansas. N:ar Manila, Arkansas, Van Arsdale et al. (1996) also document Quatemary faulting. This suggests that the Blytheville arch has been liRed over much ofits length during Quaternary time and perhaps during the December 16,1811, New Madrid earthquake. The Reelfoot fault of northwestern Tennessee is believed to be responsible for the February 7, 1812, M 8.0 New Madrid earthquake (Johnston,1996; Johnston and Schweig,1996) (Figs. A-4 through A-7). This fault is an up-to the-southwest thrust / reverse fault within the lea stepover zone of the NMSZ. The Reelfoot fault surfaces as the Reelfoot scarp, which has been mapped from the southwestern margin of Reelfoot Lake into the town of New Madrid, Missouri, for a total dis;ance of 32 km (Fig. A 5)(Van Arndale et al.,1995b; Kelson et al.,1996). Unpublished Mini Sosie seismic lines along the Obion River of western Tennessee, reveal a fault that is tentatively identified as the Reelfoot fault. Thus, the Reelfoot fault appears to continue 20 km southeast of Reelfoot Lake and, based on the current microseismicity pattern, the fault probably continues another 20 km to near Dyersburg, Tennessee for a total length of 72 km (Figs. A-2 and 1 A-4). The north-trending Reelfoot scarp forms the eastern margin of the Lake County upliA in - Tennessee and turns northwest in Kentucky to New Madrid, Missouri, paralleling the seismicity trend (Figs. A 2 and A 5). A Mini-Sosic line, acquired along the soc.hm wrgin of Reelfoot Lake, reveals that episodic movement on the Reelfoot thrust, apparently since the Cretaceous, has resulted in 70 m of displacement at the top of the Paleozoic and 10 m at the ground surface (Figs. A-5 and A-6)(Purser,1996). This same reflection line reveals that the Mississippi Valley bluffs are underlain by the down-to-the-west Ridgely fault zone. Thus, the bluffs appear to be structurally controlled east of Reelfoot Lake (Purser,1996; Van Arsdale et al.,1996). The bluffs i continue northeast along the apparent strike of the Ridgely fault sysem. However, we do not I know how far northeast the Ridgely fault system continues. Based on the presence of the Ridgely (bluff margin) fault, we speculate that other northeast-trending Mississippi Valley bluff l margins may be underlain by Reelfoot rin faults. For example, in Figure A-5 we have turned the Cottonwood Grove fault north to continue as the New Markam fault (Van Arsdale et al.,1996). A3 . z

It is also pos2ible that the Cottonwood Grove fault continues along a northeast strike in which case it would underlie a northeast trending bluffline northeast of Reelfoot Lake. The Lake county upliA is a consequence of deformation within the hanging wall of the Reelfoot fault (Purser,1996). Essentially, the Tiptonville dome is a horst stop the Lake County upliA horst, within the hanging wall of the Reelfoot fault (Fig. A 7). If the Figure A-7 model is correct, then the Lake County upliA and Tiptonville dome are bounded on their west sides by kink bands (back thrusts) that probably had displacement during the February 7 earthquake. The northern arm of the NMSZ extends from New Madrid towards Charleston, Missouri. It appears that, like the southern arm, this segment is coincident with a northeast striking fault zone (East Prairie) and that right lateral faulting is the most common type of fault plane solution. H:: wever, as revealed in Figure A 2, this arm is not parallel to the ria margin. The structure of this northem seismicity trend has not been well imaged with seismic reflection and therefore we know very little about it (Hamilton and Zoback,1982). I Seismicity near New Madrid, Missouri, is complicated by a minor westerly trend. McKeown (1982) proposed that the Bloomfield pluton is locally influencing the seismicity and that the l I - seismicity essentially wraps i ound tha. pluton. McKeown (1982) also notes that plutons in the Mississippi embayment, like the Bloomfield pluton, consist of different rhoologies and may thus be acting as stress concentrators. Although the Bloomfield pluton may be locally influencing the distribution of earthquakes there is no convincing evidence of plutons along most of the NMSZ (Nelson and Zhang,1991). The Bootheel lineament is believed to be a coseismic fault of the New Madrid 1811-1812 earthquake sequence (Schweig and Marple,1991; Marple and Schweig,1992; Johnston and Schweig,1996). However, there is no evidence of surface faulting along the lineament. It should be pointed out; however, that there is also no evidence of 1811-1812 surface faulting along the - Blytheville arch or the northern NMSZ seismicky arm. Mini-Sosie reflection data does reveal flower structures beneath the Bootheel lineament (Schweig et al.,1992; Sexton et al.,1992) and the lineament has been interpreted as a linking fault that is in the process of bypassing the unstable Reelfoot fault (Fig. A 5 inset)(Schweig and Ellis,1994). Sikeston Ridge Sikeston Ridge is a 50 km long by 12 to 5 km wide topographic ridge that trends northerly from New Madrid, Missouri (Fig. A-4). This ridge has been explained to have originated as a Pldstocene erosional remnant lea between two Pleistocene courses of the Mississippi River (Saucier,1974; 1994). However, numerous faults underlie the ridge (Sexton,1992). The faults with the most displacement are inward-dipping normal faults that are near the ridge's margins. Essentially, Sikeston Ridge overlies a graben. If Sikeston Ridge is solely crosional in origin, we would have to accept the fact that the crosional superposition of the ridge atop a graben is coincidental. This is highly unlikely. We believe that Sikeston Ridge is tectonic in origin, and that the graben has been tectonically inverted. That is, the Noxal and Farrenburg normal faults within the margins of the graben have been reactivated as reverse faults. This reactivation must have A4

=_ occurred in Pleistocene or Holocene time because the ridge is capped by Pleistocene sediments (Saucier,1974). Discussion of the Structure of the Unoer Mississippi Embayment Faults with Quaternary displacement appear to be within the northeast trending Reelfoot rift (Fig. A-4). Right lateral strike-slip motion on the rift-parallel faults is compatible with t..: contemporary stress field. Seismicity defining the northwest trending Reelfoot thrust fault appears to cr. tend from the easterr margin of the Recifoot rift to Fayond the currently mapped western rift margin (Fig. A 2). It is therefore possible that the Reelfoot fault is truncated at its western end by an unidentified rift-mugin fault. The fault bend fold model(Suppe,1985) suggests that deformation within the stepover zone is controlled by the Reelfoot fault and associated parallel kink bands (back thrusts)in the hanging wall (Fig. A-7). Crowley's Ridge has Tertiary faulting beneath its margins but it appears that the most recent movement on these faults may be Pleistocene (Van Arsdale et al.,1995a). However, it must be noted that the northern portion of Crowley's Ridge, and presumably its bounding faults, are nearly parallel to the rift margin faults and the Bootheel lineament. Therefore, we propose that the faults that bound the northern portion of Crowley's Ridge are capable of generating earthquakes. The Reelfoot and Ridgely faults have identical total displacements and they appear to have had identical displacement histories (Purser,1996). This suggests that the multiple main shocks of 1811-1812 were not unique in the history of the NMSZ and a similar mode of multiple-event seismic moment release should be anticipated for future large earthquake occurrence. Earthquake Surface Rapture and Liquefaction in the New Madrid Seismic Zone and Western Lowlands Reelfoot Fauh Although the 3 great earthquakes of 1811-1812 had moment magnitudes of 8.1, 7.8, and 8.0, the only known surface fault is the Reelfoot reverse-fault scarp Apparently, the first and second events occurred on strike-slip faults, thereby leaving no scarps. Trenches excavated across the Reelfoot scarp have revealed un east facing monocline broken by normal faults, reverse faults, and sand dikes. Russ (1978; 1982) first excavated a trench across the scarp and concluded that there had been 3 faulting events within the last 2,000 years. Subsequently, a better dated earthquake chronology has been interpreted from scarp-derived colluvium and graben fill sediments (Kelson et al.,1992; 1996). Three faulting events have occurred along the Reelfoot scarp within the last 2,400 years; between A.D. 780 and 1000, between A.D.1260 and 1650, and during A.D.1812. Thus, the recurrence interval of the Reelfoot fault is estimated to be 400-500 years (450 years) (Kelson et al.,1996). Liouefaction (Sand blows) Paleoliquefaction has been a particularly powerful tool in deciphering the palco,eismology of the NMSZ (Tuttle and Schweig,1995; 1996; Tuttle et al.,19%; Schweig and Van Arsdale,19961 Paleoliquefaction sites consisting of sand blow deposits and sand dikes are located within the A5

NMSZ and in the Western Lowlands west of Crowley's Ridge.- Paleoliquefaction studies in the Western Lowlands ofMissouri have identified 4 palcoseismic events dating from 23,WO-17,000 B.P.,13,430 9,000 B.P., A.D. 2401,020, and A.D.1,4401,540 (Vaughn, 1991; 1992; Vaughn et al.,1993). Vaughn speculates that the susade source of the Western Lowland earthquakes may have been the Commerce fault, a fault beneath the western margin of Crowley's Ridge, or the NMSZ. i Studies within the southern ponion of the NMSZ seveal a minimum of 3 paleoliquefaction events in the 2,000 years prior to 1811 (Fig. A-8)(Schweig and Van Arndale,1996). These events are i believed to have occurred between A.D. 0 500, A.D. 8001,000, and A.D.1,2001,400. At the northern end of the NMSZ, paleoearthquakes have been dated at approximately A.9. 439, A.D. $39-911, and A.D. 7701,020, The Wabash Valley Seismic Zone Structure of the Wahaah Wilev Seimmic 7ana The Wabash Valley seismic zone (WVSZ) geographically includes portions of Kentucky, Indiana, and Illinois, and geologically lies within the Illinois basin (Figs. A 9 and A-10). Principal structures within the WVSZ are the Rough Creek graben, Hicks Dome /Fluorspar district, and the Wabash Valley fault system (Kolata and Hildenbrand,1997). The Reelfoot riA and the Rome trough are part of a structurally cont'muous late Precambrian to early Paleozoic riA system that continues east into the Rome trough of centra' Kentucky. Based on geophysical evidence, Braile et al. (1982; 1986; 1997) advocate that the Reelfoot riA ectually splits into three arms; the St. Louis arm, the Rough Creek graben, and the southern Indiana arm . (Fig. A-ll). In the Braile et al. model the Reelfoot riA continues northeast to include the Watash Valley seismic zone. However, other authors have argued that there is no geologic evidence for a major riA system in southern Illinois and Indiana (Bear et al.,1997; Hildenbrand and Ravat, 1997)._ The Wabtsh Valley fault system and underlying Grayville graben is the only fault system in the region and it is simply too small to be an extension of the Reelfoot riA (Bear et al.,1997). Thus, many authors believe that the Rough Creek - Shawneetown fault system is a major structural boundary that separates the two seismic zones into discrete seismic source zones (Fig. l A.10) (Kolata and Hildenbrand,1997; Hildenbrand and Ravat,1997; Wheeler,1997; Bear et al., j 1997), Based on various geologic arguments Wheeler (1997) believes that the northernmost l-extent of the New Madrid seismic zone coincides with the bend from the northeast trending l Reelfoot riR to the east trending Rough Creek graben (Fig. A-12). However, the Commerce l Geophysical lineament (Langenheim and Hildenbrand,1997; Hildenbrand and Ravat,1997) has - been mapped from Arkansas into central Indiana. Quatemary faulting is evident along the l Commerce fault segment at Thebes Gap in Missouri where the lineament appears to be a western border fault of the Reelfoot riA. Langenheim and Hildenbrand (1997) have speculated that this lineament may be the source of historic earthquakes iri Illinois.- 1-l ~

Although there is no clear evidence that the Reelfoot rift and Illinois basin are currently structurally connected, there is stratigraphic evidence that the Ihmois basin and Rectroot rift were part of the same inland seaway for much of the Paleozoic Era. Prior to uplift of the Pascola arch, the proto-Illinois basin was a broad southward-plunging trough extending to the lapetus Ocean (Kolata and Hildenbrand,1997). Thus, it appears that the Reelfoot and Illinois basins were stmeturally contiguous for a very long time but that uplift of the Pascola arch sometime between ~ post Pennsylvanian and pre late Cretaceous time divided the trough into two separate basins. Uplift of the Pascola arch probably occurred in Permian and middle Cretaceous times (Cox and Van Arsdale,1997). Ouaternary Fauhing in the Wabash Vallev Seismic Zone No Holocene (past 10,000 years) fault displacement has been identified in the WVSZ (Nelson et al.,1997). However, geologic mapping in southern Illinois and westem Kentucky has identified Quaternary tectonic faulting in two areas; the fluorspar area fault complex in Pope and Massac counties, and the Commerce fault zone in Alexander County (Fig. A-13)(Nelson et al.,1997). Northeast striking faults in the fluorspar area fault complex displace Mounds Gravel oflate Miocene to early Pleistocene age (11 to 1 million years) and locally dispha Metropolis terrace Bravel that is believed to be Illinoisaa or older in age (> 100,000 years). No sediments of Woodfordian or younger age (< 20,000 years) are faulted within the fluorspar area fault complex. The Commerce fault zone displaces Meunds Gravel :n Illinois and Peoria loess (10,000 to 24,000 years old) in Missouri. All of the Quaternary faulting appears to be primarily right-lateral strike-slip with a minor component of normal movement. The fluorspar area fault complex consists of a series of faults of various orientations but with a preferred northeast sirike (Figs. A-10 and A-13). These faults bound horsts and grabens within the fluorspar mining district ofIllinois and Kentucky. Near New Columbia, in northern Massac County, a graben associated with the Lusk Creek fault zone has down-dropped Mounds Gravel 30 to 45 m (Figs. A-13 and A-14). The Hobbs Creek fault zone in Massac Cousy contains a blxk of Metropolis gravel 600 m long by 100 m wide that has been tectonically lowered 57 m (Figs. A-13 and A-15). Metropolis gravelis interpreted to be a gravel derived from the Mounds Gravel (Nelson et al.,1997). Also within Ma;. sac County the Barnes Creek fault zone displaces Metropolis gravel and along the fault zone are small grabens and clastic dikes 611ed with Metropolis gravel (Figs. A-13 and A-16). It appears that the most necent Barnes Creek faulting occurred after deposition of the Metropolis gravel and pre-Fcrmdale (> 35,000 yearr' In Livingston County, Kentucky, the Lockhart Pluffgraben displaces Mounds Gravel 120 m in a right-lateral sense (Fig. A-13). The Conmce fault zone discussed above as part of the Reelfoot rift continues northeast into Illinois at Thebes Gap (Fig. A-13). This fault is considered as part of the very long Commerce Geophysical lineament that extends from Arkansas into Indiana (Langenheim and Hildenbrand, 1997; Hildenbrand and Ravat,1997). The Commerce Geophysicallineament in Illinois is defined as a northeest-trending zone of subtle gravity and magnetic h:3 s. Although no continuous fault h system lies along this lineament there is discontinuous faulting in Alexander, Union, and Johnson counties that lies along the lineament (Fig. A-13). However, these faults are exposed in coal mine high walls and do not displace Pleistocene loess. l l A-7

Paleeseismology in the Wabash Valley Seismic 2,one Paleoliquefaction studies have been conducted along most of the major rivers of southern Indiana { and Illinoit. In these studies, river bank sediments are studied for evidence of earthquake-induced sand dikes, sand sills, and sand blows. Paleoliquefaction studies in the WVSZ have revealed at least eight prehistoric earthquakes within the last 20,000 years that were strong enough to cause liquefaction (Figs. A-9 and A-17)(Obermeier et al., 1991; 1993; Munson et al., 1992;1997). Of these eight earthquakes, six were probably 2 M 6 and at least two are estimated

o have exceeded M 7. The strongest prehistoric carthquakes have been located in and near the WVSZ in the lower Wabash Valley ofIndiana and Illinois; however, some prehistoric events exceeding M 6 have occurred outside of the WVSZ where there have been no historic earthquakes (Munson et al.,1997).

