Seismic Hazard Analysis for Qaraoun Dam in the Bekaa Valley, Lebanon

Tenth U.S. National Conference on Earthquake Engineering Frontiers of Earthquake Engineering July 21-25, 2014 10NCEE Anchorage, Alaska

SEISMIC HAZARD ANALYSIS FOR QARAOUN DAM IN THE BEKAA VALLEY, LEBANON

C. B. Johnson1, D. G. Murphy2, and F. I. Makdisi3

ABSTRACT

Qaraoun Dam is a 60-meter-high concrete-faced rockfill embankment built in the 1960s and located at the southern end of the Bekaa Valley in central Lebanon. Lebanon, which is situated near the tectonic boundary of the African and Arabian plates, is within the seismically-active region of the Dead Sea fault system that includes many complex fault structures with a history of producing high levels of earthquake shaking in the region. A Probabilistic Seismic Hazard Analysis (PSHA) was conducted for Qaraoun Dam that included characterizing active faults and background seismicity areal source zones in the surrounding region. The results of the PSHA are presented in terms of peak ground acceleration and response spectral values for various specified annual frequencies of exceedance, or return periods. The Yammouneh fault, located approximately 2.4 km west of Qaraoun Dam, was found to contribute the most to the seismic hazard.

A comparison of the results of this study with previous studies of the seismic hazard for the region surrounding Qaraoun Dam highlights the importance of conducting a site-specific PSHA for critical structures. The comparison of results also emphasizes the need to periodically update seismic hazard evaluations for frequent (and less frequent, more damaging) earthquakes in order to evaluate whether the critical structures are designed for appropriate levels of ground shaking.

Uniform hazard response spectra, UHRS, were developed from the PSHA to represent the frequency content and level of shaking for several return periods. Deaggregation of the results of hazard analysis was used to select appropriate earthquake scenarios which were used to develop two sets of vertical and horizontal scenario acceleration time histories for the 100-year and 2,475-year return period levels. These time histories became the input for a dynamic response analysis to estimate potential earthquake-induced deformation of the dam.

1Geologist, AMEC Environment & Infrastructure, Inc., Oakland, CA 94612 2Engineer, AMEC Environment & Infrastructure, Inc., Oakland, CA 94612 3Principal Engineer, AMEC Environment & Infrastructure, Inc., Oakland, CA 94612 Johnson CB, Gilkerson DG, Makdisi FI. Seismic hazard analysis for Qaraoun Dam in the Bekaa Valley, Lebanon. Proceedings of the 10th National Conference on Earthquake Engineering, Earthquake Engineering Research Institute, Anchorage, AK, 2014.

Tenth U.S. National Conference on Earthquake Engineering Frontiers of Earthquake Engineering July 21-25, 2014 10NCEE Anchorage, Alaska

Seismic Hazard Analysis for Qaraoun Dam in the Bekaa Valley, Lebanon

C. B. Johnson1 , D. G. Murphy2, and F. I. Makdisi3

ABSTRACT

Uniform hazard response spectra, UHRS, were developed from the PSHA to represent the frequency content and level of shaking for several return periods. Deaggregation of the results of the hazard analysis was used to select appropriate earthquake scenarios which helped develop two sets of vertical and horizontal scenario acceleration time histories for the 100-year and 2,475-year return period levels. These time histories became the input for a dynamic response analysis to estimate potential earthquake-induced deformation of the dam.

Introduction

Qaraoun Dam is located on the Litani River in the Bekaa Valley approximately 40 kilometers southeast of the Lebanese capital city of Beirut. The dam is a critical structure that provides storage for the production of electricity and irrigation water for the Bekaa Valley. The dam is a concrete-faced rockfill embankment with a maximum height of 60 meters and a length of 1,100

Johnson CB, Gilkerson DG, Makdisi FI. Seismic hazard analysis for Qaraoun Dam in the Bekaa Valley, Lebanon. Proceedings of the 10th National Conference on Earthquake Engineering, Earthquake Engineering Research Institute, Anchorage, AK, 2014. meters. The purpose of this paper is to present a seismic hazard analysis for the site, site-specific earthquake response spectra for the return periods of interest (100 and 2,475 years), and compare the results with previous studies. A more complete description of the work is presented in AMEC (2012) [1].

