FINITE FAULT MODELING OF FUTURE LARGE EARTHQUAKE FROM NORTH TEHRAN FAULT IN KARAJ, IRAN

Transcription

1 FINITE FAULT MODELING OF FUTURE LARGE EARTHQUAKE FROM NORTH TEHRAN FAULT IN KARAJ, IRAN Meghdad SAMAEI 1, Masakatsu MIYAJIMA 2, Hamid SAFFARI 3, Masato TSURUGI 4 1 Doctoral Student, Graduate School of Natural Science and Technology, Kanazawa University (Kakuma-machi, Kanazawa, Ishikawa , Japan) samaei@stu.kanazawa-u.ac.jp 2 Member of JSCE, Professor, Graduate School of Natural Science and Technology, Kanazawa University (Kakuma-machi, Kanazawa, Ishikawa , Japan) miyajima@t.kanazawa-u.ac.jp 3 Doctoral Researcher, Graduate School of Natural Science and Technology, Kanazawa University (Kakuma-machi, Kanazawa, Ishikawa , Japan) hamid.saffari@gmail.com 4 Senior researcher, Geo-Research Institute (4-3-2, Itachibori, Nishi-ku, Osaka, Japan) tsurugi@geor.or.jp The main purpose of this study is to predict strong ground motions from future large earthquake for Karaj city, the capital of Alborz province of Iran. This city is an industrialized city having over one million populations and is located near several active faults. Finite fault modeling with a dynamic corner frequency has adopted here for simulation of future large earthquake. Target fault is North Tehran fault with the length of 110 km and rupture of west part of the fault which is closest to Karaj, assumed for this simulation. For seven rupture starting points, acceleration time series in the site of Karaj Caravansary historical building- are predicted. Peak ground accelerations for those are vary from 423 cm/s 2 to 584 cm/s 2 which is in the range of 1990 Rudbar earthquake (M w =7.3). Results of acceleration simulations in different distances are also compared with attenuation relations for two types of soil. Our simulations show general agreement with one of the most well known world attenuation relations and also with one of the newest attenuation relation that hase developed for Iranian plateau. Key words: Strong ground motion prediction, Finite fault modeling, Karaj, North Tehran fault 1. INTRODUCTION Karaj city, the capital of the Alborz province with the population of 1,377,450 in the 2006 census, is the fifth-largest city in Iran. However, this city is increasingly becoming an extension of metropolitan Tehran. Karaj, Tehran and surrounding faults are shown in Fig. 1. As it is seen; the west part of the North Tehran fault passes through Karaj. So it is important to estimate the earthquake ground motion caused by this fault in the city. North Tehran fault with the length of 110 km which is counted as a thrust fault (Tchalenko et al. 1) and Berberian et al. 2) ) is the target fault in this study. The devastating earthquakes of 855 to 856 (exact year is uncertain) and 1177 could have been caused by this fault (Berberian 3) ). Strong ground motion will be predicted in the site of Karaj Shah Abbasi Caravansary which is a historical building located in the down town and is one of the oldest structures in the city. This caravansary belongs to Safavid dynasty and built between 1688 to 1698 A.C. It is one of the hundreds caravansaries along the Silk Road. Caravansaries had been roadside inns where travelers could rest and recover from the day's journey. They supported the flow of commerce, information, and people across the network of trade routes. Results of strong motion simulations in different distances will also be compared with an attenuation relation that has developed for Iran plateau and also one of the most popular attenuation relations that has developed for world. This comparison will carried out for two types of soil. I_20

2 Fig. 1 Karaj, Tehran and surrounding faults Fig. 2 Karaj Shah Abbasi Caravansary 2. FINITE FAULT MODELING WITH DYNAMIC CORNER FREQUENCY One of the most useful methods to simulate ground motion for a large earthquake is based on the simulation of several small earthquakes as subevents that comprise a large fault-rupture event. This idea for the first time was introduces by Hartzell 4) ; when he used this method to model the El Centro displacement record for the 1940 Imperial Valley earthquake. In this method a large fault is divided into N subfaults and each subfault is considered as a small point source. The contributions of all point sources are summed in the observation point and large event is produced. One of the simple solutions