151

Geological Bulletin Univ. Peshawar Vol. 35, pp. 151-161,2002

ave
TARIQ MAHMOOD, M. QAISAR & ZAHID ALI
Micro Seismic Studies Programme Ishfaq Ahmad Research Laboratories
P. 0. Nilore, Islamabad
ABSTRACT: Teleseismic body waves recorded at the IRIS Global Seismographic network are
analyzed to investigate the source mechanism ofthe Astor valley earthquakes of Nov. 20,2002
occurred in northern area of Pakistan about 60 km southeast of Gilgit near Bunji. The inversion
method ofKikuchiand Kanamori is applied to the records of 11 seismicstations that included vertical
P and horizontal SH components for thispurpose. The azimuthal coverage of seismic stations was
good enough to resolve some details ofheterogeneous moment tensor. Thefault area is estimated
porn thefirstfour days aftershock distribution as 25 x 20 km2and their distribution corresponds
to the local tectonic trends. The estimated seismic moment of the November 20,2002 main shock
is 9.4 x 1017Nrnwhichgives Mw =5,9. Thesolution ofbody waves indicates normalfaulting with
a small component ofstrike slip on sourcefault whose strike isparallel to the local trend ofMMT,
with the tension axisperpendicular to it. The location of the main shocks along with aftershock area
suggests that the nodalplane oriented southwest-northeast dipping to the northwest ispreferredfor
thefaultplane. Fault Plane solutions ofNovember 01,2002 and November20,2002 earthquakes
derivedfiomfirst motion polarity data recorded by MSSP local seismic network and someglobal
seismicstations also show normalfaulting. The results indicate that the Astor valley earthquakes
may be considered to have occurred in an uncoupled region, in response to thegravitationalpull
due to internal gravity sliding.

INTRODUCTION
On November 01,2002 a moderate earthquake
of magnitude mb= 5.3 occurred in the northern area
of Pakistan about 45 km southeast of Gilgit city. The
earthquake occurred at 3S037' 12"N and 74@39'
36"E
during local nighttime at 2209 GMT.
On N~vember20,2002 the region was rocked
again by another large shallow earthquake of magnitude mb= 6.2 located on almost at the samelocation
in the northwestern part of Astor valley, northern
Pakistan duringlocal nighttime. The main s h o d was

followed by a large number of aftershocks, few of
which were large enough to increase the damage
caused by the main shock. The occurrence of aftershock lasted 40 days. This is the largest event ever
recorded in the region. This earthquake provides a
good opportunity to study the Seismotectonics
characteristic of the region. The epicenter of
main shock was located at about 60 km southeast of
Gilgit city.
The epicenters of the earthquakes lie near the
interface of Kohistan Magmatic Arc and the Nanga

Geological control on natural hazards: earthquakes and mass movement. Khan, M.A., Abbasi, LA., &
Khattak, G.A. (Eds.). Geological Bulletin, University of Peshawar, (Special Issue), Vol. 35, pp. 151-161, 2002

The main purpose of this study is to determine the mechanism of November 20. The followingwere the hypocentral parameter of the main shock. pp. fault geometry.52"N. MAIN SHOCK AND AFTERSHOCKS The hypocenter of the November 20. 1997). Determination of these parameters is useful not only for understanding the physics of earthquakes but also for estimating the potential hazard associated with stress changes in the faults adjacent to the earthquake area. 151-161.152 Geological Bulletin Univ. near the town of Bunji (Fig. Seisrnotectonic map of Gilgit and surrounding region modified from Geological Map of Pakistan by Farah Hussain 1993 showing location ofmain shocks (stars)of November 20 and open star shows location of November Parbat Haramosh Massif. 1). 2002 earthquake including. 1999). fault area. along the MMT-Raikot fault. using waveform inversion. 74. 2002 Astor valley earthquake was relocated by using computer code HYPOELLIPSE (Lahr. seismic moment and other related source parameters.5 km . The entire area is faulted and fractured and seismically active with history of quite a few moderate-damaging earthquakes (Kazmi and Jan.68OE Depth : 21. This feature raises the significance ofthe tectonic studies and study of active seismicity of the region. 2). 1. The data recorded by local seismic network and some global seismic stations for azimuthal coverage were used to determine the hypocentral parameters (Fig. Origin time : 21:32:29. Peshawar Vol: 35.85 Epicenter : 35.The area also represents the northern margii of the under thrusting Indian plate.2002 73"00' 73' 30' 74' 00' 74' 30' 75' 00' 75' 30' 76' 00' Fig.