Tectonic Setting

Qaraoun Dam is located along the Dead Sea fault system, which is a 1000-kilometer-long left- lateral transform fault system that forms the northwestern part of the boundary between the African plate on the west and the Arabian plate on the east (Fig. 1). The Dead Sea fault system stretches between the Gulf of Aqaba, at the northern edge of the Red Sea, to the Taurus Mountains in southern Turkey. The southern and northern portions of the fault system are primarily expressed as prominent north-northeast-trending transform faults connected by a series of fault-bounded depressions. In the central portion of the Dead Sea fault system the Lebanese restraining bend, approximately 150 km long, is composed of left-lateral faults trending about North 25 degrees East. Compression within the restraining bend is accommodated by northwest and southeast-dipping reverse faults that have formed several prominent ranges. Qaraoun Dam is located in the southern part of the Lebanese restraining bend, just east of the main left-lateral strike-slip fault of the Dead Sea fault system (Fig. 1). Detailed studies of offset global positioning system (GPS) data indicate a slip rate between about 3 and 6 mm/yr across the Dead Sea fault system.

Figure 1. Left: Regional tectonic setting and independent seismicity (2150 B.C. to A.D. 2011); thick gray arrows show plate motion relative to a stable Eurasian plate. Right: Active faults within 100 km of the site included in the model.

Seismicity

The region near Qaraoun Dam has been affected by numerous moderate to large magnitude earthquakes related to activity along the tectonic plate boundary between the African and Arabian plates. The earthquake catalog used in this analysis was compiled from several global catalogs (e.g., ANSS, NEIC, IRIS) and a regional catalog of historic 20th century events [2]. A combined catalog was developed by merging these catalogs and removing duplicate events. The catalog events were all converted to a common moment magnitude (MW) and dependent events (aftershocks and foreshocks) were removed. The resulting catalog covers the time period A.D. 115 to June 2011 (Fig. 1). Details of the catalog compilation are presented in AMEC (2012) [1].

Source Characterization

Quaternary-active (i.e. from 2.6 Ma to the present) faults and historic background areal seismic source zones were included in the hazard model. The key source characteristics used to develop input parameters for the active faults that are within approximately 100 km from the site are summarized in Table 1, along with weights identified for each fault parameter.

Table 1. Characterization of fault sources. Numbers in brackets are subjective probability weights.

Fault Distance Total Rupture Slip Rate Maximum Magnitude to Site Length Length (mm/yr) Mean and Range (km) (km) (km) Yammouneh 2.4 297 80 [0.3] 4.0 [0.2] 7.2 6.8-8.2 (strike-slip) 100 [0.5] 5.0 [0.4] 150 [0.15] 6.0 [0.3] 297 [0.05] 6.5 [0.1] Roum 18 55 40 [0.2] 0.7 [0.2] 6.9 6.5-7.2 (strike-slip) 55 [0.8] 1.0 [0.4] 1.4 [0.4] Rachaya 14 51 40 [0.2] 1.0 [0.3] 6.8 6.4-7.1 (strike-slip) 51 [0.8] 1.3 [0.4] 1.5 [0.3] Serghaya 28 122 60 [0.2] 1.0 [0.3] 7.1 6.7-7.6 (strike-slip) 80 [0.6] 1.5 [0.5] 122 [0.2] 2.0 [0.2] Mount 32 138 100 [0.2] 1.0 [0.25] 7.4 7.1-7.7 Lebanon 125 [0.6] 1.5 [0.4] (reverse) 138 [0.2] 2.0 [0.3] 4.0 [0.05] Jordan Valley 51 174 100 [0.2] 2.0 [0.1] 7.3 6.9-7.9 (strike-slip) 120 [0.4] 3.0 [0.2] 150 [0.3] 4.0 [0.4] 174 [0.1] 6.0 [0.3]

The shallow (less than 35 km deep) earthquakes that cannot be spatially attributed to the mapped active faults shown on Fig. 1 are modeled as separate areal source zones, shown as polygons on Fig. 2. The size and extent of the areal source zones were delineated based on prominent geologic structures and consistent patterns of seismicity. In general, the boundaries between these morphostructural regions correspond to changes in the spatial pattern and magnitude distribution of earthquakes, such that zones are inferred to exhibit uniform seismicity.