The largest prehistoric earthquake recorded in the paleoliquefaction record occurred about 25 km west of Vincennes, Indiana (Figs. A-9 and A-17). This earthquake, the Vmeennes-Bridgeport event, is estimated to hwe been a M 2 7.5 (magnitude-bound method) and occurred 6,100 200 yr bP. Bned on geotechnical studies (energy-acceleration method) this event is estimated to have been a M 7.7 (Pond and Martin,1997). The second strongest earthquake occurred 40 km southwest of Vincennes r.nd has been called the Skeleton-Mt. Carmel earthquake. This event occurred approximately 12,00011,000 yr BP and is estimated by Munson et al. (1997) to have been a M 7.1-7.2 and M 7.3 by Pond and Martin (1997). The third greatest prehistoric earthquake is called the Vallonia carthquake and its estimated epicenter was 100 km east of the Wabash River. This event occurred 3,9501250 yr BP and its estimated magnitude ranges from M 6.9 (Munson et al.,1997) to M 7.1 (Pond and Martin,1997). The Martinsville-Waverly earthquake occurred about 30 to 50 km southwest ofIndianapolis, is estimated to have been a M 6.8 to 6.9, and occurred between 8,500 and 3,500 yr BP (Munson et al.,1997). A smaller earthquake occurred 4,000 500 about 35 km southeast of Vincennes at Inola. This earthquake is estimated to have been a M 2 6 but less than 6.7 (Munson et al.,1997). At 2,0001500 yr BP a M 6 earthquake is believed to have occurred 60 km northeast of Vincennes near the town of Einora. The only other paleoliquefaction event is dated at 20,000 2,000 yr BP and is believed to have occurred 50 km south-southwest ofIndianaoolis. However, insufficient dats is available to estimate its magnitude. Most of the paleoliquefaction dates have been acquired in Indiana with many sites in Illinois not yet fully studied and dated. However, preliminary Illinois data suggest paicocarthquake epicenters in Illinois. Paleoearthquake epicenters are tentatively proposed for central Skillet Creek in southeastern Illinois (3,760 70 yr BP), in the upper Little Wabash River valley (18,500 yr BP),35 km east of Springfield (unknown sge but M 6), along the south fork of the Sangamon River (about 10,000 yr BP), and the central and lower Easkaskia River (mid Holocene) (Munson et al.,1997). Absence of paleoliquefection in Ohio River sediments along the Indiana-Kentucky bcrder and the Illinois-Kentucky border suggest that this area has not experienced severe ground shaking in the last 4,500 years (Fig. A-17) (Munson et al.,1997). However, these authors suggest that a 5-6 m thick clay cap may have kept sand dikes from penetrating to levels above the current maintained stage of the Ohio River. A-8

u Part 2: Source Parameters and Seismic Potential Introduction The faults of the upper Mississippi embayment and the Illinois basin /Wabash Valley that 6 - been identified in this report indicate that this region experienced a significant level of seismic activity during the Quaternary and to a lesser extent during the Holocene (last ~10,000 years) (Table A-1). However, of all the faults described in this report, only those of the NMSZ (the first 5 listed in Table ml) are known to have ruptured in M = 7 earthquakes in historic times. The seismic and geologic data are sufficient to demonstrate (1) fault (s) existence at a specific location; (2) fault orientation and sense of displacement (sometimes); (3; approximate age of most recent displacement (sometimes); (4) recurrent displacements (some'imes); (4) fault length (but not fault width, although seismic profiling can confirm if a fault is deco-rooted, i.e., tectonic) This listing represents only a portion of the input data needed for PSHA 'or example, for those faults where length and type of faulting can be estimated, there are rough regression relations to recover moment magnitude M fiom fault length. But one cannot be sure that the entire length of the fault can rupture in a single earthquake (if not, M is overestimated), or whether the rupture can link to other fault strands (if so, M is underestimated). Independent measurement of ctrain rates holds the promise of constiaining the possible styles and rates of faulting, but the GPS technology and deployr r nt time spans necessary to achieve this are in their infancy and are unlikely to be useful for PSHA for at least a decade. This means that recurrence for most of the faults is constrained by only the available geologic cr paleoseismic data. Typically, these data are sparse or nonexistent, so the constraints are weak. Therefore, the best approach for seismic source zoning in this region is to (a) identify all known (Quaternary) faults (Part 1), (b) determine their most recent movement (Part 1), end (c) use historical and recent seismicity to identify and characterize other possible seismic sources, not identined as discreet faults, and (d) estimate recurrence from paleochronologies, where available and from the time of most recent movement otherwise. Hence in this Part 2 section, four types of seismic sources will be examined and characterized as to their seismic potential: (1) faults with known (historical) earthquakes, (2) faults with Holocene offset, (3) other Quaternary faults, and (4) other generalized seismic source zones. Expected Sources and Source Parameten for Future Earthquakes Eaults witMnown (Historican Earthauakes Only the faults of the NMSZ, which is defined mainly by current microseismicity, may be associated with known, historical earthquakes. As shown in Figs. A-2 and A-4 these are the Blytheville arch [ including a NE extension as the Cottonwood Grove fault (CGF)], the Bootheel lineament, the Reelfoot fault (including its northern extension to NW of New Madrid and southern extension to north of Dyersburg), and, with lower confidence, the East Prairie fault. Results of a multi year project to esti7nate the seismic moments of the principal 1811-12 earthquakes and to determine which faults they ruptured were published in 1996 (Johnston,1996; Johretet' and Schweig,1996). Although aspects of this research are ongoing, the conclusions are for the most part up to date and will be used in this report. With the exception of the East Prairie A-9

fault, we assume that the principal 1811-12 events represent maximum or characteristic eenhquakes for their respective faults. The D1 Fadhanake F2nha The three principal New Madrid earthquakes that occurred on 16 December 1811,23 January 1812, and 7 February 1812, are designated Dl, J1 and F1, rupectively (Fig. A-18). D1 was M8.1

• 0.3 and fault modelitig indicates a mpture length of

~140 km with ~10 m slip is required (Figs. A 20, A-21). Historical accounts from the town New Madrid have the vibration and sound coming from the southwest; from the Arkansas Post, far sotith on the Arkansas river, they came from the northeast; and from Mississippi River boatmen - they came from the west or northwest. Crude triangulation places the rupturejust in the region of a major crustal transpressional fault structure identi6ed with seismic reflection data--the Blytheville arch (Fig. A-2). Therefore, the D1 rupture can be placed on the arch with con 6dence. However, accour.ts from Little Prairie (near present-day Caruthersville on the CGF extension of the Blytheville arch and its seismicity band), speak ofD1 as a " distant earthquake" and how the major aftershock of 7:15 am,16 D==h~ was much more severe, especially in terms of the - massive liquefaction generated. For this reason, Johnston and Schweig (1996) completed the D1 rupture on the Bootheel lineament (Fig. A-4) and placed the major D1 aftershocks on the NE extension of the Cottonwood Grove (see the Contempormy GroumiMotion Estimates in 1811 section). For the purposes of this report, it makes no significant difference for Paducah ground motions whether the D1 rupture extended from the Blytheville arch onto the Bootheel lir=amant or continued on strike onto the CGF. Hence these faults are grouped as one with M.,x = 8.1

• 0.3.

(All quoted uncertainties on M are 1-o. To be consistent with present-day focal mechanisms, - regional stress regime, and Blytheville arch structure, D1 would have been a right-lateral strike-slip rupture. Directivity effects could be important; the Little Prairie accounts indicate the rupture - came from the SW, suggesting either SW-to-NE unilateral rupture or bilateral rupture from nucleation near the Blytheville arch-Bootheel lineament intersection. This would amplify both the S body wave and the Love surface wave along fault strike to the northeast, i.e., toward Paducah. The J1 Furthquake Fault. The source fault for J1 is unconstrained except by the following indirect argument (M. Ellis, paper in prqaration,1997). Strike-slip faulting termination concentrates high stress (and strain) levels at the ends of the rupture. A D1 termination at the - northern end of the Bootheel lineament would therefore concentrate stress on the southern end of. the East Prairie fault (EPF, labeled E in Fig. A-4). A J1 right-lateral rupture on the EPF, combined with right-lateral aftershocks of D1 on the CGF, loads compressive stress onto the left stepover zone between the two. This is just the locale of the Reelfoot fault (RF), the surface expression of a thrust fauit at depth inferred from precise microseismicity hypocenters (Chiu et al.,. 1992; Pujol et al.,1997). Also, a historical account from New Madrid (E. Bryan,1816) says that the J1 earthquake was "...as violent as the severest of the former ones", even though the far-field intensity data clearly show it was smaller than Dl, suggesting the J1 mpture was close to the town. In summary, fault mechanics arguments and meager historical data favor the EPF as the source fault for J1 (NM2 of Fig. A-21). Johnston (1996) has esthnated the following J1 source parameters: M = 7.8 + 0.3, fault length = 65 km, average slip = 7.5 m, and static stress drop of A 10 I

i l i i 65 bars (Figs. A-20 and A-21). With this fault length, the 1895 M6.6 Charleston, MO earthquake i occurred within the NE high stress fault tip zone of the J1 rupture. As with Dl, SW to NE directivity on the EPF could enhance S and Love (Lg) wave motions at Paducah. The F1 Furthayance Faults. Historical evidence would allow us to place F1 on the Reelfoot fault as a thrust faulting rupture even without the scientific evidence for compressional tectonics in this stepover zone. At least six independent accounts cuimate that the town New Madrid sank from 8 to 15 feet (relative to the Mississippi River level); no account mentions subsidence for D1 or J1. E. Bryan's (1816) report is especially valuable: _ "The site of this town (New Madrid) was settled down at least 15 feet, but not more than a half mile below town there does not appear to be any alteration of the bank of the river." This description is entirely consistent with New Madrid being on the footwall of a thrust fault-the Reelfoot fault-while just below (downstream), the river i and bank are on the hangingwall (see Fig. A-5). New Madrid may have subsided to a degree, but most of the change was the elevation of the hanging wall and the river rising to maintain its level there. The relative change at New Madrid was subsidence relative to the foot wall river level, but the absolute change was the river on the foot wall rising relative to New Madrid. The other historical evidence comes from two riverboatmen stopped for the night on the river on the eastern limb of Donaldson point, upriver of New Madrid, near where Reelfoot fault first t crosses the river north of Reelfoot Lake (Fig. A-5). In their accounts the boatmen give the correct date and time of F1 and give their precise locations (see Johnston and Schweig,1996, for details). t Both survive the F1 mainshock at ~3:45 am (local) and go on to describe both barriers and falls or i-rapids on the river that are consistent with where these features would be expected from Figure - A-5. Hence the F1 event can be pla'.:ed on Ree! foot fault with high confidence. Johnston (1996) ) i determined a M = 8.0

• 0.3 for this earthquake.

The length of the Reelfoot fault and the amount of displacement during F1 are critical issues that have not been completely resolved. Van Arsdale et al. (1995b) demonstrated the fault extends from south Reelfoot basin, which is the intersection of the CGF and the RF, across Kentucky l-bend to west of New Madrid for Lmin~32 km. High topography extends ~5 km further NW to an approximate intersection with the EPF. To the SE-beyond the left stepover zone between the EPF and CGF-current thrust-plane microseismicity extends ~32 km beyond the CGF l intersection to NE of Dyersburg, and very spw.e seismicity suggest the RF faulting may extend a funher 10 km and terminate at the Reelfoot rift's SE margin. These NW and SE extensions would increase the Reelfoot fault length to ~75-80 km; Johnston (1996) used L = 75 km to model the F1 rupture (Figs. A-20 and A-21). Seismic reflection work now in progress should answer the question of whether the Reelfoot fault can be imaged on this SE extension of RF north of Dyersburg. Currently, the best evidence that L the SE Reelfoot fault extension was included as part of the F1 rupture is Obion Lake. This lake, which no longer exists on the channelized Obion River, is shown on maps as early as the 1820's. It was 1/3 to % the size of Reelfoot Lake, and the location of the downstream terminus of the lake j is consistent with the microseismicity's indication of where the fault plane should cross the river.' I ) Furthermore, the early Rutherford survey of the region in 1785 shows that no lake existed at that time. Therefore historical evidence brackets the formation of Obion Lake between 1785 and the i l 1820's, consistent with the scenario oflake impoundment by the F1 thrust fault scarp 1 i A Il i

l We take the historical and microseismicity data cited above as evidence that F1 rupture involved the entire extended Reelfoot fault length of-75 km. This, then, revises the preferred scenario of Johnston and Schweig (1996) in which F1 dip-slip rupture was limited to the stepover zone but continued as strike-slip faulting onto the western seismicity branch (Fig. A-18). The two scenarios yield roughly the same total fault length and seismic moment but possibly could produce significantly different ground motions at Paducah since the revised scenario brings the average fault distance ~10-15 km closer to the site. It also increases the likelihood of strong directivity effects at the site since updip rupture on the SW dipping Reelfoot fault plane would enhance S and Love wave amplitudes to the NE, i.e., toward Paducah. In summary, the best model for the F1 earthquake is that it was a M8.0 0.3 thrust rupture on the Reelfoot fault. It breached the Mi:Sissippi River in three places, twice as a barder and once as a fails or rapids, and it impounded two lekes, Reelfoot and Obion. The faulting parameters (Figs. A-20 and A-21) are L = 75 km, W = 45 km, average slip = 8 m, and static stress drop = 50 bars. The distribution of slip on the fault plane and the configuration of the fault at depths greater than 12-14 km are unknown. It is possible that coseismic slip and moment release was distributed in a more complex fashion, as in the fault propagadon fold mod:1 ofFigure A-7, but at present the data are insufficient to warrant including models other than simple planar shear on RF in a PSHA. A final note concerns the vertical component of the net displacement -the throw-of the FI rupture. Except for the Reelfoot fault, the NMSZ is a strike-slip fault zone; therefore the potential for significant topographic changes is highest for the Reelfoot fault. The height of the Tiptonville dome and the contribution to its height from the F1 earthquake have an important bearing on the recurrence intervals between large or ' characteristic' earthquakes of the NMSZ faults and is discussed in the next section. Recurrence for the Faults of the NMEZ The NMSZ proper consists of the CGF/Blytheville arch,- Bootheel lineament, Reelfoot fault, East Prairie fault, and the' strike-slip westward extension of the Reelfoot fault. The developing paleoliquefaction chronology for the NMSZ (see Part 1) is not fault specific because earthquakes on the order of M8 have fault dimensions within a factor of two or so of the entire NMSZ. This means that the strong liquefaction zone for each NMSZ fault, which may extend 50 km or more from the fault, will overlap with similar zones for all NMSZ faults. Therefore, for a given liquefaction feature, there is no way to be sure of the source fault for the seismic waves that produced it. The one exception to this statement is the NMSZ south of Blytheville, AR, where the only major known fault is the Blytheville arch. A related question-critical for recurrence-is whether the normal mode for large seismic moment release in the NMSZ is by single characteristic earthquakes on individual faults at varying times, or by a concerted moment release in a sequence such as that of 1811-12, involving most or all the faults of the zone. Palcoseismological evidence to date loes have some input to this _ qu ;stion but we are still far from a definitive answer. The key h the trenching results of Kelson et al. (1996) (see Part 1). Three faulting events were identified: 780-1000 AD,1260-1650 AD, and 1812. These are consistent with the liquefaction sand blow chronology of major events in 800-1000 AD,1200-1400 AD and 1812 for the entire NMSZ. But the Kelson et al. results (also the Russ 1979,1982 results from an earlier RF trench) are fault-specific to the Reelfoot fault because they are from dating of scarp-derived colluvium and graben-fill sediments. And these Reelfoot A 12 w

