J Seismol (2007) 11:27–38 DOI 10.1007/s10950-006-9025-z

ORIGINAL ARTICLE

Normal faulting mechanisms in the Western Desert of Egypt Ali K. Abdel-Fattah & Ahmed Badawy & K. Y. Kim

Received: 20 September 2005 / Accepted: 17 July 2006 / Published online: 7 December 2006 # Springer Science + Business Media B.V. 2006

Abstract Egypt is recognized as a moderate seismicity region where earthquakes are distributed within several active regions. Owing to sparse distribution of both seismicity and seismic stations, mostly moderate-size Egyptian earthquakes were recorded by regional stations. One of such cases is the moderatesize earthquakes of moment magnitudes greater than 4.0 which struck the Western Desert of Egypt in 1998 and 1999. These events are the first instrumentally recorded earthquakes occurring in the area. In the present study, the source mechanism for these earthquakes was estimated using the waveform data recorded from one of the very broadband MedNet seismograph stations and polarities from the national short-period seismographs. An iterative technique was applied to find the best-fit double-couple mechanism by a grid search over strike, dip and rake. Regional synthetic seismograms were calculated by using f–k integration in the frequency range of 0.03–0.1 Hz. A crustal structure fitted to surface wave dispersion curves was used to compute Green’s function. Focal A. K. Abdel-Fattah (*) : A. Badawy National Research Institute of Astronomy and Geophysics, Cairo, Helwan 11421, Egypt e-mail: [email protected] K. Y. Kim Kangwon National University, 192-1 Hyoja-2-dong, Chunchon, Kangwon-do 200-701, South Korea

depths were determined through the grid search method for a range of source depths. Our results show a normal faulting mechanism with minor strikeslip component. The NNW trend has been chosen as a preferred rupture plane in consistence with surface and subsurface faults and microearthquake seismicity in the epicenteral area as well. Key words grid search method . moment magnitude . relocation . seismic moment . source mechanism . Western Desert . Egypt

Introduction Seismicity of Egypt is characterized by moderate earthquake activity due to the relative motions between the African, Arabian and Eurasian plates. Many inland or intraplate earthquake source regions have been reported such as Dahshour, southwest Cairo, Abo-Zaabal, northeast Cairo, southeast BeniSuef and Aswan (Figure 1). These intraplate earthquakes are important because they greatly expand the region of knowledge about potential seismic hazard where their relationship to regional tectonics is poorly understood. In Egypt, normal faulting mechanisms with strikeslip component are commonly identified by using polarities of P wave in several active regions as a result of dominant extensional stress field (Figure 1).

28

Indeed, few reverse faulting mechanisms with strikeslip components are identified as well (Badawy 2005a, b). Moreover, geological studies have recently revealed reverse faults in the Western Desert (WD) of Egypt (El-Toukhy et al. 1998; Guiraud and Bosworth 1999; Moustafa et al. 2002; Allard et al. 2003). Four significant earthquakes took place in the WD with moment magnitudes (Mw) less than 4.5 (estimated in Figure 1 (a) Seismicity map of Egypt and focal mechanism solution of some significant earthquakes in the 20th century. Focal mechanisms from Badawy (2005a, b). 1: Dahshour; 2: Abo-Zaabal; 3: Beni-Suef, 4: Aswan dislocation. (b) Focal mechanism solution of the EV01 was based on Badawy (2005b), revised in the present paper.

