The Dead Sea Basin

Ministry of Energy and Water Resources Geological Survey of Israel

Simulation of Seismic-Wave Propagation through Geometrically Complex Basins: The Dead Sea Basin

Shahar Shani-Kadmiel1,4, Michael Tsesarsky2, John N. Louie3, and Zohar Gvirtzman4

1 – Department of Geological and Environmental Sciences, Ben Gurion University of the Negev, Beer-Sheva, Israel. 2 – Department of Structural Engineering, Ben Gurion University of the Negev, Beer-Sheva, Israel. 3 – Nevada Seismological Laboratory University of Nevada, Reno, Nevada. 4 – Geological Survey of Israel, Jerusalem, Israel.

Prepared for the Steering Committee for Earthquake Readiness in Israel

Jerusalem, December 2012

Ministry of Energy and Water Resources Geological Survey of Israel

Simulation of Seismic-Wave Propagation through Geometrically Complex Basins: The Dead Sea Basin

Shahar Shani-Kadmiel1,4, Michael Tsesarsky2, John N. Louie3, and Zohar Gvirtzman4

Prepared for the Steering Committee for Earthquake Readiness in Israel

במסגרת הפרויקט "איפיון סיכוני תנודות קרקע באגנים סדימנטריים בישראל"

Jerusalem, December 2012

מדינת ישראל משרד התשתיות הלאומיות המכון הגיאולוגי

State of Israel Ministry of National Infrastructures Geological Survey

8.12.2012 תנודות קרקע באגני סדימנטריי בישראל

זהר גבירצמ

מניסיו שהצטבר בעול ידוע שבאגני גיאולוגיי צרי ועמוקי , שבנויי מסלעי רכי ביחס לשוליה , תנודות הקרקע בזמ רעידת אדמה מתארכות ומתחזקות בשיעור ניכר . אגני סדימנטריי עמוקי בישראל מפוזרי לאור בקע י המלח ועמקי הצפו . בתחומי האגני הסדימנטריי בישראל מצויי ריכוזי אוכלוסיה , בי היתר הערי בית שא וקריית שמונה , אזורי תעשיה ותיירות בדר ו י המלח , אזור תעשיה ומתקני רגישי בעמק זבולו ועוד . .

יחד ע זאת , נכו להיו בישראל אי בידינו מדידות שמה נית לאפיי את התופעה ולכמתה , מפני שמאז הקמתה של מדינת ישראל ובפרט מאז הצבת הרשת הסיסמית שלה לא התרחשו בישראל רעידות בינוניות או חזקות . בנוס! , דליל ותה היחסית של הרשת הסיסמית בישראל , שכמעט ואינ ה כוללת תחנות באגני הסדימנטרי , גורמת לכ שאפילו רישומי של רעידות חלשות כמעט לא קיימי . במצב זה אי אפשרות להשוות בי התנודות באגני לתנודות בשוליה ולא נית להערי את גודל התופעה והיקפה . .

למח קר זה , שמתוכנ להימש מספר שני , הוגדרו שתי מטרות : ראשית , לפרוס בכל אג סדימנטרי חשוב במדינת ישראל רשת סיסמית ניידת שתרשו רעידות אדמה במש תקופה של כמה חודשי באתרי שוני באג ומחוצה לו . שנית , לערו סימולציות נומריות של אפקט האג ולכייל אות , במידת האפשר , על ידי המדידו ת שתיאספנ ה בהדרגה . .

אנו מצפי לתרומה משמעותית בשלושה מישורי שכל אחד חשוב בפני עצמו. (1 ) רישו הקלטות בו זמני של רעידות אדמה באגני ובשוליה הסלעיי. (2 ) פיתוח מתודולוגיה והבנה תיאורטית של אפקטי פני % ובינ % אגניי. (3 ) אפיו כמותי של ההגברה באגני ישרא ל לצ ור עריכת תרחישי ולצור תקני בנייה.

בשני האחרונות הוקמה קבוצת מחקר בשיתו! פעולה בי ד"ר זהר גבירצמ מהמכו הגיאולוגי וד"ר מיכאל טסרסקי מאוניברסיטת ב גוריו ש במסגרתה התקדמנו בתחו הסימולציות הנומריות של התפשטות גלי באגני סדימנטריי. חלק ניכר מהמחקר נעשה על ידי סטודנט לדוקטורט (שחר שני % קדמיאל) שנסע ל פרופסור ג'ו לואי מה מעבדה הסיסמולוגי של נבדה בארה"ב ולמד ממנו כיצד להשתמש בשתי תוכנות שפותחו במש שני רבות בארה"ב וכוילו במסגרת מחקרי רבי. לאחר תקופת לימוד שבמהל כה בוצעו סימולציות דו % מימדיות עבור אג י המלח במחשבי המעבדה הסיסמולוגית של נבדה, הגענו למצב שמאפשר לנו עצמאות חישובית במכו הגיאולוגי.

המחקר שתואר לעיל מומ בחלקו מתקציב פרויקט רב שנתי של יציבות התשתית בי המלח וכ מכספי ועדת ההיגוי. ה מאמר המצור! בזאת מתו ה עיתו של החברה הסיסמולוגית אמריקנית מדגי תוצאות מ סימולציות דו % ממדיות שערכנו לאג י המלח . ב שנה הבאה אנו מתכנני להמשי במחקר תופעת מיקוד גיאומטרי של הגלי סיסמיי באגני כמו י המלח וכ לבצע סימולציות של עמק זבולו.

בכבוד רב

ד"ר זהר גבירצמ

.Malkhe Israel St 30 רח ' מלכי ישראל 30 דר ' זהר גבירצמן Dr. Zohar Gvirtzman Jerusalem, Israel 95501 ירושלים 95501 , ישראל 02-5314269 [email protected] Tel. 972-2-5314211 Fax. 972-2-5380688

Bulletin of the Seismological Society of America, Vol. 102, No. 4, pp. 1729–1739, August 2012, doi: 10.1785/0120110254

Simulation of Seismic-Wave Propagation through Geometrically Complex Basins: The Dead Sea Basin by Shahar Shani-Kadmiel, Michael Tsesarsky, John N. Louie, and Zohar Gvirtzman

Abstract The Dead Sea Transform (DST) is the source for some of the largest earthquakes in the eastern Mediterranean. The seismic hazard presented by the DST threatens the Israeli, Palestinian, and Jordanian populations alike. Several deep and structurally complex sedimentary basins are associated with the DST. These basins are up to 10 km deep and typically bounded by active fault zones. The low seismicity of the DST, the sparse seismic network, and limited coverage of sedimentary basins result in a critical knowledge gap. Therefore, it is necessary to complement the limited instrumental data with synthetic data based on computational modeling, in order to study the effects of earthquake ground motion in these sedimentary basins. In this research we performed a 2D ground-motion analysis in the Dead Sea Basin (DSB) using a finite-difference code. Cross sections transecting the DSB were compiled for wave propagation simulations. Results indicate a complex pattern of ground- motion amplification affected by the geometric features in the basin. To distinguish between the individual contributions of each geometrical feature in the basin, we developed a semiquantitative decomposition approach. This approach enabled us to interpret the DSB results as follows: (1) Ground-motion amplification as a result of resonance occurs basin-wide due to a high impedance contrast at the base of the uppermost layer; (2) Steep faults generate a strong edge-effect that further amplifies ground motions; (3) Sub-basins cause geometrical focusing that may significantly amplify ground motions; and (4) Salt diapirs diverge seismic energy and cause a decrease in ground-motion amplitude.