for this method was proposed by Beresnev and Atkinson 5). In this solution target fault is divided to nl nw (=N) subfaults with unique dimensions. For each subfault ground motion is being produced by stochastic 2 source point method with an ω model. By considering the effects of path and site, the produced motions for all subfaults are summed in the observation point with a proper time delay to obtain the ground motion acceleration from the entire fault: (1) where Δtij is relative delay time for the radiated wave from the ijth subfault to reach the observation point and aij is the subevent (motion) which coincides with the ijth subfault. Each a ij (t) is calculated by the stochastic point-source method of Boore 6),7). In Boore s method the acceleration I_21

3 spectrum for a subfault at a distance R ij is modeled as a point source with an ω 2 shape. The acceleration spectrum of shear wave of the ijth subfault, A ij (f), is described by: (2) where M 0ij, f 0ij, and R ij are the ijth subfault seismic moment, corner frequency, and distance from the observation point, respectively. The constant C= θφ 4FV/(4πρβ 3 ), where θφ is radiation pattern (average value of 0.55 for shear waves), F is free surface amplification (2.0), V is partition onto two horizontal components (0.71), ρ is density, and β is shear wave velocity. f 0ij is the corner frequency of the ijth subfault. D(f) is the site amplification and the term exp(-πfκ 0 ) is a high-cut filter to model near surface κ 0 effects. The quality factor, Q(f), is inversely related to anelastic attenuation. The implied 1/R geometric attenuation term is applicable for body-wave spreading in a whole space. The approach of the stochastic point-source model is to generate a transient time series that has a stochastic character, and whose spectrum matches a specified desired amplitude spectrum such as that given by Equation (2). Steps for doing this are illustrated in Figure 3. First, white Gaussian noise is generated for the duration of the motion (Fig. 3a). Duration of the motion can generally be represented as: T(R)=T 0 +dr, where T 0 is the source duration and d is the coefficient controlling the increase of duration with distance which is derived empirically (Atkinson and Boore 8) ). This noise is then windowed (Fig. 3b); the windowed noise is transformed into the frequency domain (Fig. 3c); the spectrum is normalized by the square-root of the mean square amplitude spectrum (Fig. 3d); the normalized spectrum is multiplied by the ground motion spectrum Y (Fig. 3e); the resulting spectrum is transformed back to the time domain (Figure 3f). The corner frequency of the ijth subfault, f 0ij (t), is defined as: / / / (3) where the average seismic moment of subfaults is M 0ave = M 0 /N. In identical subfaults, the moment of each subfault is controlled by the ratio of its area to the area of the main fault (M 0ij = M 0 /N, where M 0 is the seismic moment of the entire fault). If the subfaults are not identical the seismic moment of each subfault is expressed as: // (4) where S ij is the relative slip weight of the ijth subfault. It should be noted here that in this method, a subevent is produced with just one set of random numbers. (unlike the other methods that might use several sets of random numbers for producing several small earthquakes and use the average of them as a subevent.) Fig. 3 Basis of the time-domain procedure for simulating ground motions using the stochastic method. (From Boore 7) ) Despite all advantages of the presented approach by Beresnev and Atkinson 5), Motazedian and Atkinson 9) showed that the received energy at the observation point is very sensitive to subfault sizes e.g. as the subfault sizes are increased, the energy at low frequency is decreased and the energy at high frequencies is increased. They overcame this problem by introducing a dynamic corner frequency. In their model, the corner frequency is a I_22