fault paleoevents could not have produced all the paleoliquefaction features in the NMSZ, espedally those at the northern and southern extremes. We therefore have two paleoevent chronologies: one is specifically for the Reelfoot fault, the other is general for the entire NMSZ. And both chronologies are the same within the resolution . of the dating techniques. The similarity between the Reelfoot faul and the NMSZ chronologies support the second faulting scenario described above in which large seismic moment release normally involves mort or all NMSZ fault segments rupturing during a short time span sequence such as in 1811-12. This mode of activity also is supported by the Part 1 observation of similar displacement histories on the Ridgely and Reelfoot faults. The implication for recurrence is that the raleoliquefaction chronology being deveioped for the entire NMSZ will also apply to the major individual fault segments. This is the basis for our recurrence estimates in this report (Table A-1). Such estimates assume that the NMSZ faults rupture in characteristic ca-thquakes (i.e., an 1811-12-type sequence)---at least approximately-and that the paleoevents being dated are the charecteristic earthquak:s of individual faults.' These characteristic earthquakes also may be considered the maximum magnitude (Mmax) of the faults. There is one, somewhat qualitative check, on the above interpretation of the NMSZ pattem of recurrence. Reelfoot fault is the only fault of the NMSZ where we can measure the fault's throw in 1812 and can compare it with the cumulative vertical displacement on the fault. The est' nate u of 1812 vertical offset is provided by the former channel of Reelfoot River, which prior to 1812 must has e been continuous across Tiptonville dome to empty westward into the Mississippi River as shown on old maps. Today the relic channel on Tiptonville dome is 12-15 ft (~4-5 m) higher than the submerged relic channel in Reelfoot Lake. The maximum elevation difference of Tiptonville dome's eastern flank, which represents the maximum hanging wall-foot wall differential of Reelfoot fault is ~30 40 ft (9-12 m). Several meters of the height differential could have been lost to hanging wall eroson and foot wall sedimentation. Thus Tiptonville dome could have formed entirely in the last 2000 years in the three characteristic earthquakes identified by. Kelson et al (1996), each contributing ~4 m of fault throw. Three abandoned Mississippi River meander channels cross Tiptonville dome in the Reelfoot Lake vicinity. If these could be dated, they would provide much tighter constraints on the tectonic evolution of Tiptonville dome; Lacking that, the data to date are consistent with characteristic earthquake occurrence on the Reelfoot fault, each of which produce a vertical surface offset of~4 m. From the Kelson et al. (1996) trench data, such characteristic earthquakes reoccur every ~450 years. A 500-yea 9 recurrence for M8 earthquakes implies an extremely nigh strain rate, on the order of 5-10 x 10' /yr (depending on the strain storage volume) in ordei to renew the strain potential energy required. Such a rate is highly unlikelg/yr determined from the initial GPS s in a uidplate region and must be viewed with scepticism. The very preliminary rate of ~10' et al. (1992) applies to only a small portion of the NMSZ. Weber et al. (1997) cover a much larger region and find significantly lower strain rates but the uncertainties are high because the time interval between their GPS surveys was only two years. A-13

If mid-to upper-M7's-rather than M8's--could produce -4 m dip-slip fault throws, the required strain rate would be reduced significantly, This would imply the Reelfoot fault's characteristic canhquake is M7.5-8.0, and the associated recurrence interval is ~500 years. M7.5-8.0 is generally consistent with Johnston's (1996) estimated uncertainty for FI's mp/y gnitude, and when applied to the entire NMSZ, implies a more credible strain rate of-0.5x10 that the magnitudes of the earthquakes that produced the paleocarthquake features on the RF and throughout the NMSZ are unknown. Assigning them M7.5-8 0 rather than M=8.0, allows a 3 range from M7.5 with a 500-yr recurrence to M8.0 with ~10 >T recurrence and does not require an inordinately high strain rate. Given the uncertainties that attend all these arguments, we express our recurrence estimates for the faults of the NMSZ in terms of the following weighted alternative scenarios (RI= recurrence interval): n) CGF/Blytheville arch /Bootheellineament system Mmax= 8.1 0.3 RI = 1000 yr wt. = 0.70 Mmax= 8.l*0.3 RI = 500 yr wt. = 0.15 Mmax= 7.7*0.2 RI = 500 yr wt. = 0.15 b) East Prairie fault (EPF): Mmax= 7.8*0.3 RI

• 1000 yr wt. = 0.70 Mmax= 7.8i0.3 RI = 500 yr wt. = 0.15 Mmax= 7.5i0.2 RI = 500 yr wt. = 0.15 b) Reelfoot fault (RF):

Mmax= 8.0*0.3 Ri = 1000 yr wt. = 0.70 Mmax= 8.0io.3 RI = 500 yr wt = 0.15 Mmar= 7.6i0.2 RI = 500 yr wt. = 0.15 The uncertainty bounds on Mmax are symmetric in the above listing; this is purely an artifact of the statistical regression analysis approach of Johnsion (1996). Given the fault lengths available--based on current seismicity-to D1 on the CGF/Bytheville arch /Bootheel lineament system, it is much more difficult for D1 to be significantly larger than M8.1 than smaller. Therefore, we assign weights for Dl's magnitude range of M 8.4-8.1-7.8 of 0.1-0.6-0.3, respectively. The weights for J1 on the EPF remain symmetric. For F1 on the Reelfoot fault with M 8.3-8.0-7.7, the weights are (as with D1) 0.1-0.6-0.3, respectively. Finally, we note that these recurrence estimates derive from the clustered faulting scenario in which all three of the major NMSZ faults are active during a NMSZ characteristic seismic moment release episode. If this were not the case and the individual faults rupture independently, the recurrence interval far any characteristic earthquake in the NMSZ would be % to 1/3 of the above estimates. Such short intervals are simply not credible and are incompatible with any physically reasonable strain accumulation rate so we give zero weight to this alternative. Faults with Holocene Offset There are only three known faults in the study region, other than the principal NMSZ faults discussed above, for which offset of Holocene sediments has been documented. Two of these, the Commerce /Benton Hills and the Crittenden County fault zone (CCFZ) lie within or near the NW and SE margin zones, respectively, of the Reelfoot rift. Both of these are described in Part 1 A-14

as most probably rift-bounding normal faults that have been reactivated as oblique thrusts or transpressional strike-slip faults in the current ~ east-west compressional stress regime. The Benton Hills faulting is seen in both trench exposures and shallow seismic reflection profiles as subvertical, predominantly normal faults that are interpreted as the tops of flower structures overlying right-lateral / oblique faulting at depth (Palmer et al.,1997). The lengths of both the Benton Hills fault (s) and the CCFZ are poorly constrained; best estimates are given in the Table A-1 conclusions. With lengths of>20 km to 32 km, these faults could generate maximum earthquakes of M-7. We consider the possibility that Benton Hills-CCFZ-type faulting could extend for significantly greater distances along the rift flanks to be remote. No Quaternary deformation has been documented on any other segments of either the NW or SE rift flanks. If M7.5-8.0 earthquakes had occurred during the Quaternary on the rift margins, there should be a strong paleoliquefaction signature in the many river valleys that traverse the margins. To date, no such signature has been observed (although systematic searches have not been conducted). Consequently, in Table A-1 we assign a very low wei ht to the possibility that coseismic rupture could extend for lengths 6 >>30 km along either the NW or SE Reelfoot rift margins. The recurrence interval of rift margin ' characteristic' earthquakes is unconstrained. Since only two Holocene occupences are documented-one each for the NW and SE margins, we assign a Holocene (~10 yr)-level time span as the recurrence interval. This long repeat time, coupled with the M-7 limit on maximum magnitude earthquakes for the rift flanks, results in a negligible contribution to the Paducah PSHA. Other Ouaternary Faults The known faults of the study region with Quaternary offset but without demonstrable Holocene offset are listed in Table A-1 with Pleistocene as their most recent movement. All of these are short (from 2-15 km of mapped length) and are either flank faults of Crowleys ridge (Noxal, Farrenburg, Jonesboro Junction, Brookland, Bono, Valley View, Harrisburg, Weona) or small fault segments in the fluorspar zone (Lusk Creek, Hobbs Creek, Barnes Creek, Lockhart Bluff) (Fig. A-4). Based on fault length, we assign an M Of 7.0, and based on lack of Holocene f max offset, we assign a Pleistocene recurrence interval of >10 yr. Therefore, faults with this classification have a negligible contribution to the seismic hazard of the Paducah site. Other Seismic Source Zones in intraplate regions in general and stable continental regions (SCRs) in particular, the delineation of known Qu sternary faulting is an incomplete guide to where future large earthquakes will occur. This is also true of concentrations of microseismicity. Although both Quaternary faults and microseismicity are valid criteria for seismic source zone identification, in castern North America we continue to be surprised by ' rogue' earthquakes on unknown, aseismic faults (e.g., New Bmnswick, Saguenay, Sharpsburg, Ungava). This is true in other SCRs as well, especially India (Killari-Latur) and Australia (Tennant Creek). The Paducah site is certainly not exempt from the rogue earthquake problem, but it can be minimized by the use of generalized seismic source zones (GSSZs), which I define as crustal areas encompassing faulting and crustal structure with a common tectonic evolution. The tectonic basis is important, as explained below, but occasionally ) A-15

as a last resort it may be necessary to define GSSZs on the basis of seismicity patterns. Both types of criteria are used here. Since the dimensions of a GSSZ usually offer no constraint on the maximum magnitude earthquake, any possible constraints must lie elsewhere. I use the conclusions from the EPRI study oflarge earthquake potential in SCRs (Johnston et al.,1994) to provide guidance. The most important of these is that major earthquakes of M = 7 are observed only in relatively young, rifted cmstal environments, i.e., in crust such as the reactivated Reelfoot rift that has undergone significant extension during the Cenozoic to late Mesozoic. Intracratonic rifts of Paleozoic age are moderately more active than unrifted crust, and Precambrian rins' seismic activity is indistinguishable from unrifted crust. Thus, both the presence and the age of rift-extended crust are important indices of seismic potential. The stretching of the crust is accommodated primarily by faults, which form in an extensional stress regime but later may be reactivated in the compressional stress regime of SCR North America. Wabash Vallev/LaSalle GSSZ. The Wabash Valley fault system and the LaSalle deformation belt are shown in Figures A-9 and A-13b. Deformation within these grades from fault-dominated stmeture to the south in the Wabash Valley to fold-dominated (with fault cores) structure to the north in the LaSalle belt (Hamburger and Rupp,1988). The faults are imaged mainly in the Paleozoic section but are probably deep-rooted. Age of most recent movement is poorly known but is probably post-Pennsylvanian. There is no indication of any significant displacements of Quaternary units. The deformation of the Wabash/LaSalle (WVL) system is generally interpreted as rift-related, but the degree of extension is minor in comparison to the Reelfoot rift to the south. The inability to map faults from the Reelfoot rift across the Rough Creek-Shawneetown fault system, absence of a large, well-developed basement rift north of the Rough Creek-Shawneetown fault system, and the change from seismicity following discrete faults in the NMSZ to the diffuse seismicity with no apparent association with faults within and surrounding the WVL, supports our treatment of the NMSZ and the Wabash/LaSalle system as discrete seismic zones. Recent paleoliquefaction work, described in Part 1, has forced a revision in the assessment of the seismic potent al of the WVL Prior to this work the lack of known historical or instrumental i M=5.5 carthquakes and the minor degree of extension compared to the Reelfoot rift led to Mmax estimates generally less than M7 (e.g., the LLNL and EPRI studies). The two largest paleoearthquakes (M ~7.6, 7.2; Pond and Martin,1997) occurred roughly within the mapped extent of the Wabash Valley fault system. Others ranging from M6 to M7 scatter in a more wide-ranging manner through the Illinois basin of southern Illinois and Indiana (Munson et al.,1997). The paleoliquefaction constraints on magnitude, location, and time of occurrence of these events are strong enough to use in PSHA seismic source zoning. The boundaries of the WVL GSSC are uncertain except for the well-defined southern boundary at the Rough Creek-Shawneetown fault system. Since in a GSSC the M earthquake is migrated max to its closest distance from the site, the WVL southern boundary is by far the most important one for Paducah. The other boundaries can be somewhat ' fuzzy' without much effect on Paducah hazard. We therefore expand WVZ USSC beyond the mapped extent of the Wabash/LaSalle system to include the diffuse modern seismicity and all paleoseismicity of southern Illinois as shown in Figure A-22. Two alternatives for M are given in the Table A-1 summary that rely max A-16