pp. Star denotes the main shock of November 20. 2002 153 Fig 2. 15 1-1 61. 35. MSSP seismic network and IRIS broadband seismograph stations used in the present waveform inversion. 2002 earthquake and the solid triangles denote the stations. Peshawar Vol..Geological Bulletin Univ. .

3)....... Pakistan is characterized by extensive zones of high seismicityand contains several Seismotectonics featuresgenerated by an integrated networkofactive faults.. Fig... The location of the main shock along with aftershock is shown in (Fig.USA locations.- -- . . -... 1997).2002 154 GEODYNAMICS OF REGION The study area lies within High Himalayas where the Nanga Parbat Haran~oshMassif and Kohistan Magmatic Arc define the tectonic setup of the present earthquake region (Fig 1). The aftershocks are distributed in northeast and southwest direction.... It has been believed that the development of thrust and fold system in the region by continued cornpression since more than 55 million years.... The main shockand distribution ofaftershocks is mainly along the Raikot fault zone that separates the Nanga Parbat Haramosh Massif by Kohistan Magmatic Arc near the town of Bunji. 35.. 151-161. resulted in the formation of many complex antiform and synform structures linked by the micro to mega scale faults. The present earthquake occurred in northwestern extremity of Astor valley....... Epicenter of main shock (star) of Nov 20....Geological Bulletin Univ. The results were almost same as the National Earthquake Information Center (NEIC).. Peshawar Vol.2002 earthquake and it afier shocks (circles) according to the size of the-magnitude using MSSP data... This is an active seismic zone with frequent earthquakes of 3 to 5 magnitude with depth ranging from 50-100 km (Kazmi and Jan... The locations of aftershocks extend about 30 km along the MMTRaikot fault zone and the depth down to about20 km. pp.. close to the interface of Ghizar range and Ladakh mountain ranges as shown in Fig 1.. 3....

(Searle. pp. along MKT. The Kohistan MagmaticArc resulted when the TethySea had begun to shrink by the time India began its northward drift around 130 million years ago. Gravity data modeling indicates that the MMT and MKT dip northward at 35O to 50" and that the Kohistan arc terrain is 8-10 km thick (Malinconics and Lillie. 4.2002 MORC P UD 307. The northern and western part of the arc.'1979).8 Fig.Geological Bulletin Univ. resulted in cornplete consumption of the leading oceanic edge of with remnants Indian plate and its eventual~coUision of the Kohistan-Ladakh arc. 1989). The seismological data suggests that arc is underlain by the Indian crustal plate (Seeber and Armbuster.61 YSS P UD 54. .2002.6 5. . 151-161. is covered by a sequence of late Cretaceous to Paleocene volcanic and the arc terrain is mainly composed of igneous intrusions. Kandhar) during the Cretaceousera 1. 1979). 20. . 1991). Intraoceanic subduction generated a series ofvolcanic arcs (Kohistan. Kohistan Magmatic Arc is an intraoceanic island arc bounded by the Indus Suture Zone or MMT to the south and the Shyoke Suture Zone or MKT to the north (Fig.36to 65 my. Body waves recorded by IRIS stations used in the present waveform inversion for the Astor valley earthquake of Nov. Ladakh. 1). 35. Continued subduction of the Tethy sea floor beneath Kohistan Ladakh arc and Eurasia. The abrupt slowing down of India's northward movement between 55 and 50 million years ago is attributed to this collision (Powe11. Peshawar Vol.

