prev

next

out of 13

Published on

23-Jan-2015View

255Download

1

DESCRIPTION

Transcript

<ul><li> 1. International Journal of Civil Engineering and OF CIVIL ENGINEERING AND INTERNATIONAL JOURNAL Technology (IJCIET), ISSN 0976 6308 (Print), ISSN 0976 6316(Online) Volume 3, Issue 2, July- December (2012), IAEME TECHNOLOGY (IJCIET)ISSN 0976 6308 (Print)ISSN 0976 6316(Online)Volume 3, Issue 2, July- December (2012), pp. 279-291IJCIET IAEME: www.iaeme.com/ijciet.aspJournal Impact Factor (2012): 3.1861 (Calculated by GISI) IAEMEwww.jifactor.com QUALITY FACTOR OF SEISMIC CODA WAVES IN GARHWALHIMALAYASPriyamvada Singh, J.N. Tripathi Department of Earth and Planetary Sciences, University of Allahabad, IndiaEmail: priyam028@yahoo.com ABSTRACTSeismic wave propagating through the earth experiences some reduction in the energy content. This decay in the wave energy is known as the seismic wave attenuation. The study of attenuation characteristics of these waves shed light on the heterogeneous nature of the Earth. Usually, seismic wave attenuation for local earthquakes is determined from the analysis of coda waves.Digital seismogram data of 75 earthquakes that occurred in Garhwal Himalaya region during 2004 to 2006 and recorded at different stations have been analyzed to study the seismic coda wave attenuation characteristic in this region. In the present study, 90 seismic observations from local earthquake events with hypocentral distance less than 250 km and magnitude range between 1.0 and 5.0 is used to study coda Q , i.e. Qc , using the single isotropic scattering model. QC Values are estimated at 10 central frequencies 1.5, 3, 5, 7, 9, 12, 16, 20, 24 and 28 Hz using a starting lapse-time LT=50 s and four coda window-lengths , WL= 10, 20, 30, 40 s . In the considered frequency range, QC fit the frequency dependent power-law QC = Q0 f n . The frequency dependent power-law for 50 sec lapse time with 10 sec coda window length is QC = 61.8 f 0.992 and for 50 sec lapse time with 40 sec coda window length is QC = 161.1 f 0.998 . The Q0 ( QC at 1 Hz) estimates vary from about 61.8 for a 50 sec lapse time and 10 sec window length, to about 161.1 for a 50 sec lapse time and 40 sec window length combination. The exponent of the frequency dependence law n ranges from 1.016 to 0.967, which correlates well with the values obtained in other seismically and tectonically active and heterogeneous regions of the world. It is observed for the study region that QC values increases both with respect to window length and frequency. The low QC values or high attenuation at lower frequencies and high QC values 279</li></ul><p> 2. International Journal of Civil Engineering and Technology (IJCIET), ISSN 0976 6308 (Print),ISSN 0976 6316(Online) Volume 3, Issue 2, July- December (2012), IAEMEor low attenuation at higher frequency may indicate that the heterogeneity decreases withincreasing depth, in the study region.Keywords: Attenuation, Coda Q, Single backscattering model, Lapse time window, GarhwalHimalayaI. INTRODUCTIONThe attenuation of seismic wave is one of the basic physical parameter which is closely related tothe seismicity and regional tectonic activity of a particular area. This is also important forseismic hazard measurement. In this work, the seismic attenuation in the Garhwal Himalayas isstudied using local earthquakes. The amplitude of seismic waves decreases with increasingdistance from the earthquake. This reduction of the energy content cannot be explained bygeometrical spreading of the wave only. This decay in the wave energy is known as the seismicwave attenuation. Seismic wave attenuate because the earth is not a perfect elastic andhomogeneous. Usually, seismic wave attenuation for local earthquakes is determined from theanalysis of direct body waves, surface waves or coda waves. The dimensionless parameter, Q , isstudied in the present work which is defined as a measure of the rate of decay of the coda waveswithin a specified frequency band. Aki (1969) referred Coda as the tail part of seismograms oflocal earthquakes. Aki and Chouet (1975) suggested that the S Coda of local earthquakes issuperposition of incoherent backscattered S-wave and surface waves generated from numerousheterogeneity distributed randomly in the Earths crust and upper mantle. The great variety ofpaths traveled by these waves provides information concerning the average attenuationproperties of the medium instead of just the characteristics of a particular path (Aki and Chouet1975).The QC , quality factor of Coda wave has been estimated for different parts of the world (Aki andChouet 1975; Sato 1977; Ugalde et al., 2002; Tripathi and Ugalde, 2004, Ugalde et al., 2007,Pezzo et al., 2011,). Coda wave characteristics have also been estimated for different parts of theHimalayas (Gupta et al., 1995; Kumar et al., 2005; Hazarika et al., 2009; Sharma et al., 2009;Mukhopadhyaya et al., 2010; Padhy et al., 2010; Tripathi et al., 2012).In the present work the coda attenuation properties have been estimated in the Garhwal region ofHimalayas using local earthquakes. The frequency dependence of coda wave is also estimated.II. STUDY AREAThe Himalayas is the consequence of the collision of the Indian plate with the plates of centralAsia during mid to late Eocene. The Outer Himalayas, Lower Himalayas and the HigherHimalayas are the three major terrains identified in the Garhwal Himalayas [Valdiya, (1980)]. 