4 function of time, and the rupture history controls the frequency content of the simulated time series of each subfault. The rupture begins with a high corner frequency and progresses to lower corner frequencies as the ruptured area grows. In this approach the acceleration spectrum of the shear wave of the ijth subfault, A ij (f), is described by (5) where the dynamic corner frequency, f 0ij (t), is defined as a function of the cumulative number of ruptured subfaults, N R (t), at time t: / / / (6) H ij is a scaling factor for conserving the total area under the spectrum of subfaults as the corner frequency decreases with time which is: / / (7) Stochastic finite fault modeling based on a dynamic corner frequency is adopted in this study so the success of the simulation is not dependent to the subfault size any more. This new model can also implement the concept of pulsing area. Pulsing area is the active area on the fault that contributes to the dynamic corner frequency at a given moment in time, and therefore the passive area does not affect the dynamic corner frequency. The cumulative number of pulsing subfaults (N R in Equation (6)), would increase with time at the beginning of rupture but become constant at a fixed percentage of the total rupture area which is the percentage of the pulsing area. This is in agreement with generally accepted idea that the rise time of subfaults are much smaller than the duration of fault rupture. 3. ANALYSIS AND RESULTS As it pointed earlier, the target fault for our simulation is North Tehran fault. The location of the fault is decided based on the map of major active faults of Iran, published by IIEES 10). The thrust dip of this fault is highly variable from 10 to 80 degrees towards north (Tchalenko et al. 1) ). In the west part dip is roughly estimated to be about 35 (Berberian et al. 2) ) As it is seen in the Fig. 1, North Tehran Fault is divided into 3 segments; A, B and C. These 3 segments are Karaj-Mahdasht, Kan and Lashgarak segments respectively. In this study earthquake accelerograms are predicted for rupture of A segment of the fault which is closest to Karaj. The upper depth and lower depth are decided from recorded events (with M>2.5) of Institute of Geophysics, University of Tehran (IGTU) and also Table 1 Recorded data on North Tehran Fault Row Ref Date Time Latitude N Longitude E Depth Magnitude 1 IGTU 2/21/ :5: (Mn) 2 IGTU 8/26/ :50: (Mn) 3 IGTU 11/4/ :16: (Mn) 4 IGTU 1/24/2001 6:18: (Mn) 5 IIEES 2/10/2001 2:19: (Ml) 6 IGTU 10/2/ :37: (Mn) 7 IIEES 9/9/2004 1:36: (Ml) 8 IGTU 8/15/ :41: (Mn) 9 IGTU 12/18/2005 7:51: (Mn) 10 IIEES 8/15/2006 0:33: (Ml) 11 IGTU 8/15/2006 0:33: (Mn) 12 IGTU 7/31/2007 3:8: (Mn) 13 IGTU 7/31/2007 3:18: (Mn) 14 IIEES 7/31/2007 6:20: (Ml) 15 IGTU 7/31/2007 7:9: (Mn) 16 IGTU 10/22/2008 6:12: (Mn) 17 IIEES 8/21/ :41: (Ml) 18 IGTU 4/1/2011 2:40: (Mn) I_23

5 International Institute of Earthquake Engineering and Seismology (IIEES) from 1998 to This data is shown in Table 1. All the recorded events have the hypocenteral depth between 3 to 18 kilometers so the siesmogenic depth decided to be from 3 to 18 kilometers. In finite fault method, modeling of the finite source requires the orientation and dimensions of the fault plane, the dimensions of subfaults and the location of the hypocenter. The parameters used in this study are listed in Table 2. Moment magnitude is calculated from empirical relationships of Wells and Coppersmith 11) for reverse faults by considering rupture area. Stress parameter, percentage of pulsing area, Q value and high cut filter are based on the work by Motazedian 12) for earthquakes in northern Iran. Geometric spreading used here is trilinear relationship that proposed by Kanamori and Anderson 13) which is widely used in Iran and other regions of the world 5),8),14),15). Crustal shear wave velocity and crustal density are taken from the work of Radjaee et al 16). They determined a model of the crust for the central Alborz Mountains using teleseismic receiver functions from data recorded on a temporarily network. Rupture velocity is considered to be 0.8 times of shear wave velocity 12),16),17). We don t have information of slip distribution on the fault; therefore unity slip rate for all subfaults has been used. In Fig. 4, geometry of the fault and location of Karaj Caravansary are shown. Seven points are assumed as rupture starting points and acceleration time histories for those points are drawn in Fig. 5. The soil condition of Caravansary estimated to be type II of Iranian code of practice for seismic resistant design of buildings, standard ) site classes (With the average shear wave velocity of 560 m/s). The site amplification factors employed here are those of Boore and Joyner 19) for various sites which are characterized by the average shear wave velocity. Table 2 Modeling parameters Fault orientation Strike 305 ; Dip 35 Fault dimensions along strike 46 by 26 km and dip Fault depth range 3 18 km Moment magnitude 7.1 Subfault dimensions 2 by 2 km Stress parameter 68 bars Number of subfaults 299 Q (f) 87f Geometrical spreading 1/R, R 70 km 1/70, 70 < R 130 km 1 130, R > R Windowing function Saragoni-Hart Kappa factor (High-cut filter) 0.05 Pulsing area 50% Crustal shear-wave velocity 3.5 km/sec Rupture velocity 0.8 shear wave velocity Crustal density 2.8 g/cm 3 Fig. 4 Geometry of the fault I_24