pp. Coward et al... Treloar et al. however. (199 1) have.2002 Time . Results of the inversion with double couple source. 151-161. Madin (1986) and Madin et al.3 Fig. (1989) presented field evidence that along the western side of Nanga Parbat-Haramosh Massif there is an active dextral (right lateral strike slip) reverse fault-the Raikot Fault (Fig.- . According to various -iwebtiptor&Tahirkhe1il929~arhetalal1980. (A) Seismic moment release as a function of time (B) Mechanism of the best fit double couple point source (C) Observed and synthetic P and SH wave forms. 1992). the MMT is comprised'of a complex system ofthrusts and shears. 1986. the MMT surrounds this massif. I ) that truncates - - - . shown that between Raikot and Sassi. 5.sec 1.51 css P UD 281.Geological Bulletin Univ. TheNangaparbat-HaramoshMassifisamajor re-entrant into the Kohistan arc terrain. 35. Butler et al.Peshawar Vol. The numbers in the upper and lower left indicate peak-to-peak amplitude in microns of the observed records and azimuth respectively.

the MMT is comprised of a complex system of thrusts and shears. assuming no change in the mechanism during rupture. the effects ofinhomogenity in the structure. The information contained in this time window is adequate to resolve the source process. (1991) have. The inversion process was carried over different depths. Consequentlythe seismogramswere simultaneouslyinverted to double couple single point source in the least square sense for the source model parameter. The Green's function is calculated for the six element of the moment tensor at different depths beneath the epicenter with the assumption of a single point source. however. The focal depth lies within 20-25 lundepth range. 1983).Geological Bulletin Univ. On the basis of reconstruction of the original stratigraphy and subsequent postulated erosion. pp. In this distance range.02 to 5 Hz is used. the waveforms are not contaminated by strong upper mantle or core phases (Kikuchi. the teleseismic body waves data were windowed for one minute starting 10s before P-wave arrival or S-wave arrival. of 6s and the seismic moment of xlOlsNm. The data were deconvolved to ground displacement and filtered with an appropriate fre- An iterative deconvolution inversion method developed by Kiuchi and Kanamori (1991) for teleseismic data is employed to resolve the source complexity by minimizing the differencebetween the observed and synthetic waveforms. The azimuthal coverage is good enough to resolve some details of the moment distribution. For attenuation parameter Q a Futterman (1962) operator t* (ratio of travel time to average Q) of 1s was used for P waves and 4s for S waves (Helmberger. The depth yielding the minimum residual was taken as the depth of the source. This Q parameter was convolved with a triangular source time function having rise time T. Treloar et al. (1989) estimate that the actual total uplift of Nanga Parbat may have been about 24 km. 1995). For this event a point source model provides more or less equal fitsto most ofthe waveforms. A bandwidth of 0. The bandwidth is determined by considering the spectral content and noise level. Fourteen body wave records from 11 stations.of 3s. were used in this study to derive the source process of this earthquake. Using IASPEI (199 1)SeismoIogicaltravel timetables. a1 1991)with only a minor modification in the source region as given in Table 2. In order to get a more realistic waveform. 35. Zeilter's data (1985) show that this massifhas been uplifted approximately 16lunin the last 6 to 8 m. quency band. 151-161. In the first instance the source mechanism is determined with the approximations of a single point source such that synthetic waveforms are best fit with the observed ones.y. Athree layer structures (two layers of crust and a semi-infmite mantle) is used for the synthetic waveform followingthe crustal model determined by (Qaisar et. 1997). This . The Nanga Parbat-Haramosh Massif is unique due to its current extraordinarily high uplift rate of over 5 mmtyear (Kazmi and Jan. including vertical P and horizontal SH corn ponents. Peshawar Vol. Teleseismic body wave inversion BODY WAVE ANALYSIS Seismicdata The teleseismic broadband data set used in this study was retrieved from Data Management Center of the Incorporated Research Institute of Seismology (IRIS-DMC) in the epicentral distance ranges between 300 and 900. shown that between Raikot and Sassi. Madin et al. 2002 MMT and forms the western boundary of the massif. Station parameters are given in Table 1 and displacement records of P waves as well as SH waves are shown in (Fig 4). attenuation during traveling ofwaves and instruments response was considered. and that the uplift has accelerated exponentially.Fission-track and Ar401Ar39 data by Zeilter (1985) show that this uplift is 7 times higher than in the adjacent Kohistan arc terrain. source duration T.