Introduction Sedimentary basins are known to amplify ground mo- generalizations of semiquantitative rules, useful for other tions and to prolong the shaking by trapping seismic energy basins around the world. (Anderson et al., 1986; Joyner, 2000; Boore, 2004). The out- The second goal of this study is to model earthquake come of this phenomenon was observed in Mexico City ground motion in the DSB, which hosts important industrial (Singh, Mena, and Castro, 1988), the Los Angeles basin and tourist facilities in Israel, Jordan, and the Palestinian (Graves, Pitarka, and Somerville, 1998), and Kobe, Japan Authority. The lack of seismic recordings in the basin, due to (Pitarka et al., 1998), among other places. The Dead Sea relatively low seismicity of the region and relatively sparse Basin (DSB) is a unique sedimentary basin due to its extreme national seismic network, produces the need for synthetic data depth, nearly 10 km, subvertical boundary faults, and com- in order to supplement the instrumental data. This study plex geometry formed by convex salt diapirs and concave explores principally the basin effects on earthquake ground sub-basins. Several active faults within the basin provide motion. internal seismic sources in addition to external sources from neighboring basins and the Dead Sea Transform (DST) itself. Geological Setting These circumstances provide an opportunity to study the influence of different intrabasin features on earthquake The DST is one of the largest active strike-slip faults of the ground motion. Our primary goal in this study is to develop world, connecting the east Anatolian fault in the north to the a semiquantitative methodology for decomposing a complex extensional zone of the Red Sea in the south (Fig. 1a; Garfun- basin effect to individual contributions derived from specific kel, Zak, and Freund, 1981). It defines the active boundary geometrical features. Such an analysis enables better under- between the Arabian and the African plates with an estimated standing of the integrated seismic phenomenon and allows ongoing slip rate of ∼3 to ∼5 mm=year (Wdowinski et al.,

1729 1730 S. Shani-Kadmiel, M. Tsesarsky, J. N. Louie, and Z. Gvirtzman

2004; Marco et al., 2005; Le Beon et al., 2008). The ∼105 km (Fig. 1b) and the normal step faults Sedom and Ghor-Safi of left-lateral motion along the DST since its formation in the are active (Aldersons et al., 2003; Hofstetter et al., 2007; Early to Middle Miocene (Quennell, 1956; Freund, Zak, and Data and Resources). Garfunkel, 1968) has created several pull-apart basins, the lar- 100 20 gest being the DSB, km × km in size (Fig. 1b). Seismicity This study focuses on the DSB, which is bounded by active normal step faults, filled with ∼10 km of soft sedi- Moderate and strong earthquakes associated with the ments and penetrated by large salt diapirs. It is generally DST are evident in geological, historical, and archaeological accepted that both eastern and western boundary faults records. However, due to long return periods, the instrumental

(a)32° 34° 36° (b) 35°18' 35°24' 35°30' 35°36'

Jericho

36° t

31°48'l 31°48'

t u l

a u F a

ry ry o

nda i r u e o Hotels ° J ° 31 42'B 31 42' km t l 34° n r u e 0 100 200 t a s F

e i f Judea Mountains W a DST CGF S -

rranean t 31°36'h l 31°36'

Dragot u MediteSea G a F

M. 32° JLM TLV AMN Shalem 1927 JV ° °

BS 31 30'Boundary 31 30'

° Sinai F 30 A Ein Gedi Easte

ELT 31°24' 31°24' Mt. Massada NBA and plain 1995 Arabian Plate African Plate Cross-section A 28° 31°18' 31°18' Red

Sea Lisan Diapir El Mazraa

31°12' 31°12' Hotels ns

ountai Sedom Fault 31°06'DSI 31°06' Cross-section B Mt. Sedom Moab M DSI Amiaz plain Safi 31°00' 31°00' and sub-basin km 010

35°18' 35°24' 35°30' 35°36'