Figure 2. Crustal seismicity and areal source zones. Orange-filled events that are attributed to the active modeled faults are removed from recurrence calculations. See Fig. 1 for remainder of explanation.

The seismicity in the combined earthquake catalog is used to estimate the earthquake recurrence parameters for each of the areal source zones. Eight crustal areal seismic source zones that lie within about 300 km of the site were identified and modeled in the PSHA. The maximum reported earthquake in each zone and the assigned maximum magnitude distribution and weights are listed in Table 2.

Table 2. Maximum magnitude distributions for areal seismic source zones. Numbers in brackets are subjective probability weights. Seismicity from fault sources are removed from the recurrence calculations and are not considered in the magnitude distributions.

Zone Maximum Maximum Magnitude Reported Distribution Earthquake (MW) Lebanese Restraining Bend 4.9 (1991) 6.2 [0.2], 6.6 [0.6], 6.9 [0.2] Jordan Valley Zone 6.4 (A.D. 362) 7.4 [0.3], 7.6 [0.4], 7.8 [0.3] Levantine Basin 5.0 (1970 & 1984) 6.0 [0.2], 6.3 [0.6], 6.6 [0.2] Cyrus Zone 6.7 (1996) 7.6 [0.2], 7.8 [0.6], 8.0 [0.2] Cyprus Transform 7.5 (A.D. 115) 7.6 [0.3], 7.8 [0.4], 8.0 [0.3] Sinai Peninsula 5.7 (1903) 6.5 [0.2], 6.7 [0.6], 6.9 [0.2] Syria 5.2 (1996) 5.8 [0.2], 6.1 [0.6], 6.4 [0.2] Arabian Shelf 6.7 (1182) 6.8 [0.2], 7.0 [0.6], 7.2 [0.2]

Earthquake Ground Motions

The probabilistic seismic hazard analysis (PSHA) was performed to characterize earthquake ground shaking that may occur at the dam site during future seismic events in the region, using seismic hazard codes (software programs) developed by AMEC. The PSHA is based on an assessment of the recurrence of earthquakes on and within potential seismic sources in the greater Lebanon region, and on ground motion attenuation relationships appropriate for the types of seismic sources in the region. The parameters needed for each seismic source include the maximum magnitude and the rate of occurrence (frequency) of earthquakes for a range of magnitudes (earthquake recurrence relationships). The uncertainties in identifying seismic sources and identifying their model parameters are represented by a logic tree and used in this PSHA.

The maximum magnitude, MMAX, for a seismic source represents the largest possible earthquake for that source, regardless of its frequency of occurrence. Thus MMAX defines the upper limit of the earthquake recurrence relationship for the source. For fault sources, the fault geometry, rupture length, and seismogenic depth were used to evaluate MMAX using appropriate empirical relationships [3,4]. For areal source zones, the maximum historical recorded earthquake, and observed maximum events in similar types of crust and tectonic setting, were used to estimate a range of MMAX magnitudes and associated weights (Table 2). The recurrence of earthquakes on fault sources is modeled using the “characteristic earthquake” recurrence model [5]. The earthquake recurrence for the areal source zones is represented with a truncated exponential model [6]. The earthquake recurrence rates and Gutenberg-Richter b- values were estimated using the observed seismicity (from the independent catalog) that occurred within each zone.