280 3. International Journal of Civil Engineering and Technology (IJCIET), ISSN 0976 6308 (Print),ISSN 0976 6316(Online) Volume 3, Issue 2, July- December (2012), IAEMEThe major thrust fault striking parallel to the Himalayan arc, from north to south are the MainCentral Thrust (MCT), the Main Boundary Thrust (MBT) and the Himalayan Frontal Thrust(HFT) (Figure 1).The high grade metamorphic units of Higher Himalayas, situated north ofMCT, are considered to be inactive generally, due to no signs of break of Quaternary deposits(Ni and Barazangi, 1984; Brunel 1986). The outer Himalayas comprises of Tertiary rocks that isunderlain by the marine water to brackish origin subathu formation. This is followed upward bysiwalik group. The siwalik group is overlain by Quaternary gravel and sand. The LowerHimalayas are mainly made up of Precambrian sedimentary rocks with some outcrops ofCambrian Tal formation. The Higher Himalayas are made up of high grade metamorphic rockslike amphibolites to granulites grade metasedimentary rocks, auger gneisses and intrusiveleucogranite. In the Garhwal-Kumaon Himalayas region these groups of rocks are known as thevaikrita group (Srivastav and Mitra 1994)Figure 1: (a) Simplified map of the Himalya. (b) Map of the study area modified after Valdiya(1980). 281 4. International Journal of Civil Engineering and Technology (IJCIET), ISSN 0976 6308 (Print),ISSN 0976 6316(Online) Volume 3, Issue 2, July- December (2012), IAEMEIII. METHOD AND DATAThe single backscattering model of coda wave envelopes of Aki and Chouet (1975) considers thecoincidence of source and receiver. So, for the practical application of the model, we have toconsider lapse time t >2 t s , where t s is the S wave travel time (Rautian and Khalturin, 1978).Sato (1977) proposed the single isotropic scattering model for non coincident source andreceiver. Thus, we can analyse the coda window just after the S wave arrival. In this model, it isassumed that the elastic energy is radiated spherically, scatterers are distributed homogeneouslyand randomly, and the single scattering is isotropic in the media. Thus, the coda energy density E S at frequency f can be expressed as W ( f ) g 0 ( f ) ES ( f | r , t ) = 0 4r 2 [ 1 K ( ) exp 2QC ft ] (1) Where t is the lapse time measured from the origin time of the earthquake, t s is the S-wavetravel time, r is the hypocentral distance, W0 is the total energy radiated from the source, g 0 isthe total scattering coefficient, and 1 +1K ( ) =ln ,( > 1); and = t / t s . (2) 1The energy density is considered to be proportional to the mean square amplitudes of coda wavesand taking natural logarithms of Eq.1 and reshuffling the terms, we get A ( f | r, t ) f ln obs = ln C ( f ) Q t(3) k (r , ) Cwhere Aobs ( f | r , t ) represents the observed root mean square (rms) amplitude of the narrow bandpass filtered waveforms with central frequency f ; k(r , ) = (1 / r )K ( ) 0.5 and C ( f ) is a constant.Thus the QC can be easily obtained from the slope b of the least square fit straight line to themeasured ln[ Aobs ( f | r , t ) / k (r , )] versus t for a given central frequency, using the relation nQC = f / b . The frequency dependence law, QC = Q0 fis also fitted to the QC data fordifferent lapse time and window length, where Q0 is the value of QC at 1 Hz and n is frequencydependent parameter (Table 3).The digital waveform seismograms of 75 events used for coda attenuation in the present studywere recorded at 20 stations of Garhwal Himalayas during 2004 to 2006 (Figure 2, Table 1). TheCMG 40T1 triaxial broadband seismometers were used for the digital data collection. The data is 282 5. International Journal of Civil Engineering and Technology (IJCIET), ISSN 0976 6308 (Print),ISSN 0976 6316(Online) Volume 3, Issue 2, July- December (2012), IAEMEacquired in continuous mode at 100 samples per second for three components at the stations.SEISAN (version 8.1) software package (Havskov and ottemoller, 2005) was used to pick P andS wave arrival times of each earthquake recorded at the different seismic stations. Thehypocentral parameters, viz, origin time, latitude, longitude and focal depth of these events werealso computed using the SEISAN software. Most of the events are within the crust and localmagnitude ranges from 1.0 to 5.0. First of all we preformed a visual inspection of more than 450seismograms, 90 waveforms with hypocentral distances less than 250 km have been finallyprocessed for the present work.IV. DATA ANALYSIS AND RESULTSFirst of all the seismograms were band pass filtered for ten frequency