128 9.079 9. 35.O Mudanjiang China MDJ 60.7 86.079 9.40 P-UD 1.068 7.4 0.-. 151-161.Peshawar Vol.9 40.Geological Bulletin Univ.2 MORC 307.74 SH 0.8 0.50 P-UD 1.2002 TABLE 1 LIST OF STATION PARAMETERS Back Azimuth Delta P l/r Phase Weight Mathiatis Cyprus 76..50 1 P-UD I 1.3 0.0 35.077 9..075 9.0 Yuzhno Russia YSS 1 ' 54.0 Station Name Code Azimuth I 1 GVD 284. Delta = Distance in degrees .1 80.9 50.50 P-UD 1.9 40.2 Hailar China HIA 53.8 0.4 -101.85 Where Back Azimuth = Azimuth from station to source P = Ray Parameter (P = Sin ilv where i = angle of incidence) llr = Geometrical factor ( ~ 1 1 1 0km).74 P-UD 1.074 9.O Yakutsk Russia YAK 34.0 0.0 -95.80 P-UD 1.2 32.7 0.O Gardos Greece 73.8 -76.073 8.O Mudanjiang China MDJ 60.0 Yuzhno Russia YSS 54.2 -82.5 0.6 111.6 1 41.073 1 9.077 9.- ------ .70 P-UD 1.073 8.70 P-UD 1. pp.O Isparta Turkey ISP 287.5 0.0 Gardor Greece Morovsky Czech.5 35.0 Toflu Estonia TRTE 322.7 1 0.30 P-UD 1. Republic Obninsk Russia Piszker Hungary 1 PSZ 1 304.8 33.6 42.9 0.0 105.-.-.0 0.3 38.3 73.7 43.80 P-UD 1.

The final residual waveform error is 0. A slight change in therparameters of the fault plane caused an increase in the residual error. 5a-c). = P-wave and S-wave velocities (kmls) p = density (lo3kglm3).3~10'~ 6. A a . The best matching double couple has a strike of 207.2 3 6 Mw 5. 35. The aftershock area was estimated roughly as 25 x 20 km2by consideringthe aftershock occurred within 4 days. 151-161.5 3. .0 -54 21. The local stress drop is estimated by using the relation of Fukao and Kikuchi (7987).0 26. 4 3 5 (Mpa) be4 (set) 3. A. A a = 2. The local stress drop is evaluated as A o .6224.4~10~' HarvardCMT 202. synthetic and observed waveforms for the point source model is shown in (Fig. = 2. = Slip.3 -36. The P wave records as well as the SH waves. is derived for the main shock from the seismic moment M.and rupture duration 22.3" in the northwest direction and a slip of -36. T. normal faulting mechanism.0 9. =3Mpa for + = 3 sec and M. 2 . 4) also reflect a smooth rupture.e.0 43.5 9.0 kmlsec. D = thickness (km) means that a single point source is sufficient to describe the source process of this earthquake.5 MJ(Vz)' Where Vp.2 21. pp.60. inversion was done using this solution as a fxed mechanism with the result indicating a good fit to the data with a final residual error equal to 0.3 USGS 230.9 H = Depth Mo = Seismic moment. 6 = Dip.5 Mo I S The average stress drop is estimated as 0. The fit of data i. Where V is the rupture velocity and + represents the rupture duration. = Source duration MW = Moment magnitude. It is observed that the local stress drop is higher than the average stress drop over the fault plane of 25 x 20 km2. The displacement 'D'is estimated as 0.2O. at different stations (Fig. The average stress drop A o was obtained by applying the relation (Fukao and Kikuchi. TABLE 3 FINAL SOURCE PARAMETERS 6 (9 This Study h H Mo (km) (Nm) ' 207. a dip of 60.= Local Stress drop. a.8~10'~ 6. = Rise time. In the present model a triangular source time function and half the duration z that corresponds to the rise time of3 sec is assumed. 2002 Stress drop TABLE 2 NEAR SOURCE STRUCTURE USED IN THE WAVEFORM INVERSION The local stress drop AD. Peshawar Vol. Ips Where pis the rigidity (3x101•‹N/m2).4 x lo1' Nm.V. The rupture velocity is taken as V = 3.6 60. 1966) the average can be estimated by using the followdislocation (6) ing relation Moreover.063 m. 1987).159 Geological Bulletin Univ. = 9. According to the slip dislocation theory of faulting (Aki.6224. The final parameters of this solution are summarized in Table 3.0 -120 17. D = M.2 Mpa for an aftershock area.0 Where cP = Strike.

