Lapse time and frequency-dependent attenuation of coda waves in the Zagros continental collision zone in Southwestern Iran

Please download to get full document.

View again

of 8
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Document Description
Lapse time and frequency-dependent attenuation of coda waves in the Zagros continental collision zone in Southwestern Iran
Document Share
Document Tags
Document Transcript
  Lapse time and frequency-dependent attenuation characteristicsof Kumaun Himalaya Chandrani Singh a, ⇑ , V.K. Srinivasa Bharathi b , R.K. Chadha b a Department of Geology and Geophysics, Indian Institute of Technology Kharagpur, West Bengal 721 302, India b National Geophysical Research Institute (Council of Scientific and Industrial Research), Uppal Road, Hyderabad 500 007, India a r t i c l e i n f o  Article history: Received 11 May 2011Received in revised form 2 March 2012Accepted 13 March 2012Available online 7 April 2012 Keywords: AttenuationCoda  Q  Lapse timeTurbidityKumaun Himalaya a b s t r a c t Wehaveanalyzedlocalearthquakedatasetconsistsof84welllocatedevents,recordedbyadigitalseismicnetworkintheKumaunHimalayaregionduring2004–2007tostudyseismicattenuationcharacteristicsof the region. Single back-scattering assumption is used to estimate coda  Q   ( Q  c  ) values of the region at fre-quency range of 1.5, 3, 6, 8, 12 and 18Hz for different lapse time window length. The values of   Q  c   showa ubiquitous observation of frequency dependence and follow a substantially similar trend as observedinothertectonicallyactivepartsoftheHimalaya. Thelapsetimedependenciesof codawavesareinvesti-gated which reveal that by increasing lapse time window from 20 to 50s,  Q  0  ( Q  c   at 1Hz) increases from64±2to230±19whilefrequencydependentparameter n decreasesfrom1.08to0.81.Increasein Q  c  withlapse time indicates decrease in heterogeneity with depth. The observed values of   Q  0  and  n  infer that themedium beneath the study area is highly heterogeneous and tectonically very active. The variation of   Q  c  with lapse time at 1.5Hz follow a similar pattern with the theoretically predicted values of  Gusev(1995)althoughthefrequencyparameter n doesnotmatchwiththemodel.Itmaysuggestthattheturbid-ity decays rapidly with depth for the Kumaun Himalaya. Our results are comparable with those obtainedfrom tectonically active regions in the world.   2012 Elsevier Ltd. All rights reserved. 1. Introduction Knowledgeofattenuationcharacteristicsofaregionisanecessaryfactor in the earthquake hazard assessment study (Anderson et al.,1996) and for understanding the source processes (Abercrombie,1997).Theattenuationpropertyofamediumisexpressedbyadimen-sionlessquantitycalledqualityfactor Q  ,whichrepresentsthedecayof waveamplitudeduringitspropagationinthemediumcausedbyhet-erogeneityoranelasticityorbothinthecrust(LayandWallace,1995).Thequalityfactor( Q  )canbeestimatedbyusingrateoftimedecayof thecoda-waveamplitude(called Q  C  )orofthedirect P  ( Q  P  ),direct S  ( Q  S  )or  Lg   ( Q  Lg  ) wave amplitude (Aki, 1969; Sato, 1977; Yoshimoto et al., 1993; Kim et al., 2004; Mahood et al., 2009; Singh et al., 2011ab, 2012). While dealing with local earthquake data coda  Q   or  Q  c   is usedto measure the attenuation in the Earth’s crust. The term ‘‘Coda’’ isused by Aki (1969) to describe the tail portion of the seismograms.He assumed that the coda waves of local earthquakes are backscat-tered waves from numerous randomly distributed heterogeneities,and suggested a statistical treatment in which a small number of parameters characterize the average