bands, 1.5 0.5 Hz, 3 1Hz, 5 1 Hz, 7 1 Hz, 9 1 Hz, 12 1 Hz, 16 1 Hz, 20 2 Hz, 24 2 Hz, and 28 2 Hz(Table 2), using eight-pole Butterworth filters. As the sampling rate was 100 samples per second,the maximum frequency for which reliable result could be obtained was 50 Hz. Then, the rootmean squared amplitudes of the filtered seismograms were computed at an interval of 0.5 s withmoving time windows of length t 2s for the first frequency band and t 1s for the next ninefrequency bands. Then QC was estimated applying a least square regression technique to Eq.3for one starting lapse time window length LT = 50 s from the S-wave onset, having WindowLength WL=10, 20, 30 and 40s for ln[ Aobs ( f | r , t ) / k (r , )] . The QC estimates were computedonly for the amplitudes greater than signal to noise ratios. The coda wave is analyzed only thevertical component, because it has been shown that the coda analysis is independent of thecomponent of the particle ground motion analyzed (Hoshiba, 1993). The estimated QC valuesretained for further analysis which were having correlation coefficients greater than 0.5. Stations Location 35.00 34.00 33.00Latitute 32.00 Events 31.00 Stations 30.00 29.00 72.00 74.00 76.00 78.00 80.0082.00LongitudeFigure 2: Station Locations (open circles) with events used in the Study. 283 6. International Journal of Civil Engineering and Technology (IJCIET), ISSN 0976 6308 (Print),ISSN 0976 6316(Online) Volume 3, Issue 2, July- December (2012), IAEMETable 1: Station code and their location Station Longitude Latitude DEO76.67 32.09 TZG76.79 32.59 UDA76.67 32.72 CHT76.37 32.45 TSA76.14 32.82 UNA76.32 31.52 LGR75.91 32.29 BNK75.94 32.55 RJA76.24 32.00 BRM76.54 32.44 PAL78.62 30.81 GRG79.44 30.46 JKH78.4330.4 PRT78.48 30.46 YOL 76.4 32.17AMB 76.04 31.67BEED75.94 32.58 NEL78.5230.4 GYL78.51 30.36 NAD76.31 32.24The frequency dependent Coda Q relationship provides average attenuation characteristics ofthe medium. The average values of QC at different frequencies, one lapse time and four windowlengths obtained from the mean values for the whole study area are given in Table.3.Table.2: Central frequencies and frequency range as low and high cutoff. Low cutoff Central frequency High cutoff (HZ) (Hz) (Hz) 1 1.52 234 456 678 89 10101214141618182022222426262830284 7. International Journal of Civil Engineering and Technology (IJCIET), ISSN 0976 6308 (Print),ISSN 0976 6316(Online) Volume 3, Issue 2, July- December (2012), IAEMETable 3: The average numerical values of QC at different frequencies, lapse time (LT=50)and coda window length (WL=10, 20, 30, and 40 s).LTWLValues of QC at different frequencies(s) (s) 1.5 Hz 3 Hz 5 Hz 7 Hz9 Hz 12 Hz16 Hz20 Hz24 Hz 28 Hz5010121.00 171.96 287.8386.55 553.72856.37 1249.761596.241700.58 1790.045020179.10 294.54 511.71 732.391002.40 1553.47 1825.292091.452382.79 2611.675030201.03 333.08 587.83 806.851059.86 1707.75 2228.782702.073153.73202.795040279.23 500.61 863.62 1166.27 1487.98 2212.82 2718.69 3384.61 4523.86 5200.77For the study area it is observed that QC value increases both with respect to frequency andwindow length. It is observed that the QC increases with frequency. The average value of QC forthe study region varies from 121 at 1.5 Hz to 1790 at 28 Hz for lapse time 50s and windowlength 10s. When window length is 20, QC is 179 at 1.5Hz and 2611 at 28Hz. Higher values ofQC are obtained at 30 and 40s window lengths. This observation of frequency dependence ofQC is due to the degree of heterogeneity of a medium and level of tectonic activity in an area(Aki 1980). The low QC values or high attenuation at lower frequencies may indicate a highdegree of heterogeneity and decrease in rock strength at shallow parts. The high QC values orlow attenuation at higher frequencies may be related to the comparatively less hetrogeneousdeeper zones (Aki and Chouet, 1975).Gupta et al. (1995) obtained a frequency relation QC = 126 f 0.95 using records of seven microearthquake in the adjoining southwestern part of Garhwali Himalayas for 30s coda windowlength.Kumar et al., (2005) employed the time domain coda - decay method of a single - back scattering model to calculate frequency dependent values of coda QC . A total of 36 localearthquake of magnitude range 2.4 - 4.8 have been used for QC estimation at central frequencies1.5, 3.6, 6.9, 9.0, 12.0 and 18.0 Hz through eight lapse time windows from 25 to 60s starting atdouble the time of the primary S-wave from the origin time. The estimated average frequencydependence quality factor gives the relation QC = 158 f 1.05 while the average QC values vary fromthe relation 210 at 1.5Hz to 2861 at 18Hz central frequencies. The observed coda quality factor isstrongly dependent on frequency, which indicate that the region is Seismic and tectonicallyactive with high heterogeneity.Paul et.al., (2003) estimated QC for Kumaun Himalayas using data from eight micro earthquakerecord by a five station array with epicentral distance range varying between 10 km to 80 km...</p>