Figure 1. (a) Overview map of the DST, compiled after Garfunkel, 1981. Arrows indicate directions of relative motion at faults. Epicenters of the 1927 Jericho earthquake and 1995 Gulf of Aqaba earthquake marked by gray filled circles. (b) Shaded relief map based on the DTM of Hall, 2008, overlaid by major faults and significant populated settlements, industrial facilities and tourist resorts. Modeled cross sections and simulated sources are denoted by straight solid lines and stars, respectively. Faults modified after Bartov and Sagy, 1999; Smit et al., 2008. Abbreviations: CGF, Carmel-Gilboa fault; DST, Dead Sea Transform; AF, Arava fault; JV, Jordan Valley; TLV, Tel-Aviv; JLM, Jerusalem; AMN, Amman; BS, Beer-Sheva; ELT, Elat; NBA, Nuweiba; DSI, Dead Sea Industries. The color version of this figure is available only in the electronic edition. Simulation of Seismic-Wave Propagation through Geometrically Complex Basins: Dead Sea Basin 1731 record is rather limited. To date, the strongest earthquake ever 2006), the spectral-element method (SEM; Chaljub et al., M recorded in Israel was the 1995 w 7.2 Gulf of Aqaba earth- 2005), the finite-difference method (FDM; Kristek, Moczo, quake (Fig. 1a), with its epicenter located ∼80 km south of and Pazak, 2009) and another implementation of the SEM Elat, the southernmost city of the country (Hofstetter, (Stupazzini, Paolucci, and Igel, 2009). These three methods, 2003). Prior to that, the largest earthquake felt in the country together with the finite-element method (FEM), are at present was the 1927 Jericho earthquake (Fig. 1a), later estimated the most powerful numerical modeling methods for earth- M from damage reports as an w 6.2 (Garfunkel et al., 1981; quake ground motion (Chaljub et al., 2010). They concluded Shapira, Avni, and Nur, 1993; Avni et al., 2002). that no single numerical modeling method can be considered For seismic hazard assessment it has been suggested that as the best for all important medium wave-field configura- the DST is capable of producing earthquakes with magni- tions in both computational efficiency and accuracy. tudes up to 7.5. Return periods for 7:5 ≥ M ≥ 5 were esti- Our modeling employs the FDM code E3D that was mated as 50 years in the Elat area, 30 years in the Arava and developed by the Lawrence Livermore National Laboratory Dead Sea area, and 25 years in the Jordan Valley (Fig. 1a; (Larsen et al., 2001). E3D is listed by the Organization for Shapira et al., 2007). However, because these estimates Economic Cooperation and Development’s Nuclear Energy strongly depend on the sparse historical record, much Agency (see Data and Resources). The E3D software research was invested in the unique paleoseismic record of simulates wave propagation by solving the elastodynamic the Dead Sea lacustrine sediments. formulation of the full wave equation on a staggered grid. Breccia beds in the Lisan formation formed during the The solution scheme is fourth-order accurate in space and last 60,000 years were interpreted as seismites (Seilacher, second-order accurate in time (Larsen et al., 2001). In this 1984), induced by M>5:5 earthquakes (Marco and Agnon, research we employ the software in 2D mode. Although 2D 1995; Marco et al., 1996; Marco and Agnon, 2005; Hamiel mode does not allow us to model truly closed basins, a clear et al., 2009). Marco et al. (1996) presented columnar sections benefit of 2D analysis is that it allows modeling of higher of the Lisan formation from the Massada plain and Amiaz frequencies. plain (Fig. 1b), exhibiting some 30 seismites that were formed by the same set of earthquakes. The seismites found within the Model Setup Amiaz plain (Fig. 1b) are consistently thicker than those found in the Massada plain, even though according to Begin et al. Two geological cross sections were simulated in this (2005), 11 strong earthquakes from the recorded set occurred study (locations in Fig. 1b): cross-section A transects the just north of Massada, which is farther from the Amiaz plain. basin east of Mount Massada, a UNESCO world heritage Another indication of strong ground motion in the Amiaz site (Fig. 2a); cross-section B transects Mount Sedom and plain is presented by Levi et al. (2008), who studied the de- the Amiaz plain near the Ein-Bokek Hotel complexes and velopment of clastic dykes found in the Amiaz plain and the industrial facility of the Dead Sea Industries (Fig. 2b). showed that they are seismically induced. According to Levi’s The cross sections were constructed based on a compilation models, a threshold value of M ≥ 6:5 earthquake at close of available geological data, borehole data, for example, proximity is needed in order to achieve the injection veloci- Sedom deep 1 (Baker, 1994), and geophysical data, mainly ties. Alternatively, the simulations presented here raise the seismic and gravimetry surveys (ten Brink et al., 1993; possibility that dyke injection as well as other seismites at Al-Zoubi and ten Brink, 2001). the Amiaz plain may be explained by exceptionally strong For cross-section A, we used structural maps of the top ground-motion amplification. and bottom of the Sedom formation salt unit (Al-Zoubi and The paleoseismic record of the Lisan formation shows ten Brink, 2001) and a generalized north–south cross section little evidence of surface ruptures that can be directly linked of the entire basin (Sagy, 2009) and used it for correlation with seismic activity on the boundary faults. Some superfi- with cross-section B. Cross-section B was compiled based cial faulting is documented (Marco and Agnon, 1995; Marco on structure from seismic surveys and supplemented by bore- and Agnon, 2005) within the ductile sediments of the forma- hole data for mechanical properties, specifically, pressure- tion, however, these are localized features that have no con- wave velocity and density (Frieslander, 1993; Baker, 1994; tinuous spatial distribution. Al-Zoubi, Shulman, and Ben-Avraham, 2002). Mechanical properties such as shear-wave velocity and quality factors Simulation Methods were derived using empirical relations presented in Brocher (2008). The salt diapir in section B protrudes through the Modeling of basin response to wave propagation has uppermost Lisan formation and gives rise to Mount Sedom, been used to study earthquake-shaking hazard in a limited rising ∼225 m above the Dead Sea and ∼100 m above the number of basins. Chaljub et al. (2010) have compared four Amiaz plain. Thus, the Amiaz plain is bounded by the wes- numerical predictions of ground motion in the Grenoble tern boundary fault in the west and Mount Sedom in the east Valley, France. The numerical modeling methods compared and is actually a sub-basin within the DSB. were the arbitrary high-order derivative discontinuous Galer- To simplify the numerical calculations in E3D, the topo- kin method (ADER-DGM; Käser, Dumbser, and de la Puente, graphy of the cross section was flattened; the top surface of the 1732 S. Shani-Kadmiel, M. Tsesarsky, J. N. Louie, and Z. Gvirtzman

1.0 (a) Massada (f) Amiaz 0.8 Reference Reference 0.6 0.4 1 1 PGV, m/s 0.2 0.0 10 (b) 4 2 (g) 42 8 4 6 4 4 4 2 3 3 Amp. ratio 0 Mt. Massada and Amiaz plain and Massada plain sub-basin W E WEMt. Sedom 0 km/s (c) (h) SSD 0.0 0.5 LSD 5 1.0

WBF WBF 1.5 SF SF 2.0 10

EBF EBF 2.5 Depth, km

GSF GSF 3.0

3.5 Shear wave velocity 15 0 18 (d) 5 (i) 5 m/s 3 3 0.1 5

10 0.0 Time, s 6 6 15 6 6 velocity

−0.1 Horizontal ground 20 (e) 0 (j) 1 7 m/s } 7 0.4 2 0.3 3 4 0.2 5 0.1 Frequency, Hz 6 Spectral velocity 0.0 7 0 5 10 15 20 25 0 5 10 15 20 25 Distance east, km Distance east, km Lisan and Samra Fm. - Pleistocene Hazeva Formation - Miocene Basement

Amora Formation - Pleistocene Mesozoic

Sedom Formation - Pliocene Paleozoic

Figure 2. Dead Sea Basin simulation results: Left panel cross-section A (Massada) and Right panel cross-section B (Amiaz). (a, f) Hor- izontal PGV. (b, g) Amplification ratio relative to a reference model. (c, h) Shear-wave velocity model of the modeled cross section. (d, i) Time-distance plot of horizontal velocity from surface cells. Gray is no ground motion, black is positive (east) ground motion, and white is negative (west) ground motion. Scale saturates at 0:1 m=s for clarity. (e, j) Frequency-distance plot, computed as the Fourier spectra of the synthetic seismograms presented in (d, i). The scale saturates at 0:4 m=s for clarity. Abbreviations: WBF, western boundary fault; EBF, eastern boundary fault; SF, Sedom fault; GSF, Ghor-Safi fault; SSD, Sedom Salt diapir; LSD, Lisan Salt diapir. The color version of this figure is available only in the electronic edition. resulting model conforms to the average elevation of the ex- Moab Mountains to the west and east of the Dead Sea, respec- posed Lisan formation along the transecting line. The Lisan tively (Fig. 1b). Therefore, the results of our simulations formation is the top most sediment filling the basin and is at an should only be applied to the basin itself and not to its unreal average elevation of −370 m ( below sea level) along cross- boundaries, which might have a topographic effect that was section A and −260 m along cross-section B. The water of the not considered. Water effects within the lake, such as water- Dead Sea and the air surrounding it was replaced with Lisan bottom multiple reflections, were also ignored. formation sediments to fill the missing topography. Mount As part of the simulation preprocessing, the geological Sedom above the Sedom Salt diapir (−180 m) was totally re- cross sections were spatially discretized into the intended grid moved, as well as the slopes of the Judea Mountains and the spacing, depending on the modeled frequencies. Simulation Simulation of Seismic-Wave Propagation through Geometrically Complex Basins: Dead Sea Basin 1733