y I heavily on the Pond and Martin (1997) paleoliquefaction magnitudes and occurrence times. We also give two alternatives for a large diffuse WVZ and a smaller GSSZ concentrated in the Wabash/LaSalle belt. Extension of the East Prairie f=h. This seismic source zone is defined on the basis of a second-order seismicity pattern emerging slowly from the regional seismic network data. Sparse seismic resection data suggest a minimum length of-35 km. Concentrated, first-order microseismicity extends NNE ~40 km. Johnston (1996), in modeling the source fault for the M7.8 J1 earthquake, extended the fault to the_ epicentral region of the M6.61895 Charleston, MO earthquake, for a total fault length of 65 km. The second-order seismicity concentration (not evident in Fig. A-2) is seen most clearly in the time-sequenced Fig. 5 of Bra 3 et al. (1997). It appears to consist of two parallel trends ~100 km long that extend NNE from the central NMSZ to within ~15 km of Paducah [ Fig. A-7 of Wheeler (1997)). Wheeler argues for a structural northern boundary of the Reelfoot rift in this region and indeed the seismicity concentration dies out just at the Illinois-Kentucky border. The faults of the Fluorspar Area Fault Complex continue on strike with the East Prairie seismicity to the NE for another -40 km (Fig. A-13). These faults are only mapped NE of the seismicity; however, their SW extent may be masked by Mississippi embayment sediments, which would allow the possibility that the Fluorspar Area Fauh Complex is the exposed NE segment of the source faults of the East Prairie GSSZ. Because ofits proximity to the site, a maximum earthquake on the East Prairie GSSZ potentially could produce the largest ground motions at Paducah of any of the seismic sources considered in this report-short of a larp background onsite earthquake. The faulting model for the 23 Jan 1812 event is given in Figs. A-20 and A-21 from Johnston (1996). Following this modeling, - which uses greater fault depths than the source scaling of Fig. A-19, the extension of fault length from 65 to 100 km yields a M8.0 maximum earthquake. Further extension to 160 km to include the Fluorspar Area Fault Complex, which is considered unlikely (weight of 0.1), yields M of max ~8.2. The type of faulting most consistent with the regional stress regime, the linearity of the zone and the current activity on the East Prairie fault is right-lateral strike slip, Directivity effects could be important because if the fault zone is loaded to failure by other NMSZ events, as probably happened in 1811-12, rupture could be expected to nucleate to the south and propagate NE toward Paducah. There is littig basis to estimate recurrence on the East Prairie GSSZ. The interval must be long (order of 10 s years) judging by 1) the sparse current seismicity, 2) lack of Holocene fault offsets in the Fluorspar Area Fault Complex and presence of only minor Quaternary faulting, and 3) lack of discernable offset of the margins of Sikeston ridge where the East Prairy fault obliquely crosses it.- Onghe basis of these points, M8 recurrence on the order of 10 yr is given double the I weight of a 10 yr ir.terval. The Southcasdeelfoot Rift Finnk. In contrast to the northwestern flank, the southeastem flank of the Reelfoot rift of-280 km extent exhibits a moderate level of contemporary seismicity (Chiu ct al.,1997). Moreover, within its extent, it contains the Holocene Crittenden County fault zone (M ~7.2*0.2, Table A-1). As max discussed under the Commerce /Benton Hills seismic source zone for the northwest flank, we discount the possibility that the entire flank could rupture in a single great earthquake. In Table A-1, we consider two alternatives: the 6rst, that the SE flank segmentation is limited to A 17

Crittenden County fault lengths of-32 km, yielding M -7.2 and assign that a weight of 0.5. mu ) The second alternative is that M could approach the magnitude of primary NMSZfaulting, max yielding M ~7.910.2. Both alternatives have a Holocene recurrence estimate of 10 yr. Since max tL: 'E flank GSSZ extends within ~70 km of the Paducah site, it could possibly have a non-nyligible efred on its hazard. The equal division of M,x between 7.2 and 7.9 in Table A-1 reflects the large uncertainty in the assessment ofits potential to generate major-to-great earthquakes. Other GSUn The Paducah site is within the generalized boundaries of the Reelfoot rifVRough Creek graben extensional crustal zones. Their boundaries, as mapped mainly from potential field data, ere shown in Figures A-9 and A-10. The seismic potential of these crustal provinces is t unknown since no specific faults have been mapped within their boundaries other than those included in Table A-1. These provinces have no signature in the present-day seismicity. Therefore we assign a generic M7.010.4 to these crustal regions. There is no indication of 4 seismic activity in the Holocene, so we assign a recurrence interval of >10 yr to thesu background GSSZ zones. Part I and Pert 2: Summary of Results We summarize our results in Table A-1. It contains a listing of all known historical, Holocene or other Quaternary faults, and of generalized seismic source zones that could contribute to the seismic hazard at the Paducah site. Known or inferred characteristics of these faults are included along with estimates ofits maximum magnitude and recurrence interval, sornetimes weighted for alternate interpretations. A-18

i References Cited Bear, G.W., Rupp, J.A., and Rudman, A.J.,1997. Seismic interpretation of the deep structure of the Wabash Valley fault system. Seism. Res. Lett.,68, n. 4,624-640. Boyd, K.F., and Schumm, S.A.,1995. Geomorphic evidence of deformation in the northern part of the New Madrid seismic zone. US Geol. Survey Prof. Paper 1538 R,35 pp. Braile, L.W., Hinze, W.J., and Keller, G.R.,1997.' New Madrid seismicity, gravity anomalies, and interpreted ancient rin structures. Seism. Res. Lett.,68, n. 4, 599-610. Braile, L.W.', Hinze, W.J., Keller, G.R., Lidiak, E.G., and Sexton, J.L.,1986. Tectonic development of the New Madrid rift complex, Mississippi embayment, North America. Tectonics, 1,225-237. Braile, L.W., Keller, G.R., Hinze, W.J., and Lidiak, E.G.,1982. An ancient rift complex and its relation to contemporary seismicity in the New Madrid seismic zone. Tectonics,1: 225-237. Bryan, E. (1816). Letter, in Dow, L., History ofCosmopolite, Joshua Martin, Wheeling, VA. Chiu, S.C., Chiu, J., and Johnston, A.C., Seismicity of the souti,castern margin of Reelfoot rift, central United States. Seismological Research Letters, in press. Chiu, J.-M., A.C.' Johnston, and Y.-T Yang,1992, Imaging the active faults of the central New Madrid seismic zone using PANDA array data,- Seismol. Res. lett., 63, 375-393. Cox, R.T.,1988a. Evidence of Quaternary ground tilting associated with the Reelfoot rift zone, northeast Arkansas. Southeastern Geol.,28: 211-224. Cox, R.T., and Van Arsdale, R.B.,1997. Hotspot origin of the Mississippi embayment and its possible impact on contemporary seismicity. Eng. Gecl.,46(3/4): 5-12. Crone, A.J., Giardino, M., and Schweig, E.S.,1995. Paleoseismic studies of the Bootheel lineament, southeastern Missouri, and the Crittenden County fault zone, northeastern Arkansas, New Madrid seismic zone, central United States.' US Geol. Sury. Misc Field Studies Map MF-2279. Crone, A.J., McKeown, F.A., Harding, S.T., Hamilton, R.M., Russ, D.P., and Zoback, M.D., 1985. Structure of the New Madrid seismic zone in southeastern Missouri and northeastern Arkansas. Geology, 13:547-550. Dart, R.L.,1995. Maps of upper Mississippi embayment Paleozoic and Precambrian rocks. U.S.G.S. Misc. Field Studies Map MF-2284. J A-19

t Ervin, C.P., and McGinnis, L.D.,1975. Reelfoot rift Reactivated precursor to the Mississippi l Embayment.- Geol. Soc Am.-Bull.,86: 1287-1295. - Hamburger, M.W. and J.A. Rupp,1988. The June 1987 southeastern Illinois earthquake: possible tectonism associated with La Salle anticlinal belt, Seism. Res. Lett., 59, n. 4,151-157. Hamilton, R.M., and Zoback, M.D.,1982. Tectonic features of the New Madrid seismic zone from seismic-reflection profiles, in Investigations of the New Madrid, Mismun, earthquake region, F. A. McKeown and L. C. Pakiser (Editors). U. S. Geol. Sure Profess. Pap. 1236: 55-82. Harrison, R.W., and Schultz, A.,1994. Strike-slip faulting at Thebes Gap, Missouri and Illinois: implications for New Madrid tectonism. Tectonics, 13:246-257. Hildenbrand, T.G.,1985. Rift structure of the northern Mississippi Embayment from the analysis ofgravity and magnetic data i Geophys. Res.,90: 12607-12622. Hildenbrand, T.G., and Ravat, D.,1997. Geophysical setting of the Wabash Valley fault system. Seism. Res. Lett., 68, n. 4, 567-585. Howe, J.R.,1985. Tectonics, sedimentation, and hydrocarbon potential of the Reelfoot aulacogen. Master's Thesis, University of Oklahoma, Norman, OK: 109 pp. Johnston, A.C. and S.J. Nava,1985. Recurrence rates and probability estimates for the New Madrid seismic zone, J. Geophys. Res., 90, 6737-6753. Johnston, A.C.,1996. Seismic moment assessment of earthquakes in stable continental regions - III. NewMadrid 1811-1812, Charleston 1886 and Lisbon 1755. Geophys. Jour. International, 126:314-344. Jshnston, A.C., K.J. Coppersmith, L.R. Kanter and C.A. Cornell (1994). The earthquakes of stable continental regions: Assessment oflarge earthquake potential, EPRI Report TR-102261, J.F. Schneider, ed., Electric Power Research Institute, Palo Alto, vol.1(4). Johnston, A.C., and Schweig, E.S.,1996. The enigma of the New Madrid earthquakes of 1811-1812. An. Rev. Earth Planet. Sci.,24: 339-384. Kelson, K.I., Simpson, G.D., Van Arsdale, R.B., Haraden, C.C., and Lettis, W.R.,1996. Multiple late Holocene earthquakes along the Reelfoot fault, central New Madrid seismic zone. J. Geophys. Res.,101, n. B3: 6151-6170. Kelson, K.I., Van Arsdale, R.B., Simpson, G.D., and Lettis, W.R.,1992. Assessment of the style and timing of surficial deformation along the central Reelfoot sca.p, Lake County, Tennessee. Scism. Res. Lett.,63, n. 3: 349-356. A-20 i

i Kolata, D.R., and Hildenbrand, T.G.,1997. Structural underpinnings and neotectonics of the l southem Illinois Basin: an overview. Seism. Res. Lett.,68, n. 4,499-510. Langenheim, V.E., and Hildenbrand, T.G.,1997. Commerce geophysical lineament Its source, geometry, and relation to the Reelfoot riR and New Madrid seismic zone. Geol. Soc. Am. Bull.,109, n. 5: 580-595. Liu, L., Zoback, M.D., and Segall, P.,1992. Rapid intraplate strain accumulation in the New Madrid seismic zone. Science,257: 1666-1669. Luzietti, E.A., Kanter, L.R., Schweig, E.S., Shedlock, K.M., and Van Arsdale, R.B.,1992. Shallow deformation along the Crittenden county fault zone near the southeastern boundary of the Reelfoot rift, northeast Arkansas. Seism. Res. Lett.,63, n. 3: 263-275 Luzietti, E.A, Kanter, L.R., Shedlock K.M., Schweig, E.S., and Van Arsdale, R.B.,1995. Shallow deformation along the Crittenden County fault zone near the southeastern margin of the Reelfoot rift, northeast Arkansas In: K.M. Shedlock and A.C. Johnston (Editors), Investigations of the New Madrid seismic zone. US Geol. Sury. Prof. Paper 1538-J,23 PP-Marple, R.T., and Schweig, E.S.,1992. Remote sensing of alluvial terrain in a humid, tectonically active setting; he New madrid seismic zone. Photo. Eng. and Remote Sensing, 58: 209-219. Marshak, S. and T. Paulsen,1997. Structural style, regional distribution, and seismic implications of midcontinent fault and-fold zones, United States, Seism. Res. Lett., 68, n. 4, 511 520. McKeown, F.A.,1982. Overview and discussion, in Investigations of the New Madrid Missouri, earthquake region, F. A. McKeown and L. C. Pakiser (Editors). U. S. Geol. Surv. Profess. Pap. 1236:1-14. Munson, P.J., Munson, C.A., Bleuer, N.K., and Labitizie, M.D.,1992. Distribution and dating - of prehistoric liquefaction in the Wabash Valley in the central U.S. Seism. Res. Lett., 63,

n. 3, 337-342, Munson, P.J., Obermeier, S.F., Munson, C.A., and Hajic, E.R.,1997. Liquefaction evidence for Holocene and latest Pleistocene seismicity in the southern halves ofIndiana and Illinois: a preliminary overview. Seism. Res. Lett., 68, n. 4, 521-536.

Nelson, W.J., Denny, F.B., Devera, J.A., Follmer, L.R., and Masters, J.M.,1997. Tertiary and Quaternary tectonic faulting in southernmost Illinois. Eng. Geol.,46, n. 3-4,235-258. Nelson, K.D., and Zhang, J.,1991. A COCORP deep reflection profile across the buried Reelfoot rift, south-central United States. Tectonophysics, 197:271-293. Obermeier, S.F., Bleuer, N.K., Munson, C.A., Munson, P.J., Martin, W.S., McW' liams, K.M., d A-21 i

Tabaczynski, D.A., Odum, J.K., Rubin, M., and Eggert, D.S.,1991. Evidence of strong earthquake shaking in the lower Wabash Valley from prehistoric liquefaction features. Science,251,1061-1063. Obermeier, S.F., Martin, J.R., Frankel, A.D., Youd, T.L., Munson, P.J., Munson, C.A., and Pond E.C.,1993. Liquefaction evidence for one or more strong Holocene earthquakes in the Wabash Valley of southern Indiana and Illinois, with a preliminary estimate of magnitude, U.S. Geol. Sury. Prof. Pap. 1536,27 pp. Palmer, J.R., Hoffman, D., Stephenson, W.J., Odum, J.K., and Williams, R.A.,1997. Shallow - seismic reflection profiles and geological structure in the Benton Hills, southeast Missouri. Eng. Gool., 46, n. 3-4: 217-233. Palmer, J.R., Shoemaker, M., Hoffman, D., Anderson, N.L., Vaughn, J.D., Harrison, R.W.,1997. Seismic evidence of Quatemary faulting in the Benton Hills area, southeast Missouri. Seism. Res. Lett., 68, n. 4, 650-661. Pond, E.C., and Martin, J.R.,1997. Estimated magnitudes and accelerations associated with prehistoric earthquakes in the Wabash Valley region of the central United States. Seism. Resc Lett., 68, n. 4, 611-623. Pujol, J., A. Johnston, J.-M, Chiu and Y.-T. Yang,1997. Refinement of thrust faulting models for the Central New Madrid seismic zone, Engineering Geology,46,281-298. Purser, J.L.,1996. Shallow seismic reflection survey along the southem margin of Reelfoot Lake, Tennessee, and regional implications. MS thesis, University of Memphis, Memphis,- Tennessee: 76 pp. Russ, D.P.,1982. Style and significance of surface deformation in the vicinity of New Madrid, Missouri, in Investigations of the New Madrid Missouri, earthquake region, F. A. McKeown and L. C. Pakiser (Editors). U. S. Geol. Sury. Profess. Pap. 1236:95-114. Russ, D.P., Stearns, R.G., and Herd, D.G.,1978. Map of explorntory trench across the Reelfoot scarp, northwestern Tennessee. U.S. Geol. Sury. Miscellaneous Field Investigations Map MF-985, r Saucier, R.T.,1974 Quatemary geology of the Lower Mississippi Valley - Research series no. 6, Arkansas Archeological Survey, Fayetteville,6: 26 pp. Saucier, R.T.,1994. Geomorphology and Quaternary geologic history of the lower Mississippi Valley. U.S. Army Corps of Engineers, Waterways Experiment Station: 364 pp. Schweig, E.S., and Ellis, M.A.,1994. Reconciling short recurrence intervals with minor deformation in the New Madrid seismic zone. Science, 264: 1308-1311. A-22