Ground Motion Prediction Equations

A ground motion prediction equation (GMPE) computes the amplitudes of peak acceleration and response spectral acceleration at the dam site for a specified earthquake magnitude, source-to- site distance, site conditions, and style of faulting. The site conditions are represented by the shear wave velocity of the underlying subsurface rock or soil. The model was run for a VS30 (shear wave velocity within the upper 30 m) of 1,100 m/s, consistent with the National Earthquake Hazards Reduction Program (NEHRP) Site Class B designation for rock site conditions. Equal weights were assigned for each set of GMPEs used in the model (i.e., crustal, stable continent, oceanic crust). The Next Generation Attenuation (NGA) models [7,8,9,10,11] along with two region-specific attenuation models [12,13] were used to estimate the ground motions at the Qaraoun Dam site for all of the modeled crustal faults and crustal areal source zones, except for the Levantine Basin, Syria, and Arabian Shelf zones. Two GMPEs developed for stable continental regions [14,15] were used for the crustal areal source zones of the Arabian Peninsula (Syria and Arabian Shield zones). Attenuation relationships developed for intraslab earthquakes are based on events that have occurred in the oceanic crust, therefore attenuation models appropriate for calculating ground motions for intraslab earthquakes are used for the oceanic crust of the Levantine Basin [16,17,18].

Ground Motion Hazard Analysis

The mathematical formulation used in most PSHAs assumes that the occurrence of damaging earthquakes can be represented as a Poisson process. Under this assumption, the probability that a ground motion parameter, Z, will exceed a specified value, z, in time period t is given by:

P(Z > z|t) = 1=e-v(z)t ≤ v(z)t (1)

where v(z) is the average frequency during the time period, t, at which the level of ground motion parameter Z exceeds value z at the site, from all earthquakes on all sources in the region. Eq. 1 is valid provided that v(z) is the appropriate average value for time period t. The frequency of exceedance, v(z), is computed by using the input source, magnitude range, and site parameters discussed previously, together with the ground motion predication equations. The GMPEs define the level of ground motion in terms of a lognormal distribution. In the hazard computations, the fault sources were modeled by segmented planar surfaces. The source zones were modeled by closely-spaced, planar pseudo-faults distributed throughout the zone with preferred azimuth orientations consistent with geologic structures within the zone. Earthquakes for both types of sources were represented by rectangular rupture planes for the given magnitude earthquake using appropriate rupture area relationships [3,4]. At each ground motion level, the complete set of results forms a discrete distribution for frequency of exceedance, v(z). The computed distributions were used to obtain the mean frequency of exceeding various levels of peak ground motion (mean hazard curves).

Probabilities of Exceedance

The mean total hazard curve (from all sources) for peak ground acceleration (PGA) is shown on Fig. 3. Also shown on the figure are contributions from each individual seismic source. The plot is presented in terms of the peak ground acceleration (PGA) versus its mean annual frequency of exceedance. Fig. 3 shows that the PGA hazard is dominated by the Yammouneh fault located at a closest distance of about 2.4 km from the dam. PSHA results for Qaraoun Dam were calculated for PGA and response spectral accelerations at ten spectral periods (between 0.075 and 4.0 seconds). The mean hazard curves for each spectral period were interpolated to obtain values of spectral acceleration associated with return periods of 100, 475, 975, 2,475, and 10,000 years (Fig. 3). The return period in years is the inverse of the annual frequency of exceedance. These values were then connected by smooth curves to produce five-percent-damped uniform hazard response spectra (UHRS).

Figure 3. Probabilistic seismic hazard analysis results. Left: Mean total hazard and source contributions at PGA. Right: Five-percent-damped uniform hazard response spectra.

Comparison of Results with Other Studies

Commonly, available reports and maps of seismic hazard focus on PGA at return periods (RP) of 475 years and 2,475 years. For comparison, the PGA values calculated in this PSHA are 0.40g for a 475-year RP, and 0.77g for a 2,475-year RP. Studies of seismic hazard for the Mediterranean or Arabian region indicate that the dam site is in an area with a range of PGA values from 0.2g to 0.3g for a 475-year RP [19,20]. Regional studies are typically based on background seismicity zones and do not often include the effects of discrete faults. Several published studies of seismic hazard for Lebanon report much lower values of PGA hazard for the 475-year RP. However, recent studies of the Yammouneh fault indicate a higher slip rate than previously attributed to the fault. More recent studies of seismic hazard for Lebanon, that incorporate the updated Yammouneh fault slip rate, indicate the site is in a zone with PGA values of about 0.2g to 0.35g [21,22]. One study reports a PGA estimate at the site of 0.3g to 0.35g for a 2,500-year RP [21]. Differences between previous ground motion estimates, and the values reported in this study, likely result from: a) the use of different GMPEs, b) how crustal faults are modeled, and c) differences in the development and use of the seismicity catalog.