Table 1 Figure 2a,f: Horizontal peak ground velocity (PGV) Dead Sea Basin, 2D Simulation Parameters across the modeled section sampled at model resolution (absolute value). Model dimensions (km; grid cells) 26:74 × 15; 5348 × 3000 Spatial discretization (km) 0.005 Figure 2b,g: Amplification ratio across the modeled sec- Time steps (#) 40,000 tion computed relative to a reference model, which is a Time step interval (s) 0.0005 homogeneous medium with properties of the surrounding Modeled time (s) 20.0 rocks. Note that this presentation of amplification following ∼30 Simulation processor time (hours) Gvirtzman and Louie (2010), differs from the common way Minimum; maximum VP (km=s) 1.35; 5.94 Minimum; maximum VS (km=s) 0.41; 3.55 of presenting amplification relative to reference stations on Minimum; maximum density (g=cm3) 1.74; 2.70 hard rock at the basin edges. Minimum; maximum QP 46; 806 Figure 2c,h: The modeled cross section, shaded accord- Minimum; maximum QS 23; 403 ing to shear-wave velocities (listed in Table 2). Figure 2d,i: Time-distance plot of horizontal velocity parameters and mechanical properties of the geological units synthetic seismograms sampled at the surface cells. Gray are summarized in Tables 1 and 2, respectively. is no ground motion, black is positive (east) ground motion, The simulated scenario presented here (Fig. 2) is a nor- and white is negative (west) ground motion. Although PGVs 1 = 0:1 = mal-slip rupture initiating at a depth of 13 km on Sedom fault reached values of nearly m s, the scale saturates at m s near the lower limit of the seismogenic zone in the region for clarity. (Aldersons et al., 2003; Ambraseys, 2006). In our simula- Figure 2e,j: Frequency-distance plot, computed as the tions the source is described in terms of a finite-length fault Fourier spectra of the synthetic seismograms presented in 0:4 = with uniform moment. The modeled hypocenter is denoted Figure 2d,i. The scale saturates at m s for clarity. by a star and paired arrows pointing in the slip direction. The ruptured fault plane of the finite source extends 3.5 km in the Description of Results up-dip direction, and rupture initiates near the bottom (of the fault plane). For our parametric study of basin effects we kept The largest ground motions are found directly above the simple ruptures entirely within high velocity rocks below the source in both the reference and the modeled cross sections basin. The normal-faulting double-couple rupture front pro- (observation 1 in Fig. 2a,f). Ground-motion amplification pagates radially from the hypocenter along the fault plane, at however, increases toward the side of the basin opposite a constant rupture velocity of 2:8 km=s(Scholz, 2002). All the source (observation 2 in Fig. 2b,g). In both sections a the 2D elements on the fault plane were given identical mo- local minimum appears approximately at the same location ment and a Gaussian source time function with frequency regardless of diapir or fault location (observation 3 in content between 0.1 and 10 Hz. Note that the size of the Fig. 2b,g). In both sections strong ground-motion amplifica- source (i.e., its moment) is not important in this 2D analysis tion is observed near faults and subvertical boundaries of salt that allows no energy to dissipate in the third dimension. bodies (observation 4). The time-distance plots (panels d and Therefore, we only analyze the relative amplification and i) show that waves traveling through the Lisan and Sedom derive no conclusions from the absolute ground motion. Salt diapirs reach the surface faster than waves traveling through the surrounding geological units (observation 5). Simulation Results It is also noticeable that the western and eastern boundary faults act as strong reflectors channeling most of the seismic Simulation results of the modeled cross-sections A and energy into the basin (observation 6). The frequency- B are summarized and visualized in Figure 2a–j by panels for distance plots (panels e and j) reveal resonance patterns (ob- each cross section. From top to bottom they present: servation 7) at the fundamental frequency f0 VS=4h,in

Table 2 Velocity, Density, and Q Properties of the Dead Sea Basin Formations*

3 Geological Unit VP,km=s VS,km=s ρ,g=cm QP QS Lisan & Samra formations—Pleistocene 1.35 0.41 1.74 46 23 Amora formation (upper)—Pleistocene 3.75 2.20 2.25 356 178 Amora formation (lower)—Pleistocene 4.04 2.44 2.28 414 207 Sedom formation—Pliocene 4.18 2.54 2.15 440 220 Hazeva formation—Miocene 4.53 2.78 2.39 510 255 Mesozoic 4.81 2.95 2.47 564 282 Paleozoic 5.58 3.37 2.62 726 363 Basement 5.94 3.55 2.70 806 403