Schweig, E.S., Fan, Shen, Kanter, L.R., Luzietti, E. A., Van Arsdale, R.B., Shedlock, K.M., and King, K.W.,1992. Shallow seismic reflection survey of tic.botheel li===t area, southeastern Missouri. Seism. Res. Lett,,63, n. 3: 285-295. Schweig, E.S., and Marple, R.T.,1991. The Bootheel lineament: a possible coseismic fault of the great New Madrid canhquakes. Geology, 19: 1025-1028. Schweig, E.S., and Van Arndale, R.B.,1996. - Neotectonics of the upper Mississippi embayment. Eng. Geol., 45, n.1-4: 185-203. Sexton, J.L.,1992. Collaborative research SIUC-USGS high resolution seismic reflection surveying of the Bootheel lineament in the New Madrid seismic zone. NEHRP Final Report: 35 pp. Sexton, J.L., Henson, H., Dial P., and Shedlock, K.,1992. Mini-Sosie high resolution seismic reflection profiles along the Bootheel Lineament in the New Madrid seismic zone. Seism. Res. Lett., 63, n. 3: 297-307. Sexton, J.L., and Jones, P.B.,1986. Evidence for recurrent faulting in the New Madrid seismic zone from Mini Sosie high resolution reflection data. Geophysics,51, n. 9: 1760-1788. Sexton, J.L., and Jones, P.B.,1988. Mini-Sosie high-resolution reflection survey of the Cottonwood Grove fau*t in northwestern Tennessee Bull. Seism. Soc, Am.,78, n. 2: 838-854. Shedlock, K.M., and Harding, S.T.,1982. Mississippi River seismic survey. Geophy. Res. Lett., 9: 1275-1278. Spitz, W.J., and Schumm, S.A.,1997. Tectonic geomorphology of the Mississippi Valley between Osceola, Arkansas and Friars Point, Mississippi. Eng. Geol.,46, n. 3-4: 259-280. Stearns, R.G.,1957. Cretaceous, Paleocene, and Lower Eocene geologic history of the northern Mississippi embayment. Geol. Soc. Am. Bull.,68: 1077-1100. Stearns, R.G., and Marcher, M.V,1%2. Late Cretaceous and subsequent structural development of the northern Mississippi embayment area. Geol. Soc. Am. Bull.,73: 1387-1394. ) Stephenson, W.J., Shedlock, K.M., and Odum, J.K.,1995. Characterization of the Cottonwood - Grove and Ridgely faults near Reelfoot Lake, Tennessee, from high-resolution seismic reflection data. U. S. Geol. Sury. Profess. Pap. 1538-1:1-10. Suppe, J.,1985. Principles of structural geology. Prentice-Hr.ll, Inc., New Jersey, 537 pp. iuttle, M.P., and Schweig, E.S.,1995. Archeo'ogical and pedological evidence for large prehistoric earthquakes in the New Madrid seismic zone, central United States. Geology, 23:253-256. A-23

Tuttle, M.P., Lafferty, R.H., Guccione, M.J., Lopinot, N., Cande, R.F., Dyer-Williams, K., and Haynes, M.,1996. Use of archaeology to date liquefaction features and seismic events in the New Madrid seismic zone, central United States. Geoarchaeology,11, n. 6: 451-480. Tuttle, M.P., and Schweig, E.S.,1996. Recognizing and dating prehistoric liquefaction features: lessom Seamed in the New Madrid seismic zone, central United States. J. Geoph. Res., 101, n. B3: 6171-6178. Van Arsdale, R.B.,1997. Hazard in the heartland: the New Madrid seismic zone. Geotimes,42,

n. 5: 16-19.

Van Arsdale, R.B., Guccione, M., Stephenson, W., and Odum, J.,1996. Collaborative research - University of Memphis and University of Arkansas: Mini-Sosie seismic reflection surveys at Reelfoot Lake, Tennessee, and Big Lake, Arkansas. U.S.G.S. NEHRP Final Report: 25 PP. Van Arsdale, R.B., Kelson, K.I., and Lumsden, C.H.,1995b. Northern extension of the Tennessee Reclfoot scarp into Kentucky and Missouri. Seism. Res. Lett., 66, n. 5: 57-62. Van Arsdale, R.B., Williams, R. A., Schweig, E.S., Shedlock, K.M., Kanter, L.R., and King, K.W.,1994. Preliminary seismic reflection study of Crowley's Ridge, northeast Arkansas. U.S. Geol. Surv. Profess. Pap.1538-C: 1-16. Van Arsdale, R.B., Williams, R.A., Schweig, E.S., Shedlock, E.S., Odum, J.K., and King, K.W., 1995a. The origin of Crowley's Ridge, northeastern Arkansas: erosional remnant or tectonic uplift? Bull. Seism. Soc. Am.,85, n. 4: %3-985. Vaughn, J.D.,1991. Evidence for multiple generations of seismically induced liquefaction features in the Western Lowlands, southeast Missouri. Seism. Res. Lett, 62: 189. Vaughn, J.D.,1992. Active tectonics in the Western Lowlands of southeast Missouri: an initial assessment. In: L. Unfer, Jr. (Editor), Conference on the geolgoy of the mid-Mississippi Valley, Southeast Missouri State University, Cape Girardeau, Missouri. MO Dept. Nat. Res., Div. Geol. and Land Survey, Spec. Publ. n. 8, pp. 54-59. Vaughn, J.D., Hoffman, D., and Palmer, J.R.,1993. A late-Holocene surficial sandblow in the Westem Lowlands of southeast Missouri. Geol. Soc. Am.,25, n. 3: 57. Weber, J., S. Stein and J. Engeln,1997. Estimation ofIntraplate strain accumulation in the New Madrid seismic zone from repeat GPS surveys, submitted to Tectonics. Wheeler, R.L.,1997. Boundary separating the seismically active Reelfoot rift from the sparsely seismic Rough Creek graben, Kentucky and Illinois. Seism. Res. Lett., 68, n. 4, 586-598. A-24 I J

Williams, R.A., Luzietti, E.A., and Carver, D.L.,1995. High-resolution seismic imaging of Quaternary faulting on the Crittenden County fault zone, New Madrid seismic zone, northeastern Arkansas Seism. Res. Lett.,66, n. 3: 42-57. Woolery, E.W., Wang, Z., Street, R.L., and Harris, J.B.,1996. A P and SH-wave seismic reflection investigation of tbc Kentucky Bend scarp in the New Madrid seismic zone. Seism. Res. Lett., 67, n. 2: 67-74. Zoback, M.D.,1979. Recurrent faulting in the vicinity of Reelfoot Lake, northwestern Tennessee. Geol. Soc. Am. Bull., Part 1, 90: 1019-1024. d A-25 l

A.-. 1 Tcble A-1: Sumatry of Conclusions Fault Most Recent Fault Inferred Type of Maximum Recunence Weight Name [1811-1812 earthquake] Movement Length Fauhing Magnitude (M) Interval FestPrairie [J1] 1811-1812 65 km RL strike slip '7.810.3 10' yr - 0.70 7.810.3 500 yr 0.15 7.510.2 500 yr 0.15 Blytheville arch /CGF [Dl] 1811-1812 125 km RL strike slip 8.110.3 10' yr 0.70 8.110.3 500 yr 0.15 7.710.2 500 yr 0.15 Bootheellineament [D1] 1811-1812 135 km RL strike slip 8.110.3 10' yr 0.70 8.110.3 500 yr 0.15 7.710.2 500 yr 0.15 Reelfoot [Fl] 1811-1812 52-72 km thrust 8.010.3 10' yr 0.70 8.010.3 500 yr 0.15 7.610.2 500 yr 0.15 g Ridgely Holocene > 20 km RLSS/ oblique? 7.5 (?)i0.5 Hol.[10' yr] 1.0 Crittenden County Holocene 32 km RLSS/ oblique? 7.210.2 Hol.[10' yr] 1.0 Commerce (Benton Hills) Holocene > 20 km RLSS/ oblique? 7.010.5 Hol.[10' yr] 0.90 8.010.2 Hol.[10* yr] 0.10 Noral Pleistocene > 15 km react. as reverse 57.0 Pl. [>10' yr] 1.0 Farrenburg Pleistocene unknown react. as reverse l l JonseboroJunction Pleistocene unknown RLSS7 l l Brookland Pleistocene > 10 km react. as reverse l l Bono Pleistocene > 6 km react. as reverse l l Valley View Pleistocene > 8 km react. as reverse? l l Harrisburg Pleistocene >5km react. asreverse [ [ Faults that have been assigned informal names for this report are in italics. IIol. Holocene; Pl. Pleistocene

Tchle A-1: Srmm:ry cf Ccnclusio2s (continued) Fault Most Recent Fault Inferred Type of Maximum Recurrence Weight Name [18;l-1812 earthquake] Movement Length Faulting Magnitude (M) Interval Weona Pleistocene >2km react. as reverse $7.0 Pl. [>10' yr] 1.0 l Lusk Creek (fluorspar zone) P8eistocene >1km dip-slip? l l l Ilobbs Creek (fluorspar zone) Pleistocene >3km dip-slip? l l l Barnes Creek (fluorspar zone) Pleistocene >2km dip-slip? l l l Lockhart Bluff (fluorspar zone) Pleistocene unknown dip-slip? l l l Generalized Seismic Source zones (GSSZ) East Prairie Fault exten-icn 1811-1812 65 km see East Prairiefault (see Fig. 23) Pleistocene 100 km RL strike slip 8.010.3 10' 37 0.3 Pleistocene 100 km RL. strike dip 8.010.3 10' yr 0.6 Pleistocene 169 km RL strike slip 8.210.3 10'yr 0.1 SE Flank of Reelfoot Rift IIolocene ~280 km RLSS/ oblique 8.010.3 10'yr 0.5 3 8.010.3 10'yr 0.5 6 (see Fig. 23) Wabash Valley /LaSalle belt IIolocene large zone 8.010.3 5x10' yr 0.2 w (see Fig. 23) large zone 8.010.3 10*yr 0.1 IIolocene small zone - - - - - - - - - - - - - - - 8.010.3 5x 10' yr 0.5 small zone 8.010.3 10*yr 0.2 Commerce fault extension IIolocene >20 km see Comraerce (Benton IIllis) fault (see Fig. 23) Pleistocene ~500 km RLSS/ oblique? 8.010.3 10'yr 0.85 >8.0 >10' yr 0.15

Central US Seismicity: 1974 - 1994 -92* -90' -88* -86* 40' ' 40' Missourl Illinois Indiana 8 e e e*e 3 e . f.. p,. ,e*, e ,4. *. 9. *.'.

• g s(

.' ens 38-38- .. y. ..r e

"e : * * /

4. e.....

g.... -

e .e..g.. f.. p Kentucky .'op t.,, ..s. '..y,,. w., c Tennessee

. 4 ;A

.e

.J.

g.s '., y-f'4 i%, 36' 36' ..ee [Q,.e. Nill f-e_ MEMPIIIS 0 50 lb0 e M>5 Mississippi ,e 5>M24 ,. Arkansas 4>M23 3>M22 Alabama y < ~3 34'. 34' -92* 3' -88* -86* Figure A-1. Location of the New Madrid and Wabash Valley seismic zones with respect to Paducah, Kentucky. A 28

37-w

• }-

~~ I 9 4 g

• ' +

+7 4. 4 N[ g .g 4, pt, ,w.. ...-.-./.- 4 4 % +, +, * * + .\\ C+'9 W Ky_ uo g ~ ( 4 Reelfoot ,y+ *.uue r, ( -., p , f' r* +e + j .*4+ s .4 '~ .. s, ,r' 4~., + #-

• +

Sq, y ,./ Tt + .g , $I *,, ^ f* t,Opye,rsbu + +

----

fN;-{^*

~ ~ f 36- ~ ' h2P ' + ) p4 . 4 + /.' ~., D d [ ~[ +* +,

• . tarked Tree

(. \\ 4 + g (.P - Mississippi + .- d River p s.\\ m I N O 50km 0 0 ( .g Memphis 91 90 89 Figure A-2. Microscismicity of the New Madrid seismic zone indicated with crosses. The open circles represent the locations of the 3 major shocks of 1811 and 1812. A.29

Mississippi Valley .Il graben faults l MSZ . raducIh k' KY-MO'

• \\

f TN [ ^R , d_. [ ',' , '. - 1 ' Margin of the. ' Mississippi Em'a ,, <,3 .^* nampm. 7.Y Y Y, 'Oi k' q#c4, \\ 'o 4 DESHA

1. e4

\\ j BASIN k s ) '% N-MONROE // UPLIFT d SABINE UPLIFT MS AL N w m Figure A-3. The New Madrid seismic zone (NMSZ) within the upper Mississippi embayment. From Cox and Van Arsdale,1997. A-30 1

heb s' Gap.. Faults with Holocene displacement ' '. Bento.n Hills 4 Faults with Pleistocene o displacement o 37 - I* ,.. f ,,v 'udsc lj 1, Sikeston .a:. :=. w._., -

.c

..F E g} .o. w: k 8* _s _f.NO KY g/... TN [i AR / J p e app; s ~ l . ~e. g O e' fr 8 %wburg 36 - E

v. :

// JonessoM:.; ,g m 4 7g g 1 .B P. jw =lL n N i ,y a p, ' '- 4 I E Memphis A TN i"2 0 T 35 - MS o 92 91 90 89 88 Scale t_- numm c ammmmc - > O 100 km Figure A.4. Quaternary fault map of the upper Mississippi einbayment. Faults are dotted where ] displacement is probable. BL=Bootheellineament, RF=Reclfoot fault, SFR=St. Francis River, MR= Mississippi River, CCRZ=Crittenden County fault zone, CGF= Cottonwood Grove fault, RigF=Ridgely fault, CF= Commerce fault. The following are fault names are introduced in this report: H=Harrisburg, B= Bono, V= Valley View, Br=Brookland, W=Weona, J=Jonesboro Junction, E= East Prairie, N=Noxal, F=Farrenburg. A-31

f 5km ~ N g,New Madrid %g.' A

,. w, i;

7 MO

.? ';*\\ ll C; / R 2? l $ gl / gy 2..____. KY w f 36,30' 4, I y Lake County .. TN

New Mu %

Uplift ', j, j Fault d f ~ W k [f 'x3{^ / 1 ~~ ~ / Reelfoot A' N Portagevilje I 4/ k. QY+kJ$~ Tiptonvilje 4w hL.ie_ . SRI E ' A ,%F4q y v. t p. A8rit.Khy O ,,, Tp +y ;,-- ...w..., ~%, /. / ~ t,. KINICE* wood 1~ ,psissiPPi Rit. Greve town l p g\\ y a . >~'- c* t, y L....*.,[ 'y ' M g <r g?' Temaw g . O Memphts 9 50 km 's gge. Figure A.5. Location of the Lake County uplift boundaries, and interpreted fauh traces. Dashed fauh lines are possible projections of the Cottonwood Grove and Ridgely fauhs. Mooified from Purser,1996. A.32

auth calicat 1.cks Higroted Tine Ssctica, Ceiwerted to Deptli E .:..e .u n ...n e in .i n ..,n .. i n v.n n on -n a rne sasu s e.n..r u ..in .-n. i vin e..a ... i n sein i,.a i n <n.:nc ...-....y..:.:.=. -_ ;. s.>.:::-c. := -: ---. - c-. - - c _.g:; A... - -........A.j.- -- ---a-c:: .e.r_.;.%g ~:1C:. ---- ~w,.- - m &__ '. f.4~. .F-- ....w- .w::;'.. 'c : - ----- Y ' ; 3::==3..:.... M. _. r =. - .~.-..-_._y,-- . 5,4 -..,,.-~,..;...- M-r :' 'A gn--;., - :- -- - _c.,C. _... .. ~ ., --"....:,,;..,; s<...g,. .-4

-y...