6 Fig. 5 Predicted acceleration time histories for one horizontal direction for different rupture starting points and observed record of L component, Rudbar Earthquake (R (i, j) implies the location of rupture starting point in ijth subfult) There is no recorded earthquake with a big magnitude on the North Tehran Fault; so here we show acceleration time history of Rudbar earthquake of 1990/06/20 with Mw=7.3 (Berberian and Walker 20), Sarkar et al 21) ) beside time histories of our simulations in Figure 5. This earthquake was chosen because it was one of the devastating earthquakes that occurred in the last decades in Iran and its accelerogram is available. In this earthquake 40,000 to 50,000 people were killed and extensive damage and landslides observed in the Rasht-Qazvin-Zanjan area. The shown accelerogram is the L component of the record at Abbar station of BHRC ground motion stations. Average shear wave velocity of upper 30 meter in this station; is 621 m/sec as Sinaeian et al. 22) (2007) report; therefore it is also characterized as type II of Iranian code of practice for seismic resistant design of buildings 18) I_25

7 Fig. 6 A seismotectonic map of the region of Rudbar earthquake from Sarkar et al 21). Earthquake epicenter is shown with a star (*). The meisoseismal area, three fault segments, viz. Baklor (B), Kabateh (K) and Zard Goli (ZG), locations of the Abbar accelerograph station and the three completely devastated towns of Rudbar, Manjil and Lowsan are also identified. The regions near Lahijan from where soil liquefaction was reported are marked with L (standard 2800) site classes like the site of Karaj Caravansary. The seismotectonic map of the region of Rudbar earthquake is shown in Fig. 6. There are some similarities between Rudbar earthquake and our simulations. We have similar moment magnitude (7.1 and 7.3) and similar hypocentral distance (The seismogram recorded in Rudbar earthquake has hypocenter distance of 43 km and our simulations have hypocentral distances of kilometers). The similarity is also with the amplification of the site; as it pointed before, both sites are characterized as type II of Iranian code of practice for seismic resistant design of buildings 18). These similarities in input data have lead to similar output which here is PGA. As it can be seen in the Figure 5, we have the peak accelerations from 423 cm/s/s to 584 cm/s/s which are in the range of Rudbar earthquake (505 cm/s/s). The important difference between our simulations and Rudbar earthquake is the time duration of the motion. Our simulations have the time duration of 15 to 25 seconds while this for Rudbar earthquake is more than 50 seconds. The reason for big difference in time duration is the difference between the size and shape of the faults. We have generated motions for rupture of a rectangular fault with the length of 46 km while Rudbar earthquake has been associated by three segment discontinues surface rupture with total length of 80 km 20),21),23). These three segments are Baklor, Kabateh and Zard Goli and are shown in Fig. 6. These segments are arranged in rightstepping en-echolon pattern and are separated by gaps in the observed surface rupture. Three main subevents that can be seen in seismic body waves are correspond to these three main segments of surface rupture 20),21). We should point out here that the difference of the PGAs for different cases is mostly because of the stochastic process of generating the wave. Directivity effect also plays an important role on the produced PGA. Directivity effect for the point R(20,8) is quiet obvious; Although we have a further hypocentral distance (33 km) for this point compared to points R(1,13), R(4,9), R(8,10), R(12,7) and R(16,9), we see a bigger PGA on the time series of acceleration. This is because as s-waves travel from this point as rupture starting point along the fault, they are reinforced by waves generated by newly ruptured segments. Accordingly we observe bigger acceleration despite the further distance. This is also considerable for the point R(23,13) which is the furthest rupture starting point (with 44 km of distance) but yields the second biggest PGA. This issue can also be noticed trough time duration of generated motions. Generated motions of R(20,8) and R(23,13) have shorter durations compared to R(1,13), R(4,9) and R(8,10) I_26