properties of the heterogeneousmedium.Low Q  hasbeenobservedforseismicallyactiveregionscom-paredtostableareas.Theattenuationpropertiesofvariousseismicre-gionsintheworldhavebeendeterminedbyanumberofinvestigators(Aki,1969;AkiandChouet,1975;Sato,1977;Pulli,1984;Rhea,1984;  JinandAki,1986;Havskovetal.,1989;Ibañezetal.,1990;Mandaland Rastogi, 1998; Mak et al., 2004; Wu et al., 2006; Mukhopadhyay and Tyagi,2007;Mukhopadhyayetal.,2008).Itisusefultostudytheatten-uationcharacteristicsforunderstandingthesubsurfacemediumprop-erties of a given region. The dependence of   Q  c   on frequency isespecially important. Aki (1980), Roecker et al. (1982) and Pulli (1984) have determined a correlation between the dependence of   Q  c  onfrequencyandthetectoniccomplexityofaregion,therebylendingsupporttothescatteringhypothesis.Areasofstrongtectonicheteroge-neityshowastrongfrequencydependenceof  Q  c  .AnattempthasbeenmadeheretounderstandthefrequencyandlapsetimedependenceofattenuationcharacteristicsoftheKumaunHimalayabyusingthecodaoflocalearthquakes.ThispartofHima-laya is seismically very active that manifests strong deformationandreactivationofsomeofthefaultsandthrustsduringQuaternary(Valdiya, 1999). In an earlier study, Paul et al. (2003) estimated  Q  c  for this region using eight local earthquakes for a single lapse timewindow. We made a comprehensive study of   Q  c   using 84 well lo-cated local earthquakes considering lapse time windows 20–50s.Further, our results are compared with those obtained in variousparts of the Himalayas as well as in different tectonic regions of the world. 1367-9120/$ - see front matter   2012 Elsevier Ltd. All rights reserved. ⇑ Corresponding author. Tel.: +91 03222 283378. E-mail address: (C. Singh). Journal of Asian Earth Sciences 54–55 (2012) 64–71 Contents lists available at SciVerse ScienceDirect  Journal of Asian Earth Sciences journal homepage:  2. Seismotectonics of the region Himalayaisalargegeodynamiclaboratoryofnaturewhereoro-genis still in youthto early maturephases of evolution. It is one of the most active orogens of the world and consequence of the col-lision of the Indian plate with the collage of previously suturedmicrocontinental plates of central Asia during mid to late Eocene(Valdiya, 1980). A detail seismicity map of the Himalayas since1500 is provided by Mukhopadhyay et al. (2010). The Kumaun re- gion of the Himalaya lies near the center of the Himalayan fold-and-thrust belt and is situated between the Kali River in the eastand Sutlej in the west that includes a 320kmstretch of mountain-ous terrain. Fig. 1 shows the geology and tectonicsof the area. Thispart of the Himalaya exposes all the four major litho-tectonic sub-divisions: (a) Sub-Himalaya, (b) Lesser Himalaya, (c) Great Hima-laya and (d) Tethys Himalaya. In the Kumaun Himalaya regionthe groups of rocks are known as Vaikriti group (Valdiya, 1980).Extension of the Aravalli structures into the Himalaya plays a ma- jorroleinactivetectonicsoftheKumaunHimalayaanditsseismic-ity. Kumaun Himalaya evolved by an overall forelandwardprogression of thrusting, with some reactivation along the Muns-iari Thrust (MT), the Main Boundary Thrust (MBT) and the MainCentral Thrust (MCT); maximum strain-energy release is relatedto the MCT (Srivastava and Mitra, 1994). Strong deformation and reactivation of some of the faults and thrusts during Quaternaryare apparent in the Kumaun sector (Valdiya, 1980), which is evi-dent by the recurrent seismicity patterns, geomorphic develop-ments and by geodetic surveys (Valdiya, 1999). 