J Seismol (2007) 11:27–38

this study), which can be used to determine the contemporary tectonic behavior of the WD (Figure 1). The determination of focal mechanism using the first P-wave polarities was uncontrolled due to lack of azimuthal station coverage and precise crustal structures. The best-fitting focal mechanism depends on the P-wave polarity observation, earthquake location and velocity model. The stability of focal mechanism

29

Counts

J Seismol (2007) 11:27–38

Cm

T

R

Figure 2 (a) Raw data of EV01 and (b) the corresponding filtered records (0.03–0.1 Hz).

solution is most sensitive to changes in the vertical velocity gradient. In this study, we have found that so far, polarity data alone were insufficient to constrain a well-controlled fault-plane solution. Therefore, the waveform modeling is the only way to achieve significant improvements in the source mechanism solution. The regional broadband seismograms from the KEG station at a distance of 400 km were modeled to determine the fault parameters, seismic moments and source depths. Deriving a reliable source mechanism from a single three-component station using waveform modeling needs a well-known velocity structure and a high signal-to-noise ratio. Since only P-wave velocity models are available near the area of interest, the velocity model along the path from the source to the KEG station was obtained by inverting dispersion curves of fundamental modes of the Rayleigh and Love waves.

In the present study, we used an iterative technique based on the grid search method to obtain strike, dip, rake, and moment magnitude for a range of focal depths with the assumption of a doublecouple point source, using the Green’s function computed from an obtained 1D velocity model. The results contribute to the tectonic explanation for the WD earthquakes.

Tectonic setting Egypt forms a part of the North Africa Craton that underwent four different stress cycles during its geological history. Said (1962) distinguished four major geological provinces: Nubian–Arabian Shield, Stable Shelf, Unstable Shelf and Gulf of Suez–Red Sea graben. The Nubian–Arabian Shield is exposed over a large part of the Sinai Peninsula, the Eastern

30

J Seismol (2007) 11:27–38

Figure 3 Surface structure map of Egypt, epicentral distribution and cross section of microearthquake activity in the area of interest from January 2004 to December 2005 as recorded by the ENSN. The two studied events are also shown.

A’ A

A

A’

Seismic station Studied earthquakes Microearthquakes

Desert and in the extreme part of the Western Desert. The stable shelf, where the epicenteral area is situated at the edge, embraces the area north and west of the Nubian–Arabian Shield. It exhibits a gentle tectonic deformation and its sedimentary

cover represented by continental and epicontinental deposits. For the majority of the stable shield area, the exposed pre-Paleozoic basement shows a regional northward slope, with a corresponding increase of thickness of the unconformable sedimentary cover

J Seismol (2007) 11:27–38

31

Table 1 Epicenteral parameters of the studied events ID

Date

O. Time

Lat. (°)

Long. (°)

EV01 EV02

14/12/1998 30/04/1999

20:48:53.00 23:27:20.00

26.85 26.88

31.35 31.35

which is composed of Paleozoic, Mesozoic and Tertiary to recent formations. The structures consist mainly of parallel, elongated, and tilted fault blocks forming horst and half-graben structures with associated erosion of the up-thrown blocks. Depocenters have been developed with strong variations in the thickness and facies from the Late Jurassic, especially during the Upper Cretaceous. These structures resulted primarily from vertical movement of basement blocks, and consist of draped and/or faulted anticlinal features. The epicenteral area of studied earthquakes is situated at the transition zone between stable and unstable shields. In this zone, a thickening sedimentary section underlain by high basement relief due to block faulting and affected by minor compressional folding. Three major sets of fault trends were shown by El-Kottob (2002), using both gravity and magnetic data. These are the NNW to NW trend, representing the most prevailing faulting direction, the NE–SW trend coming next, and lastly, the ENE to E–W trend.

with a final goal of station spacing less than 1000 km (Boschi et al. 1991). The three-component broadband seismograms (STS-1 seismometers, 24-bit A/D, 20 Hz sampling rate) recorded at the KEG seismic-station were cut into predefined time segments. A linear trend was removed and the data were tapered using the trapezoidal rule (5–10% of window length using the Hanning ramp function). Two horizontal components were rotated about the

Rayleigh

Love

Waveform analysis Data preparation MedNet is a very broadband (VBB) seismographic network installed in countries surrounding the Mediterranean Sea by the National Institute of Geophysics and Volcanology (INGV; Rome, Italy)