8 10000 SA(cm/s/s) Natural Period (s) Fig. 7 5% damped of acceleration response spectra for estimated time series (black lines) and one of Rudbar earthquake (blue line) which shows that, for R(20,8) and R(23,13) generated waves for different subfaults reach to the receiver rather together. For further comparison, 5% damped of acceleration response spectra of estimated time series and observed record of Rudbar earthquake is also shown in Fig. 7. The figure shows that response spectra of our simulations are very close to the response spectra of observed Rudbar earthquake. Therefore the effects of simulated earthquakes on regular buildings are the same as Rudbar earthquake. Comparison shows similarity of peak acceleration and response spectra of our simulation with the actual earthquake of Rudbar. This similarity of results is in company with similarity of input factors like magnitude, hypocentral distance and soil condition of the site. Based on this comparison we can conclude that an earthquake which occurs in Karaj can be as destructive as Rudbar earthquake so considering this hazard of earthquake for construction of new structures and also retrofitting of historical buildings like Karaj Caravansary is necessary. horizontal components of motion in different distances for two types of soil conditions. Soil conditions we considered here are soil type II and III of Iranian code of practice for seismic resistant design of buildings 18) (standard 2800) with VS30 =560 m/s and VS30 =275 m/s respectively. (VS30 is the time-averaged shear-wave velocity in the top 30 m of the site profile.) Site amplification curves versus frequency for these types of soil are derived from Boore and Joyner 19) curves of NEHRP (National Earthquake Hazard Reduction Program) site classes by linear interpolation. A unity slip distribution is assumed and the same seven locations of the hypocenter on the fault plane of Figure 4 are considered. We simulate records for distances from surface projection of the fault ranging from 1 to 400. (Note, the actual distances from surface projection of the fault are as follows: 1, 2, 5, 10, 15, 20, 30, 40, 50, 60, 70, 80, 100, 120, 150, 200, 250, 300 and 400 km.) Eight lines at equally spaced azimuths spreading out from a point above the center of the top of the fault plane were defined to capture the average effect of directivity; the geometry of the simulated points is shown in Fig COMPARISON WITH ATTENUATION RELATIONS Since in recent years no earthquake of target moment has happened around Karaj and thus no strong-motion recordings are available to compare the results of the simulation with actual data, we chose to draw a comparison with attenuation laws developed for Iranian plateau or similar regions. Attenuation relationships estimate ground motion as a function of magnitude and distance. Using the stochastic finite-fault model of Motazedian and Atkinson 9) with the model parameters listed in Table 2, we generated random Fig. 8 Geometry of sites for producing synthesized earthquakes. Locations of sites step out from a point above the center of the fault plane, along eight lines equally spaced in azimuth. Only one half of the focal sphere is shown in the figure. I_27

9 1000 Soil type II (Vs30 =560 m/s) 1000 Soil type III (Vs30 =275 m/s) PGA (cm/s/s) 100 PGA (cm/s/s) Distance (Km) Distance (Km) Soil type II (Vs30 =560 m/s) Soil type III (Vs30 =275 m/s) PGV (cm/s) 10 PGV (cm/s) Distance (km) Distance (km) Fig. 9 Maximum acceleration and velocity from simulations versus closest distance to the rupture of the fault (dots) in comparison with two attenuation models; first: Campbell and Bozorgnia (blue line), second: Saffari (red line) Fig. 9 plots PGA and PGV from simulations versus closest distance to the rupture on the fault in comparison with the two empirical attenuation models. First attenuation model we chose here