Time Histories and Deformation Analyses

Time histories representative of crustal sources were selected for use in dynamic response analyses to estimate the potential for earthquake-induced deformation of the dam [23]. The time histories consist of actual strong motion recordings from the Pacific Earthquake Engineering Research Center (PEER) NGA database that have been rotated into fault-normal and fault- parallel orientations, and were modified to represent the horizontal and vertical 100-year and 2,475-year RP design response spectra.

Prior to selection of original (seed) time histories, fault directivity was addressed and vertical response spectra were developed as part of this study. The UHRS were modified to account for near-field fault directivity effects. However, the factors developed for an event on the Yammouneh fault are all near 1.0, causing very little change in the UHRS curves. A detailed description of the fault directivity effects and the development of the vertical response spectra is included in AMEC (2012) [1].

Figure 4. Deaggregation of PGA Hazard for the 100-year RP [left] and 2,475-year RP [right].

In order to select appropriate time histories, the magnitude-distance contributions to the hazard (deaggregation) were assessed to determine the specific earthquake scenarios of interest. Fig. 4 shows the deaggregation of hazard for PGA for ground motions at return periods of 100-years (annual frequency of 0.01) and 2,475-years (annual frequency of 0.0004). Deaggregation of hazard for additional spectral periods and return periods was assessed in AMEC (2012) [1]. Based on the magnitude and distance contributions, the hazard at the site is dominated by a strike-slip event on the Yammouneh fault in the range of MW 7 to 8, at a distance of 2.4 km. Because of limited space, additional details and plots of the original and modified time histories are included in AMEC (2012) [1].

Conclusions

Qaraoun Dam is located near the Arabian-African plate boundary, in a region where significant, damaging earthquakes have been known to have occurred in the past. Because of the importance of the structure’s integrity, we have completed a seismic hazard analysis for the dam including compilation of UHRS for various return periods of interest, deaggregation of hazard results, and matching of time histories to horizontal and vertical response spectra. We compared our PSHA results with previous site-specific and regional studies of seismic hazard and found that where studies do not fully address all aspects of the potential sources (i.e., inclusion of faults) or inputs (e.g., use of single GMPE), results might underestimate the hazard. In addition, it is important to update the seismic hazard model and PSHA results following changes and updates to the characterization of the seismic sources. The results of this study indicate the need to conduct appropriate site-specific seismic hazard analyses for critical structures (e.g., dams, hospitals, schools), rather than relying solely on regional studies.