. _..c :.,.. :.,_::::.. ::;-- : e ,a a. .sa-

... e ~.-a

e..m.y.-

.:....,...-. ;; r.

f.....,

._;.;p....-,..-- ,: a . :.e ;. ..3 -:5.. ;., ____ ..,. 4...d..:.. ,._= :r-:_, 3 .. e, y.,._. ~..-. ~W a . =.- - ;;,y..s w,. .....,..,a..

,:.:~; g:,.

e -::~: 3,...y m...,,.. .a.. S., :/ 4;f: ) :-... -..w 3 ,,,,g,-- U-fi;b.i,a'_'6 : ;M...-....,..~... ::. =,;-..~~-::...,;. .x-.. =r : Q-.. l:n%. '.. _ -Wz._: =.r.-i. ~.J -I.N._,,_~ '~;,;h; I ;-@,.i@M.. -,

9..

-:- _..... ;+ . _ ' ~. - - -..m .a..: ~.....:e.~~--- =..-.- .i22'. . :J 1:=~sV @-..$.Q. I.-. h5.T.c -.~.-::;e.s u.. ....~..~...:--..-.~.=a.-_ .. ~ -. ~e. :..:. T...... :..:: ~ - --.u.. ~.. = ~ ~ - = - - -... _. .~...w-

., =:

e. .-=.a.- ~. :. ./ :s. -...... .. -; -> - a.c,..-,.~:. - --.:.--....=.-=~.:- +....::-._, --..---.--.,y-.:-

- :~.~~-

... :.%.... v..>!..~ ~. :. ....g.-:~ ... --a.,~ i'.'; &;..-.: ~-. p.*:c,. ~s ... n.... - : -.. m.. --. .ua ^ :. -: w:....>-a+: --- - - .....~-... ~ - . ~... ~ ~ ~ '

,yz,

- :.~ :..: '~~ s=:: . - - -. ~ ~ - - . w^.. " : .... ~..-:. '-. -~~: .J, ". ^p:. 3 ::.... _, ;., : ; ~~ ..' ^L. . ~,,, ~ y..;. ~.,-...-,'.. T. T:" ~.:. -- y "* ?.- .... r'.i ., ~..

• s

..T.--.*.g~*-- \\..- ~- :. w..-- ~;-. . s ~~,.. :. ,.. y -~:.~.a . - - ~~ . - - - ~ ~ - - L.; ~' p.-;Q. j.J',Z.. .- M. s.., - -, [ ;.',, -- '::9 J 2.- -v. ;7 -i- ~- ~. '- '.' '.D..[~.~'. .e. #.......,-..:,,.,.,z.,.,...~.~.......>< ... ~ r., '..~.*.s.~.w)~ - ~ -.. g<..:..W. - -.r,.. t .v..'..s. .s . : ::.s';;.r..K. %...,,,.'.; T.':... A.= -~. -.__.: : :: C:.: " M. :.. ~ '.x:.&..w~,~.. e ~. y- .Q~. ~ ~ -> ~. : 2.,,.,.,.... -. : ~.,,... :. n.. ..-.. - ~.-- - - - - -. - +. - -. ~. c-

~,~.

. ~.,.....--:-..,

... -. ~.. ~,, e... nau.. .n =.. O CDP 278 358 438 518 598 678 758 838 918 998 1078 1158 1238 1318 1398 1478 1558 1638 1718 1798 1878 CDP t i I I I f f f I f f I I I I I l I I O O r 1 Tc s s 'RFZ CGFZ l Ridgely s a pz s I I -~. Tw ~~ 8 e i Tw I f a l ?- f n Tp 5(x) ~~'-- / 500 Tp I g g l e -=t 8 K ~ K i /^ I I t J ) I I I f A. ~ [L PL s, - = = * * - ' ' ~ q I 9(XI 900 Figure A-6. The 7.5 km long Reelfoot Lake Mini-Sosie (SRL) reflection profile. Vertical exaggeration is 2.7. The vertical axis is in meters. RFZ=Reelfoot fault zone, CGF= Cottonwood Grove fault, TC= Tertiary Claiborne, Tw= Tertiary Wilcox, Tp= Tertiary-Porters Creek, K= Cretaceous, Pz= Paleozoic. See Figure 5 for location. From Van Arsdale et al.,1996.

A A A' trU(Dam) m (pata) - Rs trum ) TD h> l 1 _! _ __ l -- l Surface i t I i 6 { 1ahn Couney ', UpWt l n 's E i K ~ r, 0 i i bm 0 B s-Wamers baitof m. .ty Easerra hmat of macnunnemsney ~ A tru a==> pm==> as A' 0-j Ground Surface i t is 's 4', qs, u ~ yn Pc 5- -(,,'g s I i 10 - i 0 $ lan \\ is - Figure A-7. Northeast oriented cross section of the Lake County uplift. The western margin of the Lake County uplift and Tiptonville dome are ma;ked as defined by Russ (1982) and in our data. The dashed lines are the proposed kink bands. K= top of the Cretaceous, Pz= top of the Paleozoic, RS=Recifoot scarp, RF=Reelfoot fault. Venical exaggeration is 15X in the A figure thus making the Reelfoot fault and kink bands rearly vertical. Figure B is cross section A-A' using the fault-bend fold model of Suppe (1985). Note there is no vertical exaggeration in B. K= top of Cretaceous, Pz= top of Paleozoic, Pe= top of Precambrian, LCU= Lake County uplift western margin (Russ,1982), TD=Tiptcaville dome western margin (Russ,1982), RS=Reelfoot scarp. See Figure 5 for location. From Purser,1996. A 34 { l

N S 2X0_ 14 W S K 03 f4F3 C1 &l8 B1 Y3 S2 B3 B4 65 T + 1300 _ asam ' ' a. + -e ass - m-B C e .9 3 A Nu *5 -O 33co -,.n_.-rts liihm a _.c d - ~ xm m s "M* Tm. z rc, . t= " GUr - w v a 8 m.#: q ~T + Y v m 'pg u. -<a z a n,, y, - - u.c = y - +. C C* c T3 s-E 2 C o e 5 E [ T$ 5g N-2 a s. << n c. l c 1200-3 . C. C 3 o i d 3 3 2 10C0 - o .o k ( m i-m i u ) 1 l 6 ea en o O t l6' M M iiE 300 _ m m E E too _ o o c o .2 o 0-o o l 5 ? 9 m .C 2 o c'- o sc. O g pselessed 9 M9 eof event age possa.de esent ages - tallevent i i 4040 B.C. 3340 BC Figure A-8. Sumnury of paleocarthquake investigations in the New Madrid seismic zone. Vertical bars and lines show resuhs from individual sites or studies, generally arranged from north on the left to south on the right. Thick horizontallines at top represents the 1811-1812 sequence. Horizontal shaded bands represent likely age r;.nges of earthquakes in the past 2000 years, vith darker shadine renrewnrica the mare likelv dee~ um ;-r-. :. o r.i -- - :... -- -- --- x

90 88' 86* 40' O y.o ') i I / l o Indianapolis e 3 0 ( RlW - 1 O S.uwe e a ,3 l$ 'o) o o \\ .I o f o a s. 0 M' e.,% Louis t ( +o 4 4. 09 o @k.($ e -Mw7.o g, 1 "o O "g *#ey

$ a o

o o O of M o gg C)hol5 d) IL ,bN p Jg1 %o Mw$.5 g, P j o O h, o o. g o, o e, D k s) Y Fra colsh' (*'%,/ oS / 9, ftns. o ault System.. ~ ~ ~ o ? ,g k .s o 'o S 3 oo h gy . w9tadri e,-.._.g,..-...._..-..-.._.. l a 3 s , a,,,, 60 .. e.. _ _ .a

• [

- cv O o 50 100 150 KM ,, f - o >. rs u Figure A 9. Historic earthquake epicenters in the southern Illinois Basin and vicinity from 1700 1991 are indicated in blue and yellow. The red circles indicate prehistoric canhquakes determined from paleoliquefaction studies. From Hildenbrand and Kolata,1997, l l A.36

i 4 ,et u -4 FA FIEL ~ tg,..,9 ggAstN O f , i t_ SPARTA A j j [ ]

• -Q

/ O J .,T F 1 e = j 8h* t ,l 1^ t \\ se' - ) (,' [ \\ -"" 4 9, $2. m he Nr.h, / - g g,,, g 'V 4EM 7 DOME /g.0-gg y-s s ' -r a - s. ^ g',- ,[ 'h,' / utssist orf imm y. \\ y . f.F. I g \\, , '(,' i I d New Ma*W No I 'f A J ,/ N, Jy t .uo. m ~ ~ - ' Seesmic Zone ),, cannam o W. (l. . o 50 t 0o.. yo gy 4 y w w w er Figure A-10. Major structural features in the southern part of the Illinois basin (inset). The rift system, including the Reelfoot rift and Rough Creek graben, is shaded. Dashed line is the northern margin of the Mississippi embayment of the Gulf Coastal plain. TB = Taff fault. From Kolata and Hildenbrand,1997. A 37

1 y;. _ _ 77,. 1o y. ,_ y, '4Q l L + '" ^ + h i+:::rs. 40 )j ' el' C IN'

gy 0

9Stkop..[+ m a < g% p 4 +$ 39 '+

• 3, g6

..g. *. 2 s. ..v. ' IN k i x-a E j.' J M OT,,, q i, n -20 g.) ',., .f' 38 7:. ,~ -30 w... . t'%. 37 KY -40 ' j# 4. sy p-y ?'q..i ,,....m ! 100 km _ -50 T-36o + TN. 4W mb mb mb= -60 ( en lip

• mb 4j 35

... a.- i'M t -70 92 -91 -90 -89 -88 -87 Gravity (mGal) Figure A-11. Bouger gravity anomaly map and 1974 94 earthquake epicenters in the New Madnd seismic zone. Dashed lines show location of the New Madrid rift complex. From Braile :t al.,1997. A 38

I a

1. 2.==:"=.2.?

':n., em ~ erw t* 5%~ v. u,,,,,,,,,, c-eep I , + - . nOUGH Cn'E=~K appggy srwy e ..e \\ / ,s CUS ee* y J* Ni n f s>ef'*y% o a un w ** d,,, n*, f a c 4 I4 _,j I k -~ p =r e* ,,/ unnamed fault system of twson (19911 m e l 0 50 km l i CitTAtt003 AND rJimvery itETIAW STRATA 0F tt w sset IlL881881PPIiutA11NT ww EXPLANAT1oN FeJte crinrier ten W W Te en sonnevoan mes / PeJes tanger ytan M trn. Td en soughnpan esse @ 5",n%1':Osll*O '*"* "* """ se= = =- i o - - ~ - - n.e benet si teJte er leult evetems W.ser ten M tm. me===:r.ev. --- M 1'*f7 % "::L % C iF i Figure A 12. Ends and bends oflong faults in the western end of the Rough Creek graben are interpreted to mark the northeastem termination of Reelfoot rift seismogenic faults. From Wheeler,1997. A 39

RANDOVH _. reverse fault. Sawteeth on upthrown side stnke-shp fault N W LTON

• M normal fault, ball and bar on downthrown side a

earthquake epicenter mentioned in text e study area mentioned in text / ~ 0 20 mi l Hamsburg,W I

1. S g

0 nw JACKSON WWAMSON, y, w._[/p/,,/'[j/ gg;' union JOHNSON POPE \\{,g '. u- -N hc i ico ecr k s %)%k LEXANDER T, PULASKI 7 Foot Creek / e Tamms + T D -y' .-87 asw Figure A 13a. Map of southern Illinois showing selected faults. From Nelson et al.,1997. A-40

i (a) N fault or fold trace (b) C ""* """ " 8 "* pum River _ edge Of Illinol5 \\ assu.uns(p) " g e.a.

u._ _._

.4 \\ k I' l r .\\ id \\ N % nwou l\\ c,,,, usaw s.n l t J ,).s/ f'J. s, %s+p1 'l )- h ' su e // N usuwwa h, ( Mi f wvrz L L f ,(

• k, umom f. uE Q:.t j

/ [:idE W f IN N Figure A-13b. Map of the principal faults and folds in the Illinois Basin, highlighting the LaSalle Deformation belt as a low-strain rift zone. (from Marshak and Paulsen,1997) } A-41

eras-i juan uwe ,,g; V u.w cooxes.. Os a ,,*Q,- ^ uwo : e

g.

13: V e.y }.e.

• T*

.{$ ts

u e.

. 0'. .s .3

/.' M'OT

..e. 's7 'D[ 12 l Oal l Ho6ocen. skmum / .4 I oi I pi mc.n. c.. -:rire ot.

3. j.

. uwe I onc ! N coAnem sana os hi,- !e M j.ay, j uo,no, a,,, ,6;,, N gj ucs.ry-wa, 7 [ I uwe i west rua.n sanoemn. os g ' Ot ' IL uordMg M o a.s = 8metcasa*dwa*'* m a ano con.o wn.r. cone o c.s i== ,,io ..no,e % e noruonesb.amng Figure A 14a. Map of the New Columbia structure associated with the Lusk Creek fault zone. From Nelson et al.,1997. Northwest Southeast New Columes Cnc Cal g7 M CA A h ^ Mw%e Fault Zone Cache Vanov g. .g um _x Mwo ~ [ _M," -~- Mar g 1!L older Masastppian su j h dder Mhwe ab \\ A T7 g i I cm i Hoocen anuoum ? s 1 mi .5 1 1.5 km ( oTmJ Mouncs Graw (!bt McNaq Formanon l uwo l west ano n sanasen.

- uer 2 co.n.y. aiun ana n.n=et um en.

Figure A-14b. Cross-section of the New Columbia structure and Lusk Creek fault. From Nelson et al.,1997. A-42

.-...-e..._ ,so - s_ u__ .. _- 3..-... -. uin. - . w- . ot - Pie.siocene iem.oe. .o . u m,,,,, .ao Osf - f~ b 7 .;) 7 ]. ~ . -I. c .... >:i. -:1, ;

(

um Tar Spnngs to Herthnstnsg ..s . e.

s. Km '.... -

f. :.