*VP and ρ were measured in Sedom deep 1 borehole, VS, QP and QS were calculated using empirical relations (Frieslander, 1993; Baker, 1994; Brocher, 2008). 1734 S. Shani-Kadmiel, M. Tsesarsky, J. N. Louie, and Z. Gvirtzman accordance with the shear-wave velocity and thickness of the the source (observation 1 in Fig. 3b) and the amplification model’s uppermost layer. In cross-section A, overtones at signature follows a trend similar to that presented in Figure 2 fn nf0, n 1, 3,5,7,…, are visible directly above the (observation 2 in Fig. 3b). At a distance of approximately source. In cross-section B the overtones are absent above 15 km a local minimum appears, substantiating that this phe- the salt diapir but appear on both sides. The calculated fre- nomenon is independent of intrabasin features, that is, faults quencies for vertical resonance in the Amiaz sub-basin and diapirs which are absent from this model (observation 3). bounded by the western boundary fault and the salt diapir In the faults model, ground-motion amplification increases (Fig. 2b), are ∼0:3, ∼0:9, ∼1:5, and ∼2:1 Hz for modes 1, near the basin boundary faults or edges (observations 4 2, 3, and 4, respectively, which correspond to the values and 5 in Fig. 3c). The asymmetry between the two edge- marked as observation 7 in Figure 2b. effects in opposite sides of the basin is probably related to the general trend of the ground-motion amplification that in- Model Decomposition creases toward the right side of the basin (observation 2). The PGV signature produced by the diapir model resembles that The results described previously reflect a complex inter- of the Layers model except for a small depression directly action of several effects contributed by the various geometric above the diapir (observation 6 in Fig. 3d). The basin model features in the basin. To gain an in-depth understanding of produces three distinct peaks above the sub-basin, observa- this phenomenon that will enable the interpretation of the tions 7, 8, and 9 in Figure 3e. Combining all the geometrical general amplification trend and the local minima and max- features into a single model, the resulting signal contains the ima, a geometrical decomposition method is devised in the individual signature of each feature (Fig. 3f). following list. Six different geometrical models were constructed and analyzed. The complexity of the models succes- Interpretation sively evolved, changing only one element at a time. Figure 3 shows the step-by-step evolution of the six models used as Our decomposition technique revealed that the general simulation input. The following is a short description of the trend (observation 2) of the ground-motion amplification and different models: the local minimum (observation 3) are both independent of Figure 3a, reference: A reference model with a single intrabasin features. The general trend in the amplification ra- homogeneous medium. tio reflects the fact that PGV of the reference model decays Figure 3b, layers: A series of horizontal sedimentary over a much shorter distance compared with that of the layers layers with mechanical properties of the DSB geologi- model. While in the reference model PGV at a distance of cal units. more than 15 km from the epicenter decays to nearly zero, Figure 3c, faults: A series of layers as in (b) bounded by in the layers model energy is trapped in the uppermost layer two near-vertical faults. and PGV remains approximately constant (Fig. 3b). Figure 3d, diapir: A series of layers as in (b) with a Entrapment of seismic energy in a soft layer on top of a dome-shaped intrusion (diapir), near the surface. hard substrate is a well-known phenomenon, visualized by Figure 3e, basin: A series of layers as in (b) with a deep the wave-field snapshots in Figure 4. This effect is caused sub-basin near the surface. by interference of seismic waves in several different man- Figure 3f, combined: A model combining all of the fea- ners: (1) Body waves reflected from the surface interfere with tures in Figure 3a–e. body waves reflected from the base of the uppermost layer The mechanical properties of the individual units are causing vertical resonance; (2) Body and surface waves in- summarized in Table 2. The modeled earthquake hypocenter teraction caused when body waves reflected from the base of is fixed at the same location in all simulations (see Fig. 3). the uppermost layer interfere with surface waves traveling Fault plane of the finite source extends 3.0 km in the upward across the basin; and (3) Surface-surface waves interaction dip direction and rupture initiates near the bottom (of the caused when left-traveling surface waves interfere with right- fault plane). The normal-faulting double-couple rupture front traveling surface waves. The net result of the previously de- propagates radially from the hypocenter along the fault scribed processes is significant ground motion for prolonged plane, at a constant rupture velocity of 2:8 km=s(Scholz, duration. 2002). All the 2D elements on the fault plane were given The local minimum within the generally increasing identical moment and a Gaussian source time function. amplification trend is related to the source radiation pattern. Table 3 summarizes the simulation parameters. It resides roughly on a plane rotated at 45° to the nodal planes and is visible in the time-distance plots in Figures 2 and 3b Decomposition Results (observation 3) as the first motion of shear-waves transform from left to right. The PGV signature of the feference model is straightfor- The ground-motion amplification that occurs near the ward. Strongest ground motions above the source and a gra- boundary faults of the basin is caused by the interference dual decrease with distance (observation 1 in Fig. 3a). The of surface waves and body waves to create an edge-effect layers model produces a ground-motion amplification above (Kawase, 1996; Graves et al., 1998; Pitarka et al., 1998). At Simulation of Seismic-Wave Propagation through Geometrically Complex Basins: Dead Sea Basin 1735

(a) (b) (c) 1.2 0.8 1 0.4 0.0 PGV, m/s

8 2 5 6 3 4 1 4

ratio 2 Amp. 0 0 5 10

Depth, km 15 0 003 003 3 4 5 5 10 SW BW RW 15 Time, s 20 0 2 4 6 Freq., Hz 0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35 Distance, km Distance, km Distance, km (d) (e) (f) 1.2 0.8 0.4 0.0 PGV, m/s 8 7 8 7 8 6 6 9 6 5 4 4

ratio 2 Amp. 0 0 5 10

Depth, km 15 0 5 10 15 Time, s 20 0 2 4 6 Freq., Hz 0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35 Distance, km Distance, km Distance, km km/s m/s m/s 0123 −0.1 0.0 0.1 0.0 0.2 0.4 0.6 0.8 Shear wave velocity Horizontal ground velocity Spectral velocity

Figure 3. Simulation results from six models: (a) reference, (b) layers, (c) faults, (d) diapir, (e) basin, and (f) combined. The presentation scheme follows that of Figure 2. Time-distance plot scale saturates at 0:1 m=s, and the frequency-distance plot scale saturates at 0:8 m=s for clarity. Abbreviations: PGV, peak ground velocity; Amp., amplification; Freq., frequency; BW, body wave; SW, surface wave; RW, reflected wave. The color version of this figure is available only in the electronic edition. 1736 S. Shani-Kadmiel, M. Tsesarsky, J. N. Louie, and Z. Gvirtzman

Table 3 propose that this convex body scatters body waves that are Model Decomposition Simulation Parameters reflected downward from the surface thus, preventing vertical resonance. Model dimensions (km; grid cells) 35 × 15; 1750 × 750 Spatial discretization (km) 0.02 The two peaks above the edges of the sub-basin in Time steps (#) 10,000 Figure 3e marked as observations 7 and 9 are near and far Time step interval (s) 0.002 edge-effects, respectively, caused by the subvertical walls Modeled time (s) 20.0 bounding the sub-basin. The central peak, observation 8, is ∼24 Simulation processor time (hours) caused by a geometrical convergence of the seismic waves Minimum; maximum VP (km=s) 1.35; 5.94 Minimum; maximum VS (km=s) 0.41; 3.55 by the concave structure of the sub-basin (Graves et al., Minimum; maximum density (g=cm3) 1.74; 2.70 1998; Semblat et al., 2002). We term this type of conver- Minimum; maximum QP 46; 806 gence “geometrical focusing”. Minimum; maximum QS 23; 403 Figure 5 presents PGV curves (Fig. 5a) and amplification ratios (Fig. 5b) from all six simulations plotted together above the combined model (Fig. 5d) for comparison. After the near fault (the fault nearest the source), seismic waves analyzing the individual signatures of the geometrical fea- propagate upward on both sides of the fault, the faster travel- tures, we are able to quantify their relative contribution. In ing body waves on the left side of the fault reach the surface particular, we distinguish between ground-motion amplifica- before the slower body waves propagating on the right side tion related to material properties such as that illustrated by of the fault. Surface waves formed at the basin edge propa- the layers model, and ground-motion amplification related to gate into the basin and interfere with later arriving body geometrical features such as that illustrated by the faults, waves (Gvirtzman and Louie, 2010). The development of diapir, and sub-basin models. this near-fault edge-effect (observation 4) is visualized in To accomplish this, the amplification ratio for the com- the time-distance plot of the faults model in Figure 3c (body bined model is computed relative to the layers model and waves, BW; surface waves, SW). At the far fault (on the right presented in Figure 5c. This exercise demonstrates that side of the basin), seismic waves reflected by the fault inter- material related ground-motion amplification is perturbed by fere with seismic waves trapped in the uppermost layer re- geometrical effects. Fault-related edge effect amplifies sulting in a similar edge-effect (observation 5 in Fig. 3c). The ground motion by 30% (observations 4 and 5); geometrical time-distance plot of the faults model shows the development focusing in sub-basins amplifies ground motion by 30% in time of the far-fault edge-effect (reflected waves, RW). (observation 8); and divergence of seismic waves by diapirs The diapir, an upward convex structure with shear-wave deamplifies ground motion by 50% (observation 6). As some velocity higher than its surroundings, leads to a decrease in of the seismic energy is trapped in the sub-basins, ground ground-motion amplification (observation 6 in Fig. 3d). We motion between the sub-basin and the far-fault is deamplified by 30% (observation 10).