3. Data and methodology  We have used 84 local earthquakes ( M   >4.5) recorded by atleast four three-component broadband seismic stations in the Ku-maun Himalaya during 2004–2007 (Fig. 2a). Focal depths of theearthquakesarewithin35km(Fig. 2bandc). About275highqual-ity seismograms were available for the present study. Sample seis-mograms are shown in Fig. 3.The single back-scattering model is considered to be a first or-der approximation of attenuation characteristics of the real Earth(Aki and Chouet, 1975) and this model is extensively used forregular estimation of quality factor  Q  c   of coda waves. It is foundthat estimated  Q  c   is similar to direct  S  -wave quality factor ( Qs ),and represent the attenuation characteristics of a given region(Gao et al., 1983; Frankel and Wennerberg, 1987). Aki (1985) observed agreement between  Q  c   and  Qs , and concluded that the  S  coda is formed by single  S  -to- S   backscattered waves in the1–25Hz range. We have used the single back-scattering model inthis study.Assuming that coda waves are composed of single back-scat-tered waves from randomly distributed heterogeneities, Aki andChouet (1975) obtained the following relation:  A ð  f  ; t  Þ ¼  C  ð  f  Þ t   a exp ð p  ft  = Q  C  Þ ð 1 Þ where  A(f, t  ) is the coda amplitude for a central frequency‘  f  ’ over anarrow bandwidth signal. The lapse time ( t  ) is measured from theorigin time of the seismic event and  Q  C   represents the averageattenuation property of the medium.  C(f  ) is the coda source factorwhichisconsideredasconstant,and‘ a ’isthegeometricalspreadingfactor ( a  =1 for body waves). Rautian and Khalturin (1978),however, found that equation (1) is valid only for the lapse time, t  , greaterthanabout twotimesthe S  -wavetravel time,  t  s . Assumingthesource factor,  C  (  f  ), to be independent of timeand radiationpat-tern, and the geometrical spreading parameter (a) equals to 1 forbody waves, the above equation can be rewritten as ln ½  A ð  f  ; t  Þ   t   ¼  c     bt   ð 2 Þ where  b  = p  f  / Q  C   and  c   =ln C  (  f  ). Eq. (2) represents the equation of a straight line, slope of which( b  = p  f  / Q  C  ) provides the  Q  c  , for the central frequency  f  . In order toestimate  Q  c   using this method, data are normally taken from thepart of the coda coming after lapse time of about twice the  S  -wavetravel time to avoid contamination by direct  S  -wave (Rautian andKhalturin, 1978). We filtered the seismograms using Butterworthband pass filter using six different frequency bands. The low cut-off, high cut-off, and central frequencies of these bands are givenin Table 1. The beginning of coda is considered at 2 t  s , where  t  s  isthe  S  -wave travel time from the srcin time. The filtered seismictraces within these coda windows are smoothed by calculatingroot mean square (rms) values of coda amplitudes of the filteredseismograms with a sliding window of length equal to 5cycles of each central frequency.  Q  c   is then estimated from the slope of the least-square fit straight line of the plot between ln[  A (  f  , t  ) ⁄  t  ]and lapse time ( t  ) for coda window of 20s, 30s, 40s and 50s.Havskov and Ottemoller (2005) suggest that a minimum value of coda window length is 20s for stable results. The  Q  c   value is ac-cepted only when correlation coefficients for the best- fit line forthe coda decay slope with respect to lapse time were greater than 30 ° 80 ° 29 ° 79 ° movementorogenyFaultThrust M  BT  MCT 81 ° Fig. 1.  