. :.s - Wum West Benen Sanostane - Km'. .t. A cashed m M s.

e. - - - :e :.-.

en1 Ooned whom concwed r stnke and dip of beachng J. 10 O horunntN boodng ,.; 5 *:jigy+:. -:.-,..'.. QTm-e 0..oT.m. soo 'f.- 0 weser wet, e6evehon of top of Peteazac .?. +.9' *,

• ./..,,**

.. * ".a."... ". s'.:. beorom n teet t.o .,,y 3....l <t30 bearocuac twow e6eveson of 130 feet . s.

'..,

i:si:. y

.lf.,....l,.* ;. ;:.;s4:::: .c..... ,.,.. - -: K. r.n.. QTm. 3eo,, -eeg,..p,. f,.355 ", '. "/. " ' ' el:'. 9- ~#--

• . *. *.. ".......c.g.,

,. 6 7 -:.: -

• O.**.*.*.*.*

<.,- n. c.. - ......%.', s. : s. v.s. .e, /%. ...... c..-s...s......<.... + 1 : e. s ...u.... s

.......'..; Js.

. :.. e. s... e

m.,..

.:.... :. e .-e.. .. J *,,. : CF. ..:.;.q.,e :<. -:p. :3,- .%. s ....,.......... e....%.: :., b:. ? . c.-:.

• • svu e.*.*..sie.99.-:.0. ;.

+.D:........-:- ....... :.: 3g ,\\ ?. . : :.T.. c.. ..e - - .: Km ;O '.. .,,,u-:,.. ff. "<.. 1 s -: - :-:..: .s.. .,:=.. ........ n.. ... ~. ..r ...s e..............,...... gym.,.s.,..... np ,,. s-::.. ...;(g:.... :.... :l l<.:.:g : ..: g.:... ... e. me. %..... c. Ogl o e. p., g- ..,.,....:+:+ x.... M3. ye.. : ,...:...x... 4.:+..x..;,.:+:x.+:+:+. +......... ay. o '.. :."." 5p: g,,ynnn.y, gu.,.g . m::. ....s.. , p qi,g ; ;.<. ..Q.. =

g......,.

- '.:. Km.::.. ........O. p.:.-. u.. . n.. ......u. .J. :...),.....:. e:.g m. ,.. :.:3. ; :. s., .. ce.. .y-L. :... if," " " " " "oi " " " " " " "i.i .? no y , o o - Qt ...... '. e c.: o a.5 1 1.$ km.- p ,.,,. s;;. s.. p. . 3,.ss... .............a....:.... n. O .:....... ~. I ersw Figure A 15a. Map of the Massac Creek structure (Hobbs Creek fauh zone). Mounds Gravelis now believed to be Metropolis gravel (John Nelson, personal communication). From Nelson et al.,1997. A 43

l Nortnwert Southeast Weaver ,OTm QTm borence Cal M _ _ _ _ <m._ h _ _ _ _it , 'i w,- ~ < - - - m - - - - /A Mc-r vm - 7 Mississippian be6ow Mdr ' ] _ Km.h + Mw0 '. * * ^ T-EEtol older Missastppian rocks 1iL i l C. I soiocene a- ? ,5 im' ['OTm j Mounos Gravel h j 3g l

• QTe'l Quatemary and upper Ternary. undifferensated V ? O W f******

(~8dnl Tar Sonngs e Hanansourg

• - Mg-i Gonconda Formanon West Baoen Sawsene

Mar.: oo.neys suti and noneuit umesiano Figure A 15b. Cross section of the Massac Creek structure (Hobbs Creek fault zone). Mounds Gravel is now believed to be Metropolis gravel Gohn Nelson, personal communication).

From Nelson et al.,1997. A-44

l l l .mcz. son. POPE

We

~ .i' O }h. h .2** _ y yy E p?) m* HDmb*r0 ?* ' a f h'? f h l t r e e v g:. m'" Romeeorig.

• i k

g[ bororio/so Weever une Hit 1 .g g ses cny - w,r y .+ sne

~ sne n

f4 k 4 y s 6 i. = usam crew I er4r erry Figure A 16a. Map showing Barnes Creek fault mne, Massac Creek structure, Reineking Hill structure and Rock Creek graben. Significant outcrops and drillholes are indicated. From Nelson et al.,1997. BARNES CREEK. SITE A g w . '.. ' ' ' 5,D ' '. ' y ',,......... _ me m.,..,.,c 7 L",Y,',W h !,'""*"Y Y M M' 1 M *'e*rm**** # o s son o 1 2 3m Figure A-16b. Sketch of structure exposed in south bank of Barnes creek at Site A of Figure A-16a. From Nelson et al.,1997. A-45

5 l j= Io nese i m-y/

g f'*

s.sooerm, y ep 's isoo to e soo y ea t 5~ g / - ~ ~ ~ h. q r;W9plJ]aL s 'll*,a 'Q.i. l 5 ~~ i= ~~ I ~ / [ k y) .g ' *50 50 y er v 1,, yv_.J. cA,, '.,2.- a j y / .~. s, <_. / o T - ov e ^~ ~ f ~ Q, =,J i ^ op i ** ro-n, j sA W **oc6 x DF r ,f? 6 e .mf y% 3c; _1 \\ -./ o n sa 7s son <w s ,a O STUDY AREA p-r-_.~.v.o~ -- ~ ~ ' ~ '~~ [ "~~' h g =.= = = =.= =- 1 3 Y f' O 0L

• C C L" ~~.~~~C.~'
• .-e

-.: =.===~_

fr

[ l Figure A-17. Overvie v showing locations of paleoliquefaction sites (solid circles) in southern Indiana and Illinois. Maximum dike width at a site is indicated by diameter of circle. Dashed lines encompass the area ofliquefaction effects of the four largest carthquakes in Indiana, with the associated liquefaction sites of each earthquake indicated by the ulor of the circles: orange = Skelton-Mt. Carmel earthquake (12,00011.000 yr BP); red = Vincennes-Bridgeport carthquake (6,1001200 yr BP); blue = Valonia carthquake (3,950i250 yr BP); and green = Martinsville-Waverly earthquake (3,800 to 8,500 yr BP). Black circles are liquefaction sites of other or unassigned earthquakes. Shaded area shows region of shallow bedrock with limited exposures of liquefiable sediments. From Munson et al,1997.

(a) . 4 ' ~,.

• t

~

• i* N

.~,.k ; NN (60 km) ( m, 1 N , ~ Mad rid t,,. Nw(40 kmg RF J (32. 3) ,.z. 'y. '* '2 sr + , [.E ** ~ et (70 km) RS (35 kr +< u

utti, e

rr,s, .y ,s t,. N -,rJ DFZ (55 km) .a

s. -

r .m SA (70 km) m 0 50 km S*1 I ( S#2 S S#3 f J 4 ge r I 1 O$ G 't I O 70 70 Os Os Q 7j 7 / 1 I oi I a..w I.+w. ' * * ' ' ' J1 gg P1 m 4 Figure A-18. (a) The principal fault segments of NMSZ (from Fig. A.9 of Johnston and Schweig, 1996) as outlined by microseismicity, the Blytheville arch and the Bootheellineament. (b) Three faulting scenarios for 1811 12. The authors' favored S#1 new evidence from the Obion River valley (see text) lead to a preference for the S#1 scenario for the D1 and J1 events but S#2 for the F1 event. A-47

AVERAGE STABLE CONTINENTAL EARTHQUAKE SOURCE PAPAMEERS BASED ON CONSTANT STRESS DROP SCAUNG Arch C. Johnston, CER1, Memphis State Universty, Memphis, TN 38152 Sufficient good quality data are now available from 4.0 < Mw < 6.8 intraplate earth-c uakes [iutraplate in the most restrictive sense, i.e., stable continental regions (SCR)] ttat everage source properties can be s active continental or oceanic regions. pecified without using intraplate events from I consider constant stress drop scaling most appropriate for this, based on the work of Somerville [EPRI Tech. Rpt. NP-4789, 1986) and Johnston (EPRI Tech. Rpt. TR-102261,1993). The appropriate average SCR (Brune) stress drop (do) is very sensitive to the proper relation between comer frequencyfo (or period To) and duration of he source ume function fd. For example, t Somerville (above ref.) takes fdilifo*1, yielding da = 82 bars; Hernnann and Goetz [BSSA 71,1981] derive fd = (2hrlfo 1 yielding da = 22 bars. For the average SCR earthquake, I take the average of this range, da = 52 bars, which yields a strain drop ac = d/W =1.58x10-4 (for rigidity = 3.3x1011 dyn/cm2), a high but not unreasonable value (cf, Kanamori and Anderson [BSSA 65,1975]). The source scaling below is based on the relation log (fd) = -7.88 4.3331og(Mo), Mo in dyn-cm. derived from log (Mo)-log (Td) pairs of 40 SCR earthquakes in Jonnston (above ref.). The average isoseismal radii forfelt and MM/ VI areas are from the same reference, loeMo M, leneth L width W slio d w f4T, feltradius Ivlradius 2 16.00 0.0 5.8 m 5.8 m 1mm 0.0028 s 303 Hz 17.50 1.0 18 m 18 m 3mm 0.0089 s 95 Hz 19.00 2.0 58 m 58 m 9mm 0.028 s 30 Hz 20.50 3.0 183 m 183 m 2.9 cm 0.088s 9.6 Hz 25 km 21.25 3.5 325 m 325 m 5.1 cm 0.16 s 5.3 Hz 55 km 22.00 4.0 577 m 577 m 9.1 cm 0.28 s 3.0 Hz 100 km 22.75 4.5 1.0 km 1.0 km 16 cm 0.50 s 1.7 Hz ' 220 km 10 km 23.50 5.0 1.8 km 1.8 km 29 cm 0.88 s 1.0 s 350 km 22 km 24.25 5.5 3.3 km 3.3 km 51 cm 1.6 s 1.9 s 500 km 42 km 25.00 6.0 8.0 km 4.1 km 91 cm 2.8 s 3.3 s 650 km 75 km 25.75 6.5 15 km 7.0 km 1.6 m 5.0 s 5.9 s 820 km 130 km 26.50 7.0 30 km 11 km 2.9 m 8.8s 10.4 s 1000 km 210 km 27.25 7.5 70 km 15 km 5.1 m 16 s 19 s 1200 km 340 km 28.00 8.0 150 km 22 km 9.1 m 28 s 33 s 1450 km 520 km 28.45 8.3 200 km 33 km 13 m 39 s 46 s 1600 km 670 km 28.75 8.5 250 km 42 km 16 m 49 s 58 s 1700 km 790 km There are no SCR stress droa data above Mw 6.7. The double corner that would manifest itself for rectangular ratier than square faults is not tabled here, and the aspect ratio W/L for the larger events is arbitrary but consistent with the sparse Mw>6 data. Figure A-19. Average stress drop seismic source parameters applicable to the central U.S. (abstract, Scism. Res. Lett.,1992, v. 64, no. 3-4, 261) A-48

l l Table 5. New Madnd source parameters. 1 model W model Pruna U,y,Emodgj rwi wa mi %,,Ami VIIfaw Imi h... m.,i D +., i 9 beri Xi NM1,16 December 1811: logGfoi = 28.15, M = 8.1 (t) Re mc+en Ruerure (br.n!c mt 4. 4g. *0 kwr it d e abe rement fau!t U.1"t tola dyn/cm 300 20 +4 77 8.0 10 150 95 40 1.0 x 104 250 20 40 71 96 15 180 125 N) 1.4 x 104 200 20 36 63 12.0 20 120 170 70 1.9 x 104 150 20 31 55 16 40 295 265 110 2.9 x 1&4 100 20 25 45 24 90 445 485 200 54x174 50 20 18 32 48 355 890 1370 560 13 x 173 (d)f.apanaea Rueture /bn"I* & semibm"!* nue 4. A, 31 k-F vemcal fault U.19m1011 dyn/cm2 i 300 33 56 100 4.6 5 55 45 20 46x173 250 33 51 91 5.5 to 65 60 25 6.1 x les 200 33 46 81 6.9 15 80 80 35 83xIg5 180 33 45 77 7.7 15 90 95 40 1.0 x 104 160 33 41 73 8.6 20 100 115 45 1.2 x 104 [_ 140 33 3a 68 9.9 25 115 140 55 1.5 x 104 l i 120 33 34 63 11J 35 135 175 70 1J x 104 100 33 32 57 14 55 165 230 95 2.4 x 1&4 80 V 29 31 17 85 205 320 130 3.4 x 104 65 33 26 46 21 130 250 435 180 4 6 x 104 50 33 23 41 28 215 325 645 265 6 8 x 104 NM2,23 January 1812: lopMol = 27.80, M = 7J Re me,ca a E ennaea Ruerun L,JL1 A22 stnke shp, vertical fault 150 20 31 55 51 15 105 95 40 1.0 x 104 100 to 25 43 8.5 30 160 170 70 1.tx 104 80 20 23 40 10.7 50 195 240 100 25 x 10d 60 20 20 35 14 90 265 370 150 4.1 x 104 40 20 16 28 21 195 395 680 280 73 x 104 120 33 36 63 4.1 15 50 to 25 6.5 x 173 100 33 32 57 4.9 20 60 80 35 SJ x 10'3 to 33 29 51 6.1 30 "'O 115 45 1.2 x 104 l 65 33 26 46 7.5 45 90 155 65 1.6 x 104 1 60 33 25 44 8.2 55 95 175 70 1.8 x 104 50 33 23 41 9.8 75 115 230 95 2.4 x 104 40 33 20 36 12 120 145 320 130 60x104 NM3,07 February 1812: log (Mo) = 18.00, M = 8.0 E_teanded Rucere bsch ceathe m (h y16 k_m h*. *x k-P dipsbp, dip -32*, # = 3J x 10 3 11dyn/cm2 160 45 48 85 3.8 to 30 40 15 4.4 x 10'8 140 45 45 79 4.3 10 35 50 20 5.4 x 10 5 120 45 41 73 5.0 15 40 60 25 6.8 x lo s 100 45 38 67 6.0 20 50 80 35 9.0 x 10 5 90 45 34 64 4.7 25 55 95 40 1.0 x 104 [ 75 45 33 58 8.0 40 65 125 50 1.4 x 104 i 60 45 29 52 10.0 60 80 175 70 1.9 x 104 50 45

• 7 47 12 90 100 230 95 2.5 x 104 45 45 25 45 13 110 110 265 110 3.0 x 104 32 45 21 38 19 215 155 445 185 4.9 x 104

- bold udmares fault -We. cmsidered more physcady reasonable than the otturs, teed nWem-see hg 12 rounded to nearest 5 bars - sans Figure A-20. Faulting parameters for the three principal New Madrid earthquakes of 1811-1812. The favored fault models are boxed (see Fig A 21) (Table 5 of Johnston,1996) e A 49

1 l l Viable Fault Models (a) _New Madrid _ (i) NM 1: Stnke elip vertical fault, Log (Ma) = 28.25 M. 8.1 r i i i i i i ' ' ' i i ' ' 3 8P '" NM1 L.140 ken . (16 Dec. teit) W.33km ?.I ' i 2 = 10m l l.T. N' // e,,,, i i, i i i i 's C fr g