Sedom Sedom Salt Salt Diapir Diapir

W lt W lt

u u

e e

a a

s s

te F te F

y y

r r r r

n n

a a S S B Bo d d Discussion e e

n n o d d

u u u u o o

o o n n m m

d d

B B F Fa a a lt lt

ry rn ry rn a u u

u a u a

te te

F F l l

s s t t a a fi F a fi F a u u

a E a E

lt lt S S The decomposition process presented here not only - r r-

o o

h h G t=2.0 s G t=3.5 s enables us to identify the individual contribution of various intrabasin features to the ground-motion amplification, it Sedom also allows us to reexamine the complex results of the Salt Diapir t=4.6 s DSB simulations. Material related ground-motion amplification occurs Sedom throughout the entire basin due to resonance developed Salt within the Pleistocene lacustrine sediments of the Samra Diapir t=5.2 s and Lisan formations (unified in our models). This effect is

Sedom caused by the impedance ratio across the interface between Salt the Samra–Lisan formation and the sediments of the Pleis- Diapir t=6.8 s tocene Amora formation and the Pliocene Sedom salt, which are 3.5 and 4, respectively. Material related ground-motion Sedom Salt amplification is illustrated by the simulation results of the Diapir t=10 s layers model (Fig. 3b), of which material properties follow those of the geological units of the DSB cross sections. Figure 6a presents synthetic seismograms sampled from Figure 4. Wave-field snapshots of modeled cross-section B. White is motion east; black is motion west. Time in seconds is dis- the reference and the layers models (see Fig. 3a,b for loca- played at the bottom right corner of each snapshot. The color ver- tion). The Fourier spectra of these seismograms (Fig. 6b) and sion of this figure is available only in the electronic edition. the spectral amplification ratio (Fig. 6c), reveal amplification Simulation of Seismic-Wave Propagation through Geometrically Complex Basins: Dead Sea Basin 1737

(a) (a) 1.2 1 Reference Layers 0.4 0.8 Faults 0.0 0.4 Diapir PGV, m/s −0.4 0.0 Basin Reference

(b) Combined Velocity, m/s −0.8 8 8 Layers 7 5 6 0 2 4 6 8 1012141618 4 4 Time, s 1 2 Amp. ratio 6 3 9 10 (b) (c) 0 1 10 (c) 2.0 0.5 1.6 4 7 8 5 0.2 8 1.2 0.1 6 mode 2 0.8 0.05 mode 3 } mode 4 Amp. ratio 0.4 6 0.02 10 4

0.0 0.01 mode 1 km/s

(d) Velocity, m/s 0 0.0 0.005 2 0.5 0.002 Amplification ratio 5 1.0 0.001 0 1.5 0.1 0.2 0.51 2 5 10 0.1 0.2 0.5 1 2 5 10 2.0 Frequency, Hz Frequency, Hz 10 2.5 Depth, km 3.0 15 3.5 Figure 6. (a) Synthetic seismograms of horizontal ground ve- 0 5 10 15 20 25 30 35 Shear wave velocity locity sampled from the reference model and the layers model (see Distance, km Fig. 4a,b for location). (b) Fourier spectrum of each of the synthetic seismograms. (c) Spectral amplification ratio. Figure 5. PGV data from all six simulations plotted together: (a) PGV across the modeled section. (b) Amplification ratio relative Mount Sedom, with a cross-sectional wavelength of 4 km to the reference model. (c) Amplification ratio of the combined and a shear-wave velocity of 2:54 km=s. Because topogra- model relative to the layers model. (d) Shear velocity model of the combined model. Line thickness varies so that overlapped lines phy can have significant effects on seismic waves when remain visible. The color version of this figure is available only in the incident wavelength is comparable to the size of the to- the electronic edition. pographic feature, amplification would be expected at ∼0:6 Hz (Boore, 1972). The steep shoulders of the DSB rise at the fundamental frequency of 0.2 Hz, and at its overtones 400 to 500 m above the basin with shear-wave velocity 0.6, 1.0, and 1.4 Hz. ranging from 2.95 to 3:37 km=s. To assess the topographic Comparing the synthetic seismogram from the idealized effect of these features, we follow the method presented by layers model with that from the Amiaz plain in cross-section B of the DSB simulations (see Fig. 2b for location) shows that (a) the typical resonance pattern of mode 1, 2, 3 is distorted by 0.20. ground-motion amplification at other frequencies as well 0.00. (labeled in Fig. 7 with a question mark). Specifically, note −0.2−0. the prominent peak found between the fundamental fre- Reference quency, 0.3 Hz, and the first overtone, 0.9 Hz. We suggest Velocity, m/s −0.4−0. Amiaz that these amplified frequencies are contributed by the basin 0 2 4 6 8 1012141618 deeper structure. Time, s In light of these results we suggest an explanation to the (b) (c) abundance of clastic dykes injected into the Lisan formation 1 10 0.5 in the Amiaz plain (Levi et al., 2008). Whereas seismites, 0.2 8 that is, breccia, liquefied layers, and slumps, have been 0.1 mode 1 6 observed throughout the Lisan formation, the clastic dykes 0.05 ? 0.02 mode 2 4 mode 4 are confined to the Amiaz plain above the Amiaz sub-basin. 0.01 mode 3 Emplacement of clastic dykes compared with other seismites Velocity, m/s 0.005 2 requires a higher energy threshold. We attribute the localiza- 0.002 Amplification ratio 0.001 0 tion of clastic dykes to the previously described geometrical 0.1 0.2 0.5 1 2 5 10 0.1 0.2 0.5 1 2 5 10 effect of the Amiaz sub-basin. Frequency, Hz Frequency, Hz The topographic effect on ground-motion amplification Figure 7. (a) Synthetic seismograms of horizontal ground ve- was not accounted for in our simulations; however, with the locity sampled from the cross-section A and its reference model (see results of Boore (1972) this effect can be readily estimated. Fig. 2b for location). (b) Fourier spectrum of each of the synthetic Within the DSB, the sole prominent topographic feature is seismograms. (c) Spectral amplification ratio. 1738 S. Shani-Kadmiel, M. Tsesarsky, J. N. Louie, and Z. Gvirtzman