The geology and tectonic map of the study area modified after GSI (2000). MBT: Main Boundary Thrust, MCT: Main Central Thrust. C. Singh et al./Journal of Asian Earth Sciences 54–55 (2012) 64–71  65  0.5andthe S  / N  ratioisgreaterthan2foragivendataset.Assumingthe power-law frequency dependence of   Q  c  , the relation  Q  C   = Q  0  f  n is used to estimate  Q  0  and  n  values (Fig. 4), where  Q  0  is the valueof   Q  c   at 1Hz and  n  is the frequency dependent coefficient. 4. Results and discussion 4.1. Frequency dependence of Coda Q  The frequency-dependent coda  Q   relationship provides averageattenuation characteristics of the mediumproperties of a localizedzone around the study area. The estimated  Q  c   values for theKumaun Himalaya region are determined as a function of frequency by varying lapse time windows of 20–50s as shown inFig. 4. The average values of   Q  c   obtained from the mean values of different stations and for the whole region are given in Table 2. Itis observed that  Q  c   increases with increasing frequency. The aver-age value of   Q  c   for the whole region varies from 134 at 1.5Hz to1951 at 18Hz for  t  c   =20s; from 237 at 1.5Hz to 2621 at 18Hzfor  t  c   =30s; from 318 at 1.5Hz to 2867 at 18Hz for  t  c   =40s andfrom 402 at 1.5Hz to 2983 at 18Hz for  t  c   =50s. This seems tobe a ubiquitous observation of frequency dependence of   Q  c   (Aki,1980). It depends on both degree of heterogeneity of a mediumand level of tectonic activity in an area (Aki, 1980; Pulli and Aki,1981; Roecker et al., 1982). The low  Q  c   values (high attenuation)at lower frequencies may indicate a high degree of heterogeneityand decrease in rock strength at shallower part whereas the high Q  c   values (low attenuation) at higher frequencies may be relatedto relatively more homogeneous deeper zones.  Table 1 Filter characteristics used for  Q  c   analysis in the present study. Low cutoff (Hz) Central frequency (Hz) High cutoff (Hz)1 1.5 22 3 44 6 85.34 8 10.668 12 1612 18 24 0102030 Depth(Km) 272829303132    L  a   t   i   t  u   d  e 777879808182 Longitude 0102030    D  e  p   t   h   (   K  m   ) (a)(b)(c) Fig. 2.  (a) Distribution of ray paths connecting selected 84 earthquakes (circles) and four stations (inverted triangles) used in this study. Tectonic features are alsosuperimposed in the map. (b) and (c) Depth distribution of the selected events. MBT: Main Boundary Thrust, MCT: Main Central Thrust, ITSZ: Indus-Tsangpo Suture Zone,STD: Southern Tibet Detachment. Small square in the inset represents the study area. Fig. 3.  Typical example of three-component Seismograms recorded at station ALM.66  C. Singh et al./Journal of Asian Earth Sciences 54–55 (2012) 64–71  The  n  parameter represents the level of medium heterogeneityand tectonic activity of a region. A strong correlation between thedegree of frequency dependence ( n  value) and the level of tectonicactivity was observed by several authors in different parts of the 048121620 Frequency (Hz) 04008001200160020002400       Q     c Frequency (Hz) 0100020003000       Q     c Frequency (Hz) 01000200030004000       Q     c 08121620048121620048121620 Frequency (Hz) 01000200030004000       Q     c (a) Qc= (64±2)f  (1.08±0.01)  at lapse time 20sec KSN KLT DCH ALM Whole areaKSN KLT DCH ALM Whole areaKSN KLT DCH ALM Whole areaKSN KLT DCH ALM Whole area (0.97±0.03) (b) Qc= (121±8)f at lapse time 30sec(C) Qc= (172±15)f  (0.88±0.04)  at lapse time 40sec(d) Qc= (230±19)f  (0.81±0.04)  at lapse time 50sec 4 Fig. 4.  Plot of   Q  C   values obtainedfor the wholeKumaunHimalayaregionas afunctionof frequencyfor different lapse timewindows. Standarddeviations areshown aserrorbars.  