References

1. AMEC Environment & Infrastructure. Lebanon Litani River Basin Management Support Program, Seismic Deformation Analysis of Qaraoun Dam, prepared with the International Resources Group (IRG) for the Lebanon LRBMS program, February 2012. 2. Khair K. Geomorphology and seismicity of the Roum fault as oneo f the active branches of the Dead Sea fault system in Lebanon. Journal of Geophysical Research 2001; 106 (B3): 4233-4245. 3. Wells DL, Coppersmith KJ. New empirical relationships among magnitude, rupture length, rupture area, and surface displacement. Bulletin of the Seismological Society of America 1994; 84: 974-1002. 4. Hanks TC, Bakun WH. M-log A Observations of Recent Large Earthquakes. Bulletin of the Seismological Society of America 2008; 98(1): 490-503. 5. Youngs RR, Coppersmith KJ. Implications of fault slip rates and earthquake recurrence models to probabilistic seismic hazard estimates. Bulletin of the Seismological Society of America 1985; 75: 939-964. 6. Cornell CA, Van Marcke EH. The major influences on seismic risk. Proceedings of the Third World Conference on Earthquake Engineering, Santiago, Chile, 1969; A-1: 69-93. 7. Abrahamson N, Silva W. Summary of the Abrahamson & Silva NGA ground-motion relations. Earthquake Spectra 2008; 24(1): 67-97. 8. Boore DM, Atkinson GM. Ground-motion prediction equations for the average horizontal component of PGA, PGV, and 5% damped PSA at spectral periods between 0.01 s and 10.0 s. Earthquake Spectra 2008; 24(1): 99- 138. 9. Campbell KW, Bozorgnia Y. NGA ground motion model for the geometric mean horizontal component of PGA, PGV, PGD and 5% damped linear elastic response spectra for periods ranging from 0.01 to 10 s. Earthquake Spectra 2008; 24(1): 139-171. 10. Chiou BS-J, Youngs RR. An NGA model for the average horizontal component of peak ground motion and response spectra. Earthquake Spectra 2008; 24(1): 173-215. 11. Idriss IM. An NGA empirical model for estimating the horizontal spectral values generated by shallow crustal earthquakes. Earthquake Spectra 2008; 24(1): 217-242. 12. Ambraseys NN, Douglas J, Sarma SK, Smit P. Equations for the estimation of strong ground motions from shallow crustal earthquakes using data from Europe and the Middle East: Horizontal peak ground acceleration and spectral acceleration. Bulletin of Earthquake Engineering 2005; 3(1): 1-53. 13. Akkar S, Bommer JJ. Empirical Equations for the Prediction of PGA, PGV, and Spectral Accelerations in Europe, the Mediterranean Region, and the Middle East. Seismological Research Letters 2010, 81(2): 195-206. 14. Atkinson GM, Boore DM. Earthquake Ground-Motion Prediction Equations for Eastern North Americas. Bulletin of the Seismological Society of America 2006, 96: 2181–2205. 15. Campbell KW. Prediction of Strong Ground Motion Using the Hybrid Empirical Method and Its Use in the Development of Ground-Motion (Attenuation) Relations in Eastern North America. Bulletin of the Seismological Society of America 2003; 93(3): 1012-1033. 16. Atkinson GM, Boore DM. Empirical Ground-Motion Relations for Subduction-Zone Earthquakes and Their Application to Cascadia and Other Regions. Bulletin of the Seismological Society of America 2003; 93: 1703- 1729. 17. Garcia D, Singh SK, Herraiz M, Ordaz M, Pacheco JF. Inslab Earthquakes of Central Mexico: Peak Ground- Motion Parameters and Response Spectra. Bulletin of the Seismological Society of America 2005; 95(6): 2272- 2282. 18. Zhao JX, Zhang J, Asano A, Ohno Y, Oouchi T, Takahashi T, Ogawa H, Irikura K, Thio HK, Somerville PG, Fukushima Y, Fukushima Y. Attenuation Relations of Strong Ground Motion in Japan Using Site Classification Based on Predominant Period. Bulletin of the Seismological Society of America 2006; 96: 898-913. 19. Giardini D, Grunthal G, Shedlock K, Zhang P. Global Seismic Hazard Map. Produced by the Global Seismic Hazard Assessment Program (GSHAP), a demonstration project of the UN/International Decade of Natural Disaster Reduction, conducted by the International Lithosphere Program, Robinson Map Projection, scale 1:35,000,000 at the equator, 1999. 20. Shapira A, Hofstetter R. Earthquake Hazard Assessments for Building Codes Final Report. Proposal No. M18- 057, Grant No. PCE-G-00-99-00038, report prepared for U.S. Agency for International Development, Bureau for Economic Growth, Agriculture and Trade, 2007. 21. Elnashai AS-E, El-Khoury R. Earthquake Hazard in Lebanon. Imperial College Press: London, 2004. 22. Jimenez M, Al-Nimry H, Khasawneh A, Al-Hadid T, Kahhaleh K. Assessment of Seismic Hazard in Jordan. First European Conference on Earthquake Engineering and Seismology, Geneva, Switzerland, September 2006. 23. Phalen JD, Makdisi FI, Hu J, Viala E, Amacha N. Risk of Seismic Deformation of a 1960’s Rockfill Dam in Lebanon. Proceedings of the 10th National Conference on Earthquake Engineering, Earthquake Engineering Research Institute, Anchorage, AK, 2014.