& Khan. 20. & Wheeler.. P. P. R. Crust and mantle of an abducted island arc. . Inversion of complex body waves -111.. H. 1962.. Tokyo Univ. W. 557-567. Butler. I. 151-161. pp. Nazimabad.. near the town of Bunji. Bull. in Earthquakes. N.(eds) Collision Tectonics. Pub: Geological Survey of Pakistan. Karachi. B. 73-88.2002 DISCUSSIONS AND CONCLUSIONS The earthquake of Nov. H. F. M. W. and is not appropriate to the general trend of the local tectonics that is dominated by collisional tectonics mostly strikeslip and thrust faulting. Geological Map of Pakistan. Peshawar Vol. B. A..2002 was located in northwestern extremity ofAstor valley. A. 1991. and its implications for Himalayan tectonics. Res. Kennett. Fault Plane solutions of Nov.. Petterson. Waveform analysis and aftershocks distribution reveal that the event represents normal faulting with a small component of strike-slip on source fault with strike = 207. Helmberger.6". Dispersive body wave. 35. J. 67. A. Luff.. 19. Lond. 01. Spec. Geol. J. D. Kazmi. 160. Quetta. Fukao. JavedIqbal Senior Engineer for assistingwith various aspects of this study.V. 1980.. they are likely to become unstable under the gravity pull. D. Research School of Earth Sciences. C. Peshawar.. Rex. 203-219. Source retrieval for mantle earthquakes by iterative deconvolution of long period P-waves. Journal of Geophysics. 144. P. J. Kikuchi. M.. 1992. Q.. M.. The seismic moment is 9.. & Ries. Pudsey. 87-93. J. 13. 2002 earthquakes derived from first motion polarity data recorded by MSSP local seismic network and some global seismic stations also show normal faulting. K. M. Seismol. Canberra ACT 2601. Futterman.e. Observation. The movement along Raikot fault zone where the strike slip component is dominant can provide threshold energy to activate the gravity sliding of high-dip-angle limb of an antiformlup-lifted structure ofthe region.2". Soc. Maluski. northern Pakistan... Broughton. Publ.. Jones. normal.Geological Bulletin Univ. Graphic Publishers.. P. 670-689. IASPEI 1991 Seismological Tables. & Proust. & Kikuchi. 1964. George.. The Kollistan sequence. G. Inst. 174-222.2002 earthquake as determined by inversion method indicates normal faulting. Generation and propagation of G waves from the Niigata earthquake at June 16. Coward. Bologna.. 1986. Bard. P.3" northwest direction and slip = -36. Windley.. M. L. Pakistan. The strike of both the source faults is nearly close to the trace of Raikot fault but nature of fault i. Australian National University. B. Geol. Harris. R.. Hussain. This ultimatelyformed the source of the Astor valley seismic activity.. Australia. 1983. F. Soc.. Acknowledgement: The authors are thankful to Mr. Collision tectonics in the NW Himalayas.. &Jan. This may ultimately initiate normal faulting by internal "gravity sliding". 1966. C. Italy). W. H.9.. Theory and Interpretation (Societa Italiana di Fisica. Am. In: Coward. D. 1991. Treloar. 410-413. W. 79. Bull.. Theory and application of synthetic seismograms. Prior. J.M. Soc. H. & Kanamori. This uplift provides sufficient base for the activation of internal gravity sliding. 1997. Tectonophysics. 5279-5291. As describe earlier that the Nanga Parbat-Hararnosh Massif has been uplifted about 16 km during the last 6 to 8 may.. 249-269. M.. 1993. D.Geologyand Tectonics of Pakistan. Y. The focal mechanism solution of Nov 20. Lond. The possible reason of normal faulting in Nanga Parbat-Haramosh Massif is that when the limbs of antiform reach at high-dip-angle. Earthquake Res. 149.. M. F. 20. Univ. J. C. N.4 x 10i7Nmthat correspond to a moment magnitude of 5. Geol.I. Geology of the northern part of Nanga Parbat massif. dip = 60.. Matte. 44. Bull. REFERENCES Aki. 1987. C. 2002 and Nov..