Ashford et al. (1997), yielding topographic amplification at Acknowledgments ∼0:65 Hz. Our study explores the ground-motion effects of This research was partially funded by the Ministry of National Infra- basin structure between 0.1 and 7 Hz, hence, our results are structures of the State of Israel, Grant #210-17-001, and by the Geological limited at the lower end of this frequency band, where topo- Survey of Israel as part of a project assessing the instability factors in the graphic effects are expected to occur. Dead Sea Infrastructure. Water-bottom multiple reflections in the Dead Sea would be expected to affect the vertical resonance discussed References previously. The density of the briny Dead Sea water is ∼1:2 = 3 Al-Zoubi, A., and U. S. ten Brink (2001). Salt diapirs in the Dead Sea basin g cm , hence, pressure-wave velocity is slightly higher and their relationship to Quaternary extensional tectonics, Mar. Petrol. than 1:5 km=s, and the fundamental resonant frequency for a Geol. 18, no. 7, 779–797. water column of 200 m is on the order of 1.9 Hz. Our models, Al-Zoubi, A., H. Shulman, and Z. Ben-Avraham (2002). Seismic reflection which substitute shallow sediments for lake water, do not profiles across the southern Dead Sea basin, Tectonophysics 346, – – show this pressure-wave resonance. no. 1 2, 61 69, doi 10.1016/S0040-1951(01)00228-1. Aldersons, F., Z. Ben-Avraham, A. Hofstetter, E. Kissling, and T. Al-Yazjeen (2003). Lower-crustal strength under the Dead Sea basin from local Summary and Conclusions earthquake data and rheological modeling, Earth Planet. Sci. Lett. 214, no. 1–2, 129–142, doi 10.1016/S0012-821X(03)00381-9. The dimensions of the Dead Sea Basin (DSB), 10 km Ambraseys, N. N. (2006). Comparison of frequency of occurrence of earth- deep × 20 km wide × 100 km long, its steep boundary faults quakes with slip rates from long-term seismicity data: The cases of Gulf of Corinth, Sea of Marmara and Dead Sea fault zone, Geophys. and salt diapir intrusions result in a complex geometry. Due J. Int. 165, no. 2, 516–526, doi 10.1111/j.1365-246X.2006.02858.x. to the low seismicity of the Dead Sea Transform (DST) com- Anderson, J. G., P. Bodin, J. N. Brune, J. Prince, S. K. Singh, R. Quaas, and bined with limited instrumental coverage of the DSB, there is M. Onate (1986). Strong ground motion from the Michoacan, Mexico, a critical knowledge gap in terms of expected ground mo- earthquake, Science 233, no. 4768, 1043–1049. tions during a strong earthquake. However, moderate and Ashford, S. A., N. Sitar, J. Lysmer, and N. Deng (1997). Topographic effects on the seismic response of steep slopes, Bull. Seismol. Soc. Am. 87, large seismic events from instrumental (Data and Resources), no. 3, 701–709. archaeological and historical (Ellenblum et al., 1998), and Avni, R., D. Bowman, A. Shapira, and A. Nur (2002). Erroneous interpreta- geological (Marco et al., 1996; Levi et al., 2008) records tion of historical documents related to the epicenter of the 1927 Jericho are well documented. In this research we performed a 2D nu- earthquake in the Holy Land, J. Seismol. 6, no. 4, 469–476. merical ground-motion analysis of the DSB with particular Baker, S. (1994). Sedom Deep #1, Geological Completion Report, The Israel National Oil Co. Ltd, Geol. Surv. Israel. consideration of the geometrical complexity. Specifically, Bartov, Y., and A. Sagy (1999). Late Pleistocene extension and strike-slip in we studied the individual contribution of each geometrical the Dead Sea basin, Geol. Mag. 141, no. 5, 565–572, doi 10.1017/ feature in the basin to surface ground motion. S001675680400963X. We show that via a semiquantitative decomposition Begin, Z. B., J. N. Louie, S. Marco, and Z. Ben-Avraham (2005). Prehistoric approach, the contribution of the individual intrabasin fea- seismic basin effects in the Dead Sea pull-apart, The Ministry of Na- tional Infrastructures, Geol. Surv. Israel, Report GSI/04/05. tures to ground motion can be identified in an otherwise com- Boore, D. M. (1972). A note on the effect of simple topography on seismic plex signal. This process not only allows identification of the SH waves, Bull. Seismol. Soc. Am. 62, no. 1, 275–284. individual signature of each feature, but also conveys a phy- Boore, D. M. (2004). Can site response be predicted? J. Earthquake Eng. 8, sical understanding of how these signatures interact to form no. 1, 1–41. the complete signal. Ground-motion amplification in geome- Brocher, T. (2008). Key elements of regional seismic velocity models for long period ground motion simulations, J. Seismol. 12, no. 2, 217– trically complex sedimentary basins occurs (1) basin-wide 221, doi 10.1007/s10950-007-9061-3. due to resonance and anelastic effects, (2) above steep struc- Chaljub, E., C. Cornou, P. Guéguen, M. Causse, and D. Komatitsch (2005). tures, that is, faults and diapir flanks due to an edge- Spectral-element modeling of 3D wave propagation in the Alpine effect, and (3) above sub-basins due to localized geometrical Valley of Grenoble, France, in Proc. from Geophysical Research focusing. Narrow and upward convex bodies with a relatively Abstracts, Wien, Austria, vol. 7, 05225 Chaljub, E., P. Moczo, S. Tsuno, P. Y. Bard, J. Kristek, M. Käser, high seismic-wave velocity, such as salt diapirs and magmatic M. Stupazzini, and M. Kristekova (2010). Quantitative comparison intrusions, may cancel ground-motion amplification. of four numerical predictions of 3D ground motion in the Grenoble Valley, France, Bull. Seismol. Soc. Am. 100, no. 4, 1427–1455, doi 10.1785/0120090052. Data and Resources Ellenblum, R., S. Marco, A. Agnon, T. Rockwell, and A. Boas (1998). Crusader castle torn apart by earthquake at dawn, 20 May 1202, Description of E3D is available from http://www Geology 26, no. 4, 303–306. .oecd‑nea.org/tools/abstract/detail/ests1300 (last accessed Freund, R., I. Zak, and Z. Garfunkel (1968). Age and rate of the sinistral December 2011). The Earthquake Catalog of the Geophysi- movement along the Dead Sea Rift, Nature 220, no. 5164, 253– cal Institute of Israel can be found at www.gii.co.il (last 255, doi 10.1038/220253a0. accessed December 2011). Data on the digital terrain Frieslander, U. (1993). Velocity survey, Sedom Deep-1 Well: Israel Petro- leum Research of Geophysics Report, 849, no. 297, 92. model of Israel (Hall, 2008) were retrieved from www Garfunkel, Z., I. Zak, and R. Freund (1981). Active faulting in the Dead Sea .sciencefromisrael.com/index/914317320tx22207.pdf (last rift, Tectonophysics 80, no. 1–4, 1–26, doi 10.1016/0040-1951(81) accessed December_2011). 90139-6. Simulation of Seismic-Wave Propagation through Geometrically Complex Basins: Dead Sea Basin 1739