is well known model of Campbell and Bozorgnia 24) which is developed for world earthquakes. Second one is model of Saffari et al 25) which is newest model that has developed for Iranian plateau. Curves are drawn for two types of soil profile with V S30 =560 and V S30 =275 and for every soil profile total number of 1064 simulations have done (19 distances 8 azimuthally equal lines 7 rupture starting points). Saffari et al s curve is not recommended for distances below 15 km because they had few data for regression in those distances, but we have brought it here for a better comparison. (The curve is shown with dashed line in this range.) It should be noted here that our definition of distance here is M4 of the distances defined by Joyner and Boore 26). As can be seen in the Figure 9, our simulations show a general agreement with both curves especially with the one of Campbell and Bozorgnia, although the shapes of the curves are different. The I_28

10 difference of the shapes is expected because our simulations are based on a seismological model which is a theoretical issue while the attenuation relations are based on regression of observed data which are empirical. This is shown by other researches too 8),17),27). For PGA, our simulations show a bit of smaller amounts in distances less than 70 kilometers compared to both attenuation models; this difference is more pronounced in distances between 10 and 70 km. However for PGV, simulations with finite fault modeling show very good agreement with attenuation models in distances less than 70 km. It can be noticed that geometrical spreading has an important effect on the shape of the arrangement of dots in Figure 9. The constant value of PGA or PGV can almost be seen clearly between the distances 70 to 130 km. A better estimation of geometric spreading in Iran can make the results of finite fault simulation closer to attenuation models. 5. CONCLUSIONS Acceleration time series of strong ground motion are predicted for Karaj city in the site of Karaj Caravansary. Finite fault modeling with a dynamic corner frequency adopted for simulation. Target fault is North Tehran fault with the length of 110 km and the rupture of west part of it which is closest to Karaj, assumed for this simulation. For seven rupture starting points acceleration time series are predicted. Peak ground accelerations for those are vary from 423 cm/s 2 to 584 cm/s 2 which are in the range of recorded Rudbar earthquake at Abbar with the similar magnitude, hypocentral distance and soil characteristics of the site. 5% damped response spectra of those simulated accelerograms are also compared with the one of Rudbar earthquake. Similarity of response spectra shows that effect of earthquake from North Tehran fault on the buildings in Karaj and its surrounding area can be as destructive as Rudbar earthquake on the buildings in Gilan province. We also compared results of our simulations in different distances with attenuation relationships. We showed that PGAs and PGVs of finite fault simulations are in general agreement with attenuation models. In this study a unity distribution of slip for all subfaults considered for simulation. Further studies should be done by modeling asperities of the fault or by considering proper slip distribution over the fault. This is possible with the inversion analysis of middle sized earthquakes caused by rupture of the target fault. Doing this has been difficult so far for North Tehran fault because this fault has not been that active in recent decades and we have no strong or middle sized earthquake from this fault (Ashtari et al. 28) ). ACKNOWLEDGMENT: Thank to Dr. Dariush Motazedian for sharing his code EXSIM, with us. REFERENCES 1) Tchalenko, J.S., Berberian, M., Iranmanesh, H., Baily, M. and Arsovsky, M.: Tectonic framework of the Tehran region, Geological Survey of Iran, Rep. 29, pp ) Berberian, M., Qorashi, M., Arzhangravesh B., and Mohajer-Ashjai, A.: Recent tectonics, seismotectonics, and earthquake-fault hazard study of the Greater Tehran region (Contribution to the Seismotectonics of Iran, Part V), Geological Survey of Iran, Report No. 56, p (in Persian) 3) Berberian, M.: Natural hazards and the first earthquake catalog of Iran, Vol. 1: historical hazards in Iran prior 1900, I.I.E.E.S. Report ) Hartzell, S.: Earthquake aftershocks as Green s functions, Geophysical Research