• NM 2yj

'W (ii) NM 2: Stnke-elip, vertical. Loa (Mo). 27.80, M. 7.8 ....f r. TTTT r i ' i i ', NM2 L. 6Shm " g / NM3 ,(23 Jan.1812) W. 33km, \\ 3 7.5m \\ A t i,, i i i5 j A' y 4, (ill) NM3: Dir-olip, thrvetJp.32*. Loa (MA.28.00 M.Sc - NM1 cuo/,p[, r, , i,,. m f[- NuS t.75.m NM1 cn... iem w..ehm e. e.o. y t, ,3 ,* j ' y

./ o s'

A so,. NM S,[ 'O // f rr 5 25km ~ Figure A-21. Locations of the preferred fault models of Fig. A 20 within the NMSZ. The rationale for these locations is given in Johnston & Schweig (1996), (Fig. A 12 of Johnston, 1996) l t 5 A 50 ~

Central US Seismicity: 1974 - 1994 -92* -90' -88* -86* 40* ' 40' Missouri' Illinals Indiana r. e ~ e WVL, a e, 6 S S C. e.g e ( e .e. .e' .e

• .g e

.e * / J . (. 5 Kentucky , S ?... 3 _,,,{. 't Tennessee ..i ,a. ':g.,. - , j. & sea 3M f'2-36' a. 36' , j{e ,s G ~ "L".' r gg pij, y $Adf ed.. ,e c,.ss.c k & McMPitts 0 50 100 . )d g M>5 o / Mississippi e e 5>M24 s , Arkansas 4>y 3 3>M22 e Alabama y<3 34' . u' -92* -90' -88* -86* Figure A-22. The gene.ahzed seismic source zones added to Figure A 1 A Si

Appendix D l PRECESSING AND ANALYSIS OF TIIE EARTisQUAKE CATALOG B.1 CATALOG SELECTION AND MODIFICATION This study utilizes the earthquake catalog by Mueller et al. (1997) as the starting point for development of an earthquake catalog. The Mueller et al. nationwide catalog is obtained by combining a number of regional and national catalogs, removing duplicate events, and remosing enershocks; it covers the time period from 1626 to 1995. In the New Madrid aree, most events in this catalog come from the NCEER catalog (Seeber and Armbruster,1991), which.. urn is based on the EPRI(1986) catalog. Two modifications w ere made to this catalog. First, the magnitudes in the catalog were converted from m% o moment magnitude M, using the relationship by Johnston (1996): t logi, Af = 17.76 +0.36m,q+0.14m *q (B1) o 3 together with the llanks and Kanamori (1976) definition of moment magnitude; i.e., Af= logioAf -16.l' (B-2) o Second, events with magnitudes 5.5 or greater in the study region were checked against the corresponding events in the listing of Johnston et al. (1993). If the event appears in that reference, the magnitude and location appearing there are used in the catalog. B.2 CALCULATION OF SEISMICITY PARAMETERS Seismicity parameters for each seismic source were calculated using the method of maximum likelihood (Weichert,1985). Following EPRI (1986), the maximum likelihood calculations use a modified magnitude (the " uniform magnitude) to account for uncertainty in magnitude conversion. This magnitude is calculated as Af*=if bin (10)0 (B3) 2 whercif is the best-estimate magnitude in the catalog and o is the standard deviation associated with magnitude conversion. The b value is taken as unity in this calculation. Because the Mueller et al. catalog does not include o information for each event, the NCEER (1990) catalog was examined for the dependence of 0 as a function of magnitude and time. The resulting step-wise 2 functions of magnitude and time were used in the determination of o. B-1

i Catalog completeness is characterized using the EPRI (1986) probability of detection for the study region (region 4). These detection probabilities were used to compute the equivalent period of completeness, which is then used in the maximum-likelihood calculations. B.3 UNCERTAINTY IN SEISMICITY PARAMETERS The Weichert (1985) procedure also yields estimates of the epistemic uncertainty in the seismicity parameters. If the rate parameter is der:ned as the rate of earthquakes above the minimum magnitude used in the calculations, then this rate is independent of the b value. Uncertainty in these parameters is characterized by discretizing their distributions (assumed normal) into tisee-point distributions. The resulting nine combinations of rate and b value are used in the PSHA j calculations. B.4 REFERENCES Electric Power Rescatch Institute (l986). Seismic Hazard Methodologyfor the Central and Eastern United States, Vols. I through 10, EPRI NP-4726, Final Report. Johnston, A.C., K.J. Coppersmith, L.R. Kanter and C.A. Cornell (1993). The Earthquakes of Stable ContinentalRegions, Electric Power Research Institute, Final Repon No. TR-102261 December. Mueller, C., M. Hopper, and A. Frankel (1997). Prcparation ofEarthquale Catalogsfor the NationalSeismic-Ha:ardMaps: contiguous 48 Statec. U.S.G.S. Open-File Report 97-464. B2

I Appendix C ADDITIONAL SENSITIVITY RESULTS This appendix presents the results from sensitivity analyses for the major parameters and assomptions in the logic trees of Section 3 and to attenuation equations. Sensitisity calculations were performed for PGA r.nd 1 llz spectral acceleration on rock. Sensitivities to a large number of global and source parameters were calculated and examined; all important sensitivities are shown here. These results are calculated by computing the mean value of the total hazard given each assumption and then comparing the results obtained with the various assumptions. The weights assigned to the various assumptions or branches of the logic tree are also an important element of these comparisons and are included (in parentheses) in all figures. Figures C-1 through C-6 show the sensitivity to the main parameters of the East Prairie Exunsion (EPE). Figures C-7 and C-8 show the sensitivity to the rate oflarge (characteristic) events in the East Prairie fault Figures C-9 and C-10 show sensitivity to the epistemic uncertainty in ground motion amplitude (c,in Equation 4-1). Figures C-11 and C-12 show sensitivity to the alternative magnitude saturation models used for fault seismic sources. The most important contributors to uncertainty in the hazard are the epistemic uncertainty in ground-motion amplitude and the rate characteristic events on the East Prairie extension. Sensitivity to the geometry of the EFE is low because the assumption of a large source is assodstul with a lower rate per tnit area. Sensitivity to maximum magnitude is lower than sensitivity to the rate oflarge events. Sensitivity to the seismicity parameters (exponential portion of the magnitude-recurrence law; not shown)is small for the EPE and negligible for other satirces. Sensitivity to the magnitude-saturation medels in the r.ttenuation equations for faults is small. Ct\\9720\\rept_2\\ APP _C.WPD Wovember 25, 1997 C-l

l PGA - Sensitivity to New Madri. Char. Rate 1E-01 i i i i i i su YR (.700) . N YR (.300) --- 1E42 \\ 7 \\ W s 8 's 8 's% O \\s d 1E43 E N s s's s a s, f 1E-04 1E45 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Peak Acceleration (g) Figure C-1. Sensitivity of seismic hazard (PGA) to the rate of clusters oflarge NMSZ events. C:\\9728\\rept_2\\ APP _C.WPD November 25, 1997 b*2

1-Hz PSA - Sensitivity to New Madrid Char. Rate 1E-01 i i 1000 YR (.700) 500 YR (.300) -- lE42 - p \\ 7 8 \\s 8 s s's%s 8u d IE43 - '%s's E s**,- w a ~,

• • % g 1

1E-04 IE45 0 0.2 0.4 0.6 0.8 1 Spectral Acceleration (g) Figure C-2. Sensitivity of seismic hazard (1-Hz PSA) to the rate of clusters oflarge NMSZ events. C \\9728\\rept_2\\ APP,C.WPD November 25, 1997 C-3

PGA - Sensitivity to Characteristic Rate, EPE 1E-01 1000 YR (.300) 10000 YR (.700) -- 1E - t 7 \\ 8 8 's 8 \\ u s 5 1E43 N C N a N o s 5 s's s's s s s'%g' % *.,~~;- 1E-04 +\\ 1E45 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Peak Acceleration (g) Figure C-3. Sensitivity of seismic hazard (PGA) to the rate oflarge events on the East Prairie - Extension. The rate of 1 in 10,000 years represents the assumption that the NMSZ does not extend all the way to the site. C:\\9728\\rept,2\\ APP,C.WPD November 25, 1997 C-4 0

1-Hz PSA - Sensitivity to Characteristic Rate, EPE lE-01 i i i i i 1000_YR (.300) 10000 YR (.700) --- 1E42 - l 4 -l \\ \\ 8 \\ 8 \\ u \\ 's d IE-03 E \\ a \\ m IE-04 s'*s s %g IE45 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Spectral Acceleration (g) Figure C-4. Sensitivity of seismic hazard (1 Hz PSA) to the rate oflarge events on the East Prairie Extension. The rate of 1 in 10,000 years represents the assumption that the NMSZ de.. not extend all the way to the site. Ci\\9728\\rept,2\\ APP,C.WPD November 25, 1997 C-5

l PGA - Sensitivity to hi of EPE max 1E-01 max =8.0 i0.3 (.900) hi max =8.2 i0.3 (.100) -- hi 1E - 'IT 8 8 8 a s d 1E43 N s E N \\ 'iG %s

s s

%s % \\ %g en, een g % 1E-04 1E-05 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Peak Acceleration (g) Figure C-5. Sensitivity of seismic hazard (PGA) to the maximum magnitude on the East Prairie Extension. C \\9728\\rept_2\\ APP,C.WPD November ?$, 1997 C-b

l-1-Iz PSA - Sensitivity to hi of EPE max 1E-01 i i i i i i i himax=8.0 i0.3 (.900) M =8.210.3 (.100) --- max IE42 - 4 8u \\ 's d IE43 it' 's 3 o 1E-04 %~~ %g %g IE45 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Spectral Acceleration (g) Figure C-6. Sensitivity of seismic hazard (1-Hz PSA) to the maximum magnitude on the East Prairie Extension.. C:\\9728\\rept,2\\ APP,C.WPD Noverber 25, 1997 C*7

PGA - Sensitivity to Geometry of EPE 1E-01 100 km (.900) 160 km (.100) -- IE42 - ~ 8 u \\ d IE43 N s E N \\

• iG

s

's% %g %g % "'9 IE-04 1E-05 ' 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Peak Acceleration (g) Figure C-7 Sensitivity of seismic hazard (PGA) to the geometry of the East Prairie Extension. C:\\9726\\rept_2\\ APP _C.WPD Noverr.be r 25, 1997 C-8

~ 1-Hz PSA - Sensitivity to Geometry of EPE 6 1E-01 100 km (.900) 160 km (.100) -- 1E. 9 h 8 E 's 5 IE43.- 's C s's*

• g s

o s, sg*%s %g %g*%, l IE-04.- %'9 1E45 O 0.2 0.4 0.6 0.8 1 Spectral Acceleration (g) Figure C-8 Sensitivity of seismic hazard (1-Hz PSA) to the geometry of the East Prairie Extension. / ) C:\\9728\\rept_2\\ APP C.WPL November 25, 1997 C-9 'I

1 PGA - Sensitivity to Characteristic Rate, EP IE-01 10000 YR (.120) 1000 YR (.760) -- 500 YR (.120) - - - 1E - 4 g T s'- g s, s g s S 's 's,~ 6 g_o3 - ~,, a 1E - i l l 1E-05 O 0.2 0.4 0.6 0.8 1 1.2 1.4 Peak Acceleration (g) Figure C-9 Sensitivity of seismic hazard (PGA) to the rate of characteristic events on the East Prairie fault. 2 l C?\\9728\\rept,2\\ APP,C.WPD November 25, 1997 C-10

1-Hz PSA - Sensitivity to Characteristic Rate, EP 1E-01 10000 YR (.120) 1000 YR (.760) -- 500 YR (.120) - - - 1E - T 8 s. P:$ \\'. ' 's \\ s s 1E-03 C

g

.'s ,,~. s a E s,' ',~~,,' 1E-04 1E-05 0 0.2 0.4 0.6 0.8 1 Spectral Acceleration (g) Figure C-10 Sensitivity of seismic hazard (1-Hz PSA) to the rate of characteristic events on the East Prairie fault. C:\\9728\\rept_2\\ APP _C.WPD November 25, 1997 C-lI l 1 s ___J

b PGA - Sensitivity to Epistemic Unc. in Attenuation 1E-01 i i i i i i i mean - 2.33 o, (.046) mean - 0.74 o (.454) --- mean + 0.74 o,, (.454) - - - q mean + 2.33 o, (.046) .u... Ptft l'.i. i.. 1E-02

'g', \\,

s..

t.. \\s \\s..*...* ts \\ 's e s e s o s o s u s fa 1E % w \\ a. s g a s a s c s k %s ~ ~ s s s s's s* s* 1E-04 %,~, g% 1E45 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Peak Acceleration (g) Figure C-11 Sensitivity of seismic hazard (PGA) to epistemic uncertainty in the median ground-motion amplitude (represented by the ground-motion residual e., which has a standard deviation o,). C:\\9728\\rept_2\\ APP,C.WPD November 25, 1997 C-12

1-Hz PSA - Sensitivity to Epistemic Unc. in Attenuation 1E-01 i i i i mean - 0.74 o (.454) --- { mean - 2.33 o, (.046) t, mean + 0.74 o,, (.454) - - - E, mean + 2.33 o, (.046)....... b.r. @l.- 1E-02 ' LJ. ' t.

g..

g.., t ', t. 'iT \\ ', 8 .g N,.......,,,...~....""""- g s o s 's td 1E-03 g \\ ~......,,,,: a s N m s ss N s s % e, 1E-04 %s y 1E-05 0 0.2 0.4 0.6 0.8 1 Spectral Acceleration (g) Figure C-12 Sensitivity of seismic hazard (1-Hz PSA) to epistemic uncertainty in the median ground-motion amplitude (represented by the ground-motion residual e, which has a standard deviation o.). Ct\\9728\\rept_2\\ APP,C.WPD November 25, 1997 C-13

I l PGA - Sensitivity to Attenuation Saturation Model 1E-01 i i i i i i i Modeling (.400) Empirical (.400) -- None (.200) - - - 1E42 - e 6 1BM, ' :. ~.,,, ~~~___ 1E-04 1E-05 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Peak Acceleration (g) Figure C-I3 Sensitivity of seismic hazard (PGA) to magnitudc-saturation i todels used with fault seismic sources. Ci\\9At\\rept_2\\ APP _C.WPD November 25, 1997 C-I4

l 1-Hz PSA - Sensitivity to Atttenuation Saturation Model 1E, Modeling (.400) Empirical (.400) -- None (.200) - - - 1E42 - 7 8 4 B b IE N %g% it' 'E ' E',,.,,, ~ 1E : 1E45 0 0.2 0.4 0.6 0.8 1 Spectral Acceleration (g) Figure C-14 Sensitivity of seismic hazard (1-Hz PSA) to magnitude-satura' ion models used with fault seismic soarces. C:\\9728\\rept,2\\ APP _C.WPD November 25, 1997 C-15 l . _}}