Graves, R. W., A. Pitarka, and P. G. Somerville (1998). Ground-motion am- Nanbu (Kobe), Japan, earthquake, Bull. Seismol. Soc. Am. 88, no. 2, plification in the Santa Monica area: Effects of shallow basin-edge 428–440. structure, Bull. Seismol. Soc. Am. 88, no. 5, 1224–1242. Quennell, A. (1956). The structural and geomorphic evolution of the Dead Gvirtzman, Z., and J. N. Louie (2010). 2D analysis of earthquake ground Sea Rift, Q. J. Geol. Soc. London 114, 1–24. motion in Haifa Bay, Israel, Bull. Seismol. Soc. Am. 100, no. 2, Sagy, A. (2009). Cross-sections along and across the Dead-Sea basin, Geol. 733–750. Surv. Isr., TR-GSI/07/2009. Hall, J. K. (2008). The 25-m DTM (Digital Terrain Model) of Israel, Isr. J. Scholz, C. H. (2002). The Mechanics of Earthquakes and Faulting Second. Earth Sci. 57, no. 3, 145–147. Ed, Cambridge University Press, New York. Hamiel, Y., R. Amit, Z. B. Begin, S. Marco, O. Katz, A. Salamon, Seilacher, A. (1984). Sedimentary structures tentatively attributed to seismic E. Zilberman, and N. Porat (2009). The seismicity along the Dead events, Mar. Geol. 55, no. 1–2, 1–12. Sea fault during the last 60,000 years, Bull. Seismol. Soc. Am. 99, Semblat, J. F., P. Dangla, M. Kham, and A. M. Duval (2002). Seismic site no. 3, 2020–2026. effects for shallow and deep alluvial basins: In-depth motion and fo- Hofstetter, A. (2003). Seismic observations of the 22/11/1995 Gulf of Aqaba cusing effect, Soil Dynam. Earthquake Eng. 22, no. 9–12, 849–854. earthquake sequence, Tectonophysics 369, no. 1–2, 21–36, doi Shapira,A.,R. Avni,and A. Nur (1993). A new estimate for the epicenter of the 10.1016/S0040-1951(03)00129-X. Jericho earthquake of 11 July 1927, Isr. J. Earth Sci. 42, no. 2, 93–96. Hofstetter, R., Y. Klinger, A.-Q. Amrat, L. Rivera, and L. Dorbath (2007). Shapira, A., R. Hofstetter, A. Q. F. Abdallah, J. Dabbeek, and W. Hays Stress tensor and focal mechanisms along the Dead Sea fault and re- (2007). Earthquake hazard assessments for building codes, U.S. lated structural elements based on seismological data, Tectonophysics Agency for International Development, Bureau for Economic Growth, 429, no. 3–4, 165–181, doi 10.1016/j.tecto.2006.03.010. Agriculture and Trade, Washington, D.C. Joyner, W. B. (2000). Strong motion from surface waves in deep sedimen- Singh, S. K., E. Mena, and R. Castro (1988). Some aspects of source tary basins, Bull. Seismol. Soc. Am. 90, no. 6B, S95–S112. characteristics of the 19 September 1985 Michoacan earthquake Käser, M., M. Dumbser, and J. de la Puente (2006). An efficient ADER-DG and ground motion amplification in and near Mexico City from strong method for 3-dimensional seismic wave propagation in media with motion data, Bull. Seismol. Soc. Am. 78, no. 2, 451–477. complex geometry, in Proc. from ESG 2006, Third International Smit, J., J. P. Brun, X. Fort, S. Cloetingh, and Z. Ben-Avraham (2008). Salt Symposium on the Effects of Surface Geology on Seismic Motion, tectonics in pull-apart basins with application to the Dead Sea basin, Grenoble, France. Tectonophysics 449, 1–16, doi 10.1016/j.tecto.2007.12.004. Kawase, H. (1996). The cause of the Damage Belt in Kobe: “The basin-edge Stupazzini, M., R. Paolucci, and H. Igel (2009). Near-fault earthquake effect,” constructive interference of the direct S wave with the basin- ground-motion simulation in the Grenoble Valley by a high- induced diffracted/Rayleigh waves, Seismol. Res. Lett. 67, no. 5, performance spectral element code, Bull. Seismol. Soc. Am. 99, 25–34. no. 1, 286–301, doi 10.1785/0120080274. Kristek, J., P. Moczo, and P. Pazak (2006). Numerical modeling of earth- ten Brink, U. S., Z. Ben-Avraham, R. E. Bell, M. Hassouneh, D. F. Coleman, quake motion in Grenoble basin, France, using a 4th-order velocity- G. Andreasen, G. Tibor, and B. Coakley (1993). Structure of the Dead stress arbitrary discontinuous staggered-grid FD scheme, in Proc. from Sea Pull-Apart Basin from gravity analyses, J. Geophys. Res. 98, ESG 2006 Third International Symposium on the Effects of Surface 21,877–21,894. Geology on Seismic Motion, Grenoble, France. Wdowinski, S., Y. Bock, G. Baer, L. Prawirodirdjo, N. Bechor, S. Naaman, Larsen, S., R. Wiley, P. Roberts, and L. House (2001). Next-generation R. Knafo, Y. Forrai, and Y. Melzer (2004). GPS measurements of numerical modeling: Incorporating elasticity, anisotropy and attenua- current crustal movements along the Dead Sea fault, J. Geophys. tion, Soc. Explor. Geophys. Ann. Internat. Mtg., Expanded Abstracts, Res. 109, B05403. doi 10.1029/2003JB002640. 1218–1221. Le Beon, M., Y. Klinger, A. Q. Amrat, A. Agnon, L. Dorbath, G. Baer, J.-C. Department of Geological and Environmental Sciences Ruegg, O. Charade, and O. Mayyas (2008). Slip rate and locking depth Ben Gurion University of the Negev from GPS profiles across the southern Dead Sea Transform, J. Geo- Beer-Sheva, Israel phys. Res. 113, no. B11, doi 10.1029/2007JB005280. (S.S.-K.) Levi, T., R. Weinberger, Y. Eyal, V. Lyakhovsky, and E. Heifetz (2008). Velocities and driving pressures of clay-rich sediments injected into Department of Structural Engineering clastic dykes during earthquakes, Geophys. J. Int. 175, no. 3, Ben Gurion University of the Negev 1095–1107, doi 10.1111/j.1365-246X.2008.03929.x. Beer-Sheva, Israel Marco, S., and A. Agnon (1995). Prehistoric earthquake deformations near (M.T.) Masada, Dead Sea graben, Geology 23, no. 8, 695–698. Marco, S., and A. Agnon (2005). High-resolution stratigraphy reveals repeated earthquake faulting in the Masada fault zone, Dead Sea Nevada Seismological Laboratory Transform, Tectonophysics 408, no. 1–4, 101–112. University of Nevada Marco, S., T. Rockwell, A. Heimann, U. Frieslander, and A. Agnon (2005). Reno, Nevada (J.N.L.) Late Holocene activity of the Dead Sea Transform revealed in 3D palaeoseismic trenches on the Jordan Gorge segment, Earth Planet. Sci. Lett. 234, no. 1–2, 189–205, doi 10.1016/j.epsl.2005.01.017. Marco, S., M. Stein, A. Agnon, and H. Ron (1996). Long-term earthquake Geological Survey of Israel clustering: A 50,000-year paleoseismic record in the Dead Sea Graben, Jerusalem, Israel J. Geophys. Res. 101, no. B3, 6179–6191. (Z.G.) Pitarka, A., K. Irikura, T. Iwata, and H. Sekiguchi (1998). Three-dimensional simulation of the near-fault ground motion for the 1995 Hyogo-Ken Manuscript received 8 September 2011