Table 2 The used crustal structure in determining epicentral location from Makris et al. (1979) Vp (km/s)

ρ (g/cm3)

Thickness (km)

3.50 6.00 6.35 8.00

2.42 2.82 2.90 3.35

3.5 18.5 11 –

Figure 4 An example output from multiple filter analysis (MFA) of the vertical component of EV01 (1998) earthquake. Left panel contains the dispersion diagram windowed in the 1– 5 km/s group velocity range and 4–100 s period range. Contours outline the portion of the spectrum with the largest arrivals; squares identify the group velocity of the dominating wave for each frequency. Right panel displays the filtered seismogram in the period range of 5–50 s.

32

J Seismol (2007) 11:27–38

1

0 EV

2

0 EV 01

EV

2

0 EV

(a) Figure 5 Velocity models and group-velocity dispersion curves. (a) Dashed line: initial model; dashed-point line: final Vs model; solid line: Vp model derived by Makris et al. (1979). (b) Symbols show the measured dispersion curves for the fundamental modes

Figure 6 An example of variance misfit versus strike, dip and rake obtained from the fine grid search. The minimum misfit refers to strike 334°, dip 32°, and rake −129°. The horizontal lines define a range of possible solutions. Values of error below the bound manifest possible mechanism ranges.

(b) of the Rayleigh and Love waves. Solid line corresponds to the final model. For Love waves, the dispersion curve of EV02 is much shorter because of the low signal-to-noise ratio.

J Seismol (2007) 11:27–38

33

Table 3 Possible mechanism range in strike, dip, rake and moment magnitude derived from the grid search method ID

Strike

Dip

Rake

Mw

EV01 EV02

334–344 170–197

32–40 50–70

−129 to −123 −153 to −135

4.4–4.45 4.15–4.25

relocate epicenters. The arrival times of Pn waves were identified from 20 seismic stations that are shown in Figure 3. The location was determined taking into consideration different velocity models and larger data set than that used in the previous location by the Egyptian National Seismographs Network (ENSN). Different velocity models (Gaulier et al. 1988; Marzouk 1988; El-Sayed et al. 1998; Makris et al. 1979) were proposed to represent the layered elastic medium of the entire area of recorded stations. Among them, the crustal velocity model of Makris et al. (1979) gives the smallest residuals of the arrival times and low uncertainty either in the latitude and longitude. This model consists of three homogeneous layers representing the crust upon a homogeneous half space. The epicenteral parameters are listed in Table 1 and the crustal structure of Makris et al. (1979) is listed in Table 2.

source–receiver back azimuth to obtain radial and transversal components. The data were corrected to ground displacement. The frequency range of 0.02–0.4 Hz was used to minimize effects of the low- and high-frequency noise upon the measurement of the surface wave dispersion. For waveform modeling, a second-order Butterworth band-pass filter with corners at 0.03 and 0.1 Hz was applied after integration. Then, data were resampled with a 1-Hz rate. Figure 2 shows the raw data and filtered records. The filtered records exhibited P, Rayleigh and Love waves.

Shear-wave velocity structure Earthquake location The single-station determination of the source mechanism requires careful path calibration of the velocity structure between the source and receiver. Therefore, the group velocities of fundamentalmodes Love and Rayleigh waves were estimated using the multiple filter technique (Dziewonski et al.

The determination of earthquake source parameters depends strongly on accurate locations and wellknown crustal structure. We fixed the focal depth at 10 km, the normal intraplate earthquake depth in Egypt, and used the first-arrival times of Pn to

Amp (cm)

OBS SYN T

8E-005 0E+000

R

-8E-005

Z

0

50

100

150

200

250

EV02

Time (s)

Amp (cm)

Figure 7 Observed (continuous) and best-fitting synthetic (dashed) seismograms of the events. The obtained focal mechanisms show prefect agreement with the first-motion polarities of short-period network of Egypt.

4E-004

T

0E+000

R

-4E-004

Z

0

50

100

150

Time (s)

200

250

EV01

34

J Seismol (2007) 11:27–38

Figure 8 Optimal best double-couple orientations for a range of source depths: (a) EV01 and (b) EV02.

(a)

1969; Herrmann 1987; Levshin et al. 1989) and then the measured group velocity curves were inverted to determine the one-dimensional velocity model. The applied inversion procedure is an iterative technique in a set of computer programs written by Herrmann and Ammon (2002). The vertical and transverse components of two earthquakes, EV01 and EV02, were used to measure the group velocities of fundamental modes. Figure 4 presents the results of applying the multi-filter analysis (MFA) to the vertical and transverse component of the EV01. The original band passfiltered seismogram is displayed to the right. The data trend was removed and the seismograms were band-pass filtered with corners at 0.02 and 0.2 Hz of second order. The interpretation of contour plots produced by the MFA is often problematic at long periods because of the flattening of the dispersion curves and the decreased signal-to-noise ratio. We were able to measure group velocities of the Rayleigh wave at periods between 2.5 and 30 s, and of the Love wave at periods between 7 and 40 s as shown in Figure 4. Once dispersion curves were measured, we then inverted their group velocities of fundamental modes for structural model. The starting model used in the inversion is a half-space velocity model with a fixed interval of 2 km thick. The S-wave velocity is the only free parameter in each layer. Consequently, the P-wave velocity is estimated from an assumed Vp/Vs ratio. The assumed ratio (1.72) is obtained from several

(b)

trials during waveform inversion that gives the best alignment between early and later phases. The velocity structure fitted to the dispersion curves is displayed in Figure 5a. Both observed and theoretical dispersion curves corresponding to the obtained velocity model are shown in Figure 5b. Finally, we relocate the studied events using the model with a too small difference to be mentioned. Waveform modeling Determination of the seismic moment tensor is now becoming a common approach in monitoring regional seismicity (e.g., Dreger and Helmberger 1993; Romanowicz et al. 1993; Thio and Kanamori 1995; Fukuyama et al. 1998). The least squares approach of Dreger and Helmberger (1990, 1993) and Dreger (1995, 2000) has been demonstrated to be reliable in analyzing events with local magnitudes as low as 3.5. Many authors have successfully evaluated different local earthquakes recorded by either multistations (e.g., Ichinose et al. 1998; Shomali and Slunga 2000; Hofstetter et al. 2003) or single station (e.g., Ichinose et al. 1998; Kim et al. 2000) by waveform inversions. We use the waveform from a single broadband station (KEG), the only available broadband station which recorded the respective events. Green’s functions were computed for the point source in a layered elastic medium by using a fast frequency–wavenumber (f–k) integration technique (Zeng and Anderson 1995). The synthetic seismo-

J Seismol (2007) 11:27–38

35

grams for an arbitrary double-couple mechanism are computed by combining with Green’s function. The grid search method over moment, strike, dip and rake was used to determine the best double-couple mechanism, scalar moment (thus also the moment magnitude), and focal depth. A combined crosscorrelation was done and showed zero time shifts for all three components. The goodness of fit between the observed and synthetic data (optimally aligned using the cross correlation) was defined by var ¼

n Z X i¼1

minima are observed for the fault plane and the auxiliary plane. The source parameters was obtained near one of the two minima by applying the finer grid search over 3° increments in both strike and rake, and over 2° increments in dip. An example of the resulting model space is shown in Figure 6. To obtain an estimate of the range of possible solutions, we followed Walter (1993). A window of preevent noise was added to the best-fit residual to give the extremes of strike, dip and rake. The result is plotted as dotted horizontal lines in Figure 6. Solutions below these lines constitute a range of possible mechanisms. It should be emphasized that not all combinations of these ranges will fall below the dotted line. The possible mechanisms range in strike, dip, rake and moment are listed in Table 3. The waveform match and the resulting focal mechanism are shown in Figure 7. The focal depths were determined by executing the grid search in depth over 2-km intervals (Figure 8), providing the focal

½syni ðt Þ  obsi ðt Þ2

where var is variance, syn(t) and obs(t) are the synthetic and observed waveforms, respectively, and n is the number of components (in this case n=3). Initially, a coarse search was conducted over 30° increments in strike, dip and rake at different depths in the 2–16 km range with 2-km increments. Two Figure 9 Residuals between observed and synthetic seismograms for depths of 10, 14, and 20 km.

10 km 14 km 20 km T

Residual

0.0006

R

-0.0006

Z

0

50

100

Time (s)

150

200

250

36

depths of 14 and 4 km for EV01 and EV02, respectively. An example of the residual seismograms of the best-fitting synthetics at 10-, 14- and 20-km depths is shown in Figure 9 for EV01. The 14 km solution has a lower misfit; the 10 km solution has a larger misfit for the Rayleigh waves, while the 20 km solution has a larger misfit for the Love waves.

Discussion Earthquake focal mechanisms can provide important information about slip, fault structure at depth, and the stress field in the epicentral region. The centroidmoment tensors (CMT) are routinely estimated at Harvard University for globally observed earthquakes of M≥5.5. However, the vast majority of earthquakes recorded by local and regional seismic networks are too small to be included in the Harvard CMT catalog. Because of their frequent occurrence, these small earthquakes are particularly important for characterizing regional tectonics and constraining stress orientation. Egypt is in fact a region of relatively moderate seismicity where earthquakes above the magnitude threshold of the Harvard CMT catalog have been only few since 1977 and only one event since ENSN started operations in 1997. Therefore the contribution given by P-wave polarities is also important. Indeed, polarities in some cases are insufficient to obtained well-controlled focal mechanism solution, so they should be combined with the regional waveform studies. Four moderate earthquakes of Mw≤4.5 have been recorded for the first time in the WD of Egypt during 1998 and 1999 (Figure 1). For two of them (with good signal-to-noise ratios), we present the fault plane solutions, which were found from the waveform modeling constrained by the P-wave polarities. The fault plane solution from a single three-component station requires a very well known velocity structure. For this purpose, the velocity model along the path from the source to the KEG station of Mednet has been derived by inverting the dispersion curves of the surface waves. The same waveforms have been used for the source and structure modeling. The surface wave dispersion is very sensitive to shear-wave velocity and the influence of P-wave velocity being much smaller on radial seismograms. It is expected

J Seismol (2007) 11:27–38

that the synthetic seismograms computed from the shear-wave velocity model obtained from the dispersion curves should show perfect match for Rayleigh and Love waves at least. The agreement between the observed and synthetic seismograms shown in this study for all phases (P, Raleigh and Love waves) gives the confidence in the assumed Vp/Vs ratio, thus also confidence in the obtained Pvelocity model. Matching all those waves is essential for claiming that the mechanism is reasonable (Figure 7). Besides, the grid search method was followed by combined cross-correlation to measure delay time between synthetic and observed waveforms taking into account first and later arrivals. This process is repeated at a range of source depths. The maximum correlation coefficient at zero-delay time indicates that the Vp/Vs ratio of 1.72 is appropriate. The obtained fault plane solutions of the respective earthquakes exhibited normal faulting mechanisms with minor strike-slip component, stable with varying the depth. The focal mechanism solution of the first event (EV01) was also determined by Pwave polarities and shows reverse faulting mechanism (Badawy 2005a, b). Although, the misfit of the polarity solution is only 8%, it is uncontrolled. We think that the poorly known crustal structure is the main source behind this discrepancy. In contrast, changes in event location, source depth, seismicvelocity structure and consequent takeoff angle can alter the best-fitting focal mechanism. Moreover, the dip-slip events appear to be more sensitive to takeoff angle changes than the strike-slip events (Hardebeck and Shearer 2002). Regardless of the sense of motion, the common result is that a NNW-oriented nodal planes are in good agreement with both surface and subsurface faults affecting the epicentral region. Monitoring of earthquake activity in Egypt in 2004 and 2005 reveals a NNW active trend in the epicentral region that coincides with the newly estimated location of the EV01 and EV02 (Figure 3). This recognized activity has been recorded at short period seismograph with microsize magnitude. Of course, the accuracy of ENSN’s location is highly enhanced by a complete construction of network stations around the epicentral area by the end of 2003. We have relocated these events by using both Pand S-arrivals and the obtained velocity model. The

J Seismol (2007) 11:27–38

cross section in Figure 3 shows two distinguished clusters at depths of 3–8 and 10–15 km. Focal depths obtained from the inversion, for EV01 and EV02, suggest that the first event took place at the deeper cluster while the second one represents the shallower one.

Conclusions From our waveform modeling and relocation technique using two sets of earthquake data in the WD, we draw up the following conclusions: 1. Their moment magnitudes are 4.5 and 4.2, which are different from the local magnitudes of 5.4 and 4.7 determined by the ENSN. The discrepancy mainly arose from uncertainties in the estimation of local magnitudes. 2. Their focal depths determined by grid search method are 14 and 4 km for EV01 and EV02, respectively, with a possible uncertainty of 2 km. 3. The relocated earthquake parameters are highly consistent with the annual microearthquakes monitored particularly by the end of 2003 due to the complete installation of the ENSN stations. 4. The waveform modeling yields normal faulting mechanism with strikes of 334±10° and 174± 27°, dips of 32±15° and 50°, ±12°, rakes of −129± 8° and −141±20°, and seismic moments of 5.89× 1016 and 2.09×1016 N m, for EV01 and EV02, respectively. The obtained results of normal faulting mechanism represent the only earthquake-derived stress indicators in the epicentral area for a long time. Acknowledgments We thank Jan Sileny at the GFU for his valuable suggestions and discussions to improve the original manuscript. Comments and constructive criticism given by the associate Editor and two anonymous reviewers greatly improved the manuscript. This work was supported by Korea Science and Engineering Foundation (KOSEF). Figures were generated using the Genetic Mapping Tool.

References Allard D, Phelps D, Bedingfield J, Barker G (2003) Egyptian western desert Jurassic and Cretaceous structural styles.

37 Abstract at the AAPG International Conference. Barcelona, Spain, 21–24 September 2003 Badawy A (2005a) Seismicity of Egypt. Seismol Res Lett 76 (2):149–160 Badawy A (2005b) Present-day seismicity, stress field and crustal deformation of Egypt. J Seismol 9(3):267–276 Boschi E, Giardini D, Morelli A (1991) MedNet: the Broad Band seismic Network of the Mediterranean. Il Cigno Galileo Galilei, Roma, Italy Dreger DS (1995) Using the time domain moment tensor inversion program: MT-INV. 1995 IRIS workshop on moment tensor inversion, IRIS Dreger DS (2000) Time domain moment tensor inverse program. Berkley Automatic Seismic Moment Tensor Code 1.0, Berkeley, CA Dreger DS, Helmberger DV (1990) Broad Band modeling of local earthquakes. Bull Seismol Soc Am 80:1162–1179 Dreger DS, Helmberger DV (1993) Determination of source parameters at regional distances with single station or sparse network data. J Geophys Res 98:8107–8125 Dziewonski A, Bloch S, Landisman M (1969) A technique for the analysis of transient seismic signals. Bull Seismol Soc Am 59:427–444 El-Kottb A (2002) Geophysical study around the epicenter of earthquake of 1 May 1999, Sohag, Western Desert, Egypt. Msc. Thesis, Al-Mansoura Univ.: 200 pp El-Sayed A, Arvidsson R, Kulhanek O (1998) The 1992 Cairo earthquake: a case study of a small destructive event. J Seismol 2:293–302 El-Toukhy M, Charmy H, Studer M (1998) Structural evolution of the eastern part of the western desert and its implications for hydrocarbon exploration. Proc. 14th Petroleum Conference, Vol. (1) Egyptian General Petroleum Corporation, Cairo, pp156–177 Fukuyama E, Ishida M, Dreger DS, Kawai H (1998) Automated seismic moment tensor determination by using on-line broadband seismic waveforms. Zisin 51 (1998), pp149– 156 (in Japanese with English abstract) Gaulier JM, Le Pichon X, Lyberis N, Avedik F, Geli L, Moretti I, Deschamps A, Hafez S (1988) Seismic study of the crust of the Northern Red Sea and gulf of Suez. Tectonophysics 153:55–88 Guiraud R, Bosworth W (1999) Phanerozoic geodynamic evolution of northeastern Africa and the northwestern Arabian platform. Tectonophysics 315:73–108 Hardebeck JL, Shearer PM (2002) A new method for determining first-motion focal mechanisms. Bull Seismol Soc Am 92:2264–2276 Herrmann RB (1987) Computer programs in seismology, Volume IV: Surface waves inversion. Saint Louis University, Missouri Herrmann RB, Ammon CJ (2002) Surface waves, receiver functions and crustal structure, version 3.25. Saint Louis University, Missouri Hofstetter A, Thio HK, Shamir G (2003) Source mechanism of the 22/11/1995 Gulf of Aqaba earthquake and its aftershock sequence. J Seismol 7:99–114 Ichinose GA, Smith KD, Anderson J (1998) Moment tensor solutions of the 1994 to 1996 Double Spring Flat, Nevada. Earthquake sequence and implications for local tectonic models. Bull Seismol Soc Am 88:1363–1378

38 Kim SG, Kraeva N, Chen Y-T (2000) Source parameter determination of regional earthquakes in the Far East using moment tensor inversion of single-station data. Tectonophysics 317:125–136 Levshin AL, Yanovskaia TB, Lander AV, Bukchin BG, Barmin MP, Ratnikova LI, Its EN (1989) In: KeilisBorok VI (ed) Surface waves in vertical inhomogeneous media, in seismic surface waves in a laterally inhomogeneous earth. Kluwer Academic Publisher, Dordrecht, pp131–182 Makris J, Stoefen B, Vees R, Allam A, Maamoun M, Shehata W (1979) Deep seismic sounding in Eqypt, Part I Unpublished technical report, Inst. für Geophys. Univ. Hamburg, Hamburg, 20 pp Marzouk I (1988) Study of crustal structure of Egypt deduced from deep seismic and gravity data. Ph.D. thesis, p 118, Hamburg Univ., Germany Moustafa AR, Abdel-Aziz AMT, Reda T (2002) Effect of Basin geometry on locating promising hydrocarbon traps: Tiba Natrun–Kattanyia inverted basins, northwestern Desert

J Seismol (2007) 11:27–38 (Egypt). AAPG International Conference & Exhibitions. Cairo, Egypt Romanowicz B, Dreger D, Pasyanos M, Uhrhammer R (1993) Monitoring strain release in central and northern California using broadband data. Geophys Res Lett 22:1643–1646 Said R (1962) The geology of Egypt. Elsevier, Amsterdam, p 337 Shomali ZH, Slunga R (2000) Body wave moment tensor inversion of local earthquakes: an application to the South Iceland Seismic Zone. Geophys J Int 140:63–70 Thio H-K, Kanamori H (1995) Moment-tensor inversion for local earthquakes using surface waves recorded at TERRAscope. Bull Seismol Soc Am 85:1021–1038 Walter WR (1993) Source parameters of the June 29, 1992 Little Skull Mountain earthquake from complete regional waveforms at a single station. Geophys Res Lett 20 (5):403–406 Zeng Y, Anderson JG (1995) A method of direct computation of the differential seismogram with respect to the velocity changes in a layered elastic solid. Bull Seismol Soc Am 85:300–307