A. Soc.. The north-western Nanga Parbat-Haramosh Massif. Tahirkheli. P. 1979. P. M.) Geodynamics of Pakistan. Seismicity of the Hazara arc in northern Pakistan: decollement versus basement faulting. 1986. Tertiary Extensional Features. Peshawar Vol. Searle. B. K. K.. Oregon State Univ. S. In: Farah. 1985. &Wright. Geology and neotectonics of the north-western Nanga Parbat-Haramosh Massif. Geol. Am. 358. M.) Geological Survey of Pakistan. Univ. & DeJong... Corvallis. Lawrence. S.. Madin. thesis. P. Geol. Indianplate motion and shape: constrains on the geometry of the Himalayan Orogen. K. L. M..A... 209-214. Tariq. Soc. New York. 1-97. & Lillee. L. 151-161. 1991. Eastern California.. Japan. Qaisar. Surv.) Tectonics of the Western Himalayas. evidence for crustal uplift at the northwestern corner of the Indian Craton. M. & DeJong. Wiley & Sons. C. & Mubarak. Seeber. 1995. Quetta. M. (eds. M. P. pp. 4. 99-23. 1979. 131-142. 1991. J. A. L. 35. 191. Madin. P. Farah. & Armbruster. Magnitude and First Motion Pattern.. K. A.. A Speculative Tectonic History of Pakistan and Surroundings: some constrains from the Indian Ocean. JICA. Treloar. Powell. Cooling history of the NW Himalaya. Quetta. I. D. Lahr.C. 1979.. "Notes on Earthquake Source process" IISEE.. Pak. &Coward.. Am. MSSP-4119 1. & Rehman. (eds.2002 Kikuchi. A. Peshawar. 5-24. R.. A. Pakistan. R. Geology of Kohistan and Adjoining Eurasian and Indo-Pakistan Continents. Shahid. M. J. Simultaneous inversion of velocity struc- 161 ture and hypocentral locations: Application to the area around the proposed Kalabagh Dam site. J. 127-151. .Geological Bulletin Univ. J. I. W. L.. A. In: Malinconics. 169182. Zeilter. Geol. Troxel. "A computer Programme for Determining Local Earthquake Hypocentral Parameters. Pap. Centennial Field GuideCordilleran section. Geology and tectonics of the Karakoram Mountains. Tectonics. M. R. Pakistan. Bull. In Geodynamics of Pakistan. 232. B. 20. M. 160. 1991. (Eds. HYPOELLIPSE: 1999. Geol. Spec. I. Death Valley Region.." USGS Open File Report. 1987. P. 1989. Pakistan. Tectonophysics.. 189-190.