letters. Vol. 5, pp. 1-14, ) Beresnev, I. A. and Atkinson, G. M.: Modeling finite-fault radiation from the ω 2 spectrum, Bulletin of the Seismological Society of America, Vol. 73, pp , ) Boore, D. M.: Stochastic simulation of high-frequency ground motions based on seismological models of the radiated spectra, Bulletin of the Seismological Society of America, Vol. 73, pp , ) Boore, D. M.: Simulation of ground motion using a stochastic method. Pure applied geophysics, Vol. 160, pp , ) Atkinson, G. M. and Boore, D. M.: Ground motion relations for Eastern North America, Bulletin of the Seismological Society of America, Vol. 85, pp , ) Motazedian, D., and Atkinson, G. M.: Stochastic Finite-Fault Modeling Based on a Dynamic Corner Frequency, Bulletin of the Seismological Society of America, Vol. 95 pp , ) Hessami, K., Jamali, F. and Tabbasi, H.: Major Active Faults of Iran, International Institute of Earthquake Engineering and Seismology, ) Wells, D. L. and Coppersmith, K. J.: New empirical relationships among magnitude, rupture length, rupture width, rupture area and surface displacement Bulletin of the Seismological Society of America, Vol. 84, pp ) Motazedian, D.: Region-specific key seismic parameters for earthquakes in Northern Iran, Bulletin of the Seismological Society of America, Vol. 96, pp , ) Kanamori, H. and Anderson, D.L.: Theoretical baisis of some empirical relations in seismology, Bulletin of the Seismological Society of America, Vol. 65, pp , I_29

11 14) Nicknam, A., Yaghmaei Sabegh, S. and Yazdani, A.: Estimating the strong motion of the December 26, 2003 Bam (Iran) earthquake using stochastic techniques, IUST International Journal of Engineering Science, Vol. 19, pp , ) Lam, N., Wilson, J. and Hutchinson, G.: Generation of synthetic earthquake accelerograms using seismological modeling: A review, Journal of Earthquake Engineering, Vol. 4, pp , ) Beresnev, I. A. and Atkinson, G. M.: Generic finite fault model for ground-motion prediction in Eastern North America, Bulletin of the Seismological Society of America, Vol. 89, pp , ) Atkinson, G. M. and Boore, D. M.: Earthquake ground motion prediction equations for Eastern North America, Bulletin of the Seismological Society of America, Vol. 96, pp , ) Building and Housing Research Center (BHRC): Iranian Code of Practice for Seismic Resistant Design of Buildings (standard 2800), Building and Housing Research Center (BHRC), p. 465, ) Boore, D. M. and Joyner, W. B.: Site Amplification for Generic Rock Sites, Bulletin of the Seismological Society of America, Vol. 87, pp , ) Berberian, M. and Walker, W.: The Rudbar M w 7.3 earthquake of 1990 June 20; seismotectonics, coseismic and geomorphic displacements, and historic earthquakes of the western High-Alborz, Iran, Geophysical Juornal International, Vol. 182, pp , ) Sarkar, I., Hamzeloo, H., and Khattri, K. N.: Estimation of causative fault parameters of the Rudbar earthquake of june 20, 1990 from near field SH-wave data, Tectonophysics, 364, pp , ) Sinaeian, F., Zare, M., Mirzaei Alavijeh, H., and Farzanegan, E.: Study of site amplification and soil characterizing of accelerometric stations. The 5th International Conference on Seismology and Earthquake Engineering (SEE5), International Institute of Earthquake Engineering and Seismology. (IIEES), Tehran, Iran, (In Persian) 23) Berberian, M., Qorashi, M., Jackson, J. A., Priestley, K. and Wallace, T.: The Rudbar-Tarom Earthquake of 20 June 1990 in NW Persia: Preliminary field and seismological observations, and its tectonic significance, Bulletin of the Seismological Society of America, Vol. 82, pp , ) Campbell, K. W. and 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. Vol. 24:1. pp ) Saffari, H., Kuwata, Y., Takada, S., and Mahdavian, A.: Updated PGA, PGV and spectral acceleration attenuation relations for Iran, Earthquake spectra, 28, pp , ) Joyner, W. B., Boore, D. M.: Measurement, characterization, and prediction of strong ground motion, Proceedings of Earthquake Engineering and Soil Dynamics II, ) Atkinson, G. M. and Boore, D. M.: Modification of existing ground-motion prediction equations in light of new data, Bulletin of the Seismological Society of America, Vol. 101, pp , ) Ashtari, M.; Hatzfeld, D.; Kamalian, N.: Microseismicity in the region of Tehran, Tectonophysics, No. 395: pp , (Received December 16, 2011 Revised March 2, 2012 Accepted March 6, 2012) I_30