Table 2 Values of   Q  C   at five central frequencies for (a) 20 s, (b) 30 s (c) 40 s and (d) 50 s lapse time windows for each station and the whole study area along with their average values. Stations 1.5Hz 3Hz 6Hz 8Hz 12Hz 18Hz  Q  c  (  f  ) n (a) KSN 116 253 555 768 1216 1925 (53±5)  f   (1.13±0.05) KLT 134 285 606 871 1362 1825 (63±4)  f   (1.09±0.03) DCH 134 283 596 812 1258 1947 (63±8)  f   (0.08±0.05) ALM 150 307 630 848 1293 1969 (73±7)  f   (1.04±0.05) Whole area 134 283 598 814 1261 1951 (64±2)  f   (1.08±0.01) (b) KSN 206 410 817 1087 1627 2436 (103±23)  f   (0.99±0.08) KLT 231 461 920 1225 1836 2751 (116±13)  f   (1.00±0.05) DCH 275 509 943 1217 1747 2505 (148±17)  f   (0.89±0.05) ALM 236 474 950 1267 1906 2863 (118±10)  f   (1.00±0.04) Whole area 237 463 906 1196 1771 2621 (121±8)  f   (0.97±0.03) (c) KSN 312 561 1009 1287 1816 2560 (173±31)  f   (0.85±0.09) KLT 316 594 1116 1449 2097 3033 (168±13)  f   (0.91±0.03) DCH 334 599 1071 1364 1918 2696 (187±27)  f   (0.84±0.06) ALM 317 601 1137 1482 2154 3130 (167±44)  f   (0.92±0.11) Whole area 318 587 1085 1398 2003 2867 (172±15)  f   (0.88±0.04) (d) KSN 421 676 1084 1319 1739 2293 (263±50)  f   (0.68±0.09) KLT 455 747 1227 1506 2014 2692 (277±31)  f   (0.72±0.05) DCH 377 669 1188 1507 2109 2950 (212±22)  f   (0.83±0.05) ALM 425 773 1407 1804 2562 3637 (233±47)  f   (0.86±0.08) Whole area 402 703 1230 1550 2151 2983 (230±19)  f   (0.81±0.04) C. Singh et al./Journal of Asian Earth Sciences 54–55 (2012) 64–71  67  world (e.g., Aki, 1980; Pulli and Aki, 1981; Roecker et al., 1982; Van-Eck, 1988; Akinci et al., 1994). They ascertained that  n  valueis higher for tectonically active regions compared to the tectoni-cally stable regions. The obtained values of   Q  0  and  n  in the presentstudy indicate that the Kumaun Himalaya region is highly hetero-geneousandtectonicallyveryactive.Strongfrequencydependencebehavior could be related to the size of heterogeneities (Mayedaet al. 1992). This heterogeneity may be due to the convergence of the north dipping Indian plate under the southward movementof the Eurasian plate which results long structural faults/thrustsand folds in the Himalaya in addition to many localized thrusts,faults and minor lineaments. On the basis of travel time tomogra-phy Sharma (2008) showed that the crust in the Garhwal andKumaun area is highly heterogeneous. 4.2. Lapse time dependence Sato (1978) and Pulli (1984) have shown that the scatterers responsible for the generation of coda waves are generally as-sumedto be distributedover the surfacearea of anellipsoidwhichcan be estimated using the following equation:  x 2 v  t  c  2   2  þ  y 2 v  t  c  2   2   R 2 4    ¼  1  ð 3 Þ Depth 100200300400500    Q    C Lapse time (S)Depth (Km) 200300400500600700800 Lapse time (S) (a) 1.5Hz(b) 3Hz Depth (Km) 400600800100012001400 Lapse time (S) (c) 6Hz Depth (Km) 60080010001200140016001800 Lapse time (S) (d) 8Hz Depth (Km) 1200160020002400 Lapse time (S)Depth (Km) 16002000240028003200360080 100 120 140 16050 60 70 80 90 50 60 70 80 90 Lapse time (S) (e) 12Hz(f) 18Hz    Q    C    Q    C    Q    C    Q    C    Q    C 50 60 70 80 9050 60 70 80 90 50 60 70 80 9050 60 70 80 9080 100 120 140 16080 100 120 140 160 80 100 120 140 16080 100 120 140 16080 100 120 140 160 Fig. 5.  Plot of   Q  C   values versus average lapse time/depth for the central frequencies of (a) 1.5Hz, (b) 3Hz, (c) 6Hz, (d) 8Hz, (e) 12Hz and (f) 18Hz. Standard deviations areshown as error bars.68  C. Singh et al./Journal of Asian Earth Sciences 54–55 (2012) 64–71
Similar documents
View more...
Search Related
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks