“P- AND S-WAVE ATTENUATION IN THE KOPILI REGION OF NORTHEASTERN INDIA„

We comprehensively analyze attenuation factor in Kopili region and its neighboring area. Corresponding to influence of frequency on attenuation, QP and QS are estimated from direct P and S waves. The results establishes higher values of Q for S waves (QS=(101±4)f(1.4±0.05)) than for P waves (QP=(45±3)f(1.3±0.07))in the 1.5-12 Hz frequency band. It is seen that our findings agree with other published results on tectonically active regions that is characterized by high degree of heterogeneity. Hence, it can be useful for further studies in ground motion prediction in Kopili region. The low value of quality factors corresponding to P and S waves refers to seismically actives areas with complex heterogeneity. The estimated value of QS/QP and their subsequent association with frequency affirm the dominance of scattering as a major contribution to diminution of seismic amplitudes. This is further corroborated by local geological settings. 1980; Sato and Matsumura, 1980; Hough and Anderson, 1988; Masuda, 1988; Takemura et al., 1991; Campillo and Plantet, 1991; Scherbaum and Sato, 1991; Yoshimoto et al., 1993; Yoshimoto et al., 1998]. To mitigate the seismic hazard of future calamitous earthquakes, a good understanding of attenuation and properties of the medium is required. To measure the attenuation, various methods have been devised from different portions of the seismograms [e.g., Aki, 1969; Hermann, 1980; Mitchell, 1995]. Wu [1985] suggested a technique from the dependence of total S-wave energy on hypocentral distance, for the assessment of the relative impact of scattering and intrinsic attenuation. Frankel and Wennerberg [1987] used the energy flux model of coda portion of the seismogram to obtain attenuation quality factor based on decay of coda amplitude. Hoshiba et al. [1991] developed a method based on Monte Carlo simulations of the temporal shape of the coda envelope. Aki [1980] introduced the coda normalization method, based on the idea that coda waves consist of scattered S-waves from random heterogeneities in the Earth. Sato and Matsumura [1980] adopted Aki’s method to a set of deep borehole data. This method was extended by Yoshimoto et al., [1993] for simultaneous measurement of QP and QS. In this paper, we compute frequency-dependent QP and QS in Kopili region, India. This part is one of the highly seismically active part in North-east India. To gain further insight into the seismic attenuation mechanism of this region, the extended coda-normalization method proposed by Yoshimoto et al., [1993] was used to determine QP and QS. This is the first application of the technique to this region. We also examine QS/QP ratio and its possible implications for its variation. The findings are also being matched with existing data of dissimilar regions of the sphere and inferred. 2. SEISMOTECTONICS OF KOPILI REGION North east India has been established as one of the seismically active regions in the world with complex tectonic settings. As per reports [Kayal et al., 2006], seismic activity could be attributed to considerable deformation and reactivation of some faults during Quaternary period in Northeastern India. There were occurrences of two great earthquakes 1897 (MS 8.7) [Oldham, 1899] and 1950 (MS 8.6) [Poddar, 1950] and 16 large (MS=7.0) earthquakes within the last century. The region under study, Kopili Fault (KF) is encircled by diverse geological units. As for instance, the Himalayan frontal is situated to the north. Again, towards east, there exists highly folded Burmese arc. Besides, towards north, the main constituents comprise of thick sediments of Bengal. Among those earthquakes, KF had witnessed major event having magnitudes greater than seven [Biswas et al., 2018 a,b]. KF region is bounded by latitude 25.5° to 27° N and longitude 92° to 93° E. Additional evidence of seismic activity is borne by the earthquake of magnitude ~7.8 (MS), occurring at 60 km along KF on 10th of January 1869 [Kayal et al., 2010; Bora et al., 2017 a]. This wrecked a huge portion of areas between Silchar and Nagaon. On the same note, another event of 7.3 (MS) (27.5°N: 93.5°E) seemed to originate along KF on the year 1943. Recently, Kumar et al., [2016] reported that the entire region originated during the transition from Mesozoic to Tertinary period. Within this transition, rivers Kopili and Kalang had been channeling through this part, causing broad topographic depression (Figure 1). This KF possesses approximately width and length of 50 km and 300 km, respectively, covering Manipur, Arunachal Pradesh, Bhutan and Assam. According to BORA ET AL. 2 FIGURE 1. Map of Kopili Fault region [after Biswas et al., 2013a,b; Bora and Biswas, 2019]. The red circles symbolize the used events for this study, while triangles represents the temporarily deployed stations. The Great and major earthquakes occurred in and around this region was represented by the blue star symbol. (In inset the India map was shown). Kayal et al., [2006] and Kayal et al., [2010], KF is characterized by high seismic activity with events at a depth ~50 km. Bhattacharya et al., [2008] showed the depth of Kopili fault to range from 20 to 30 km. A high VP value at a depth of 40 km beneath Mikir Hills was also assessed by them. Meanwhile, Kundu and Gahalaut [2013] computed the active deformation of this fault as 2.9±1.5 mm per year (Slip rate). As reported by Kayal et al., [2006 and 2010], this fault was a normal strike slip fault, having a dip towards the N-E side.


INTRODUCTION
Seismic waves are in general influenced by path effect, source effect and site effect.While propagating, there occurs diminution of amplitudes as well as energy of these seismic waves.This is referred to as seismic attenuation, designated by the term Q -1 .Seismic attenuation arises basically from two constituents, viz, body and coda waves.Between them, attenuation from body waves bears signature of the formation and material state of the interior part of the Earth [Sato, 1992].
Again, attenuation from body waves has got paramount importance in the prediction of ground motion parameters.The causes for attenuation can be various factors.Geometrical spreading and redistribution of seismic energy due to its passage through diverse medium need special mention [Tsujiura, 1966;Aki, 1969Aki, ,1980;;Aki and Chouet, 1975;Sato and Fehler, 1998;Mukhopadhyay et al., 2010].
The quality factor Q attempts to quantify the diminution of progressive seismic amplitudes from source to receiver site.Q also relates to the energy disipation in accordance with Knopoff and Hudson [1964], Aki and Chouet [1975].Sato and Fehler [1998] reported that seismic activity of a region can be classified with magnitude of Q where lower Q characterizes higher activity and lower activity is implied by greater magnitude of Q.
Q also bears direct correspondence with total harmonic wave energy.The harmonic wave energy is mathematically expressed as: 2π/Q=-DE/E [Knopoff and Hudson, 1964], where E denotes total energy available in the harmonic wave and DE is the energy lost per cycle.Since P and S wave are manifested by varying seismic energies, body wave quality factors also exhibit different characteristics.Throughout the world, researchers have adopted diverse techniques to study quality factors [Aki, 1980;Sato and Matsumura, 1980;Hough and Anderson, 1988;Masuda, 1988;Takemura et al., 1991;Campillo and Plantet, 1991;Scherbaum and Sato, 1991;Yoshimoto et al., 1993;Yoshimoto et al., 1998].
To mitigate the seismic hazard of future calamitous earthquakes, a good understanding of attenuation and properties of the medium is required.To measure the attenuation, various methods have been devised from different portions of the seismograms [e.g., Aki, 1969;Hermann, 1980;Mitchell, 1995].Wu [1985] suggested a technique from the dependence of total S-wave energy on hypocentral distance, for the assessment of the relative impact of scattering and intrinsic attenuation.Frankel and Wennerberg [1987] used the energy flux model of coda portion of the seismogram to obtain attenuation quality factor based on decay of coda amplitude.Hoshiba et al. [1991] developed a method based on Monte Carlo simulations of the temporal shape of the coda envelope.Aki [1980] introduced the coda normalization method, based on the idea that coda waves consist of scattered S-waves from random heterogeneities in the Earth.Sato and Matsumura [1980] adopted Aki's method to a set of deep borehole data.This method was extended by Yoshimoto et al., [1993] for simultaneous measurement of Q P and Q S .
In this paper, we compute frequency-dependent Q P and Q S in Kopili region, India.This part is one of the highly seismically active part in North-east India.To gain further insight into the seismic attenuation mechanism of this region, the extended coda-normalization method proposed by Yoshimoto et al., [1993] was used to determine Q P and Q S .This is the first application of the technique to this region.We also examine Q S /Q P ratio and its possible implications for its variation.The findings are also being matched with existing data of dissimilar regions of the sphere and inferred.

SEISMOTECTONICS OF KOPILI REGION
North east India has been established as one of the seismically active regions in the world with complex tectonic settings.As per reports [Kayal et al., 2006], seismic activity could be attributed to considerable deformation and reactivation of some faults during Quaternary period in Northeastern India.There were occurrences of two great earthquakes 1897 (MS 8.7) [Oldham, 1899] and1950 (MS 8.6) [Poddar, 1950] and 16 large (MS=7.0)earthquakes within the last century.The region under study, Kopili Fault (KF) is encircled by diverse geological units.As for instance, the Himalayan frontal is situated to the north.Again, towards east, there exists highly folded Burmese arc.Besides, towards north, the main constituents comprise of thick sediments of Bengal.Among those earthquakes, KF had witnessed major event having magnitudes greater than seven [Biswas et al., 2018 a,b].
KF region is bounded by latitude 25.5° to 27° N and longitude 92° to 93° E. Additional evidence of seismic activity is borne by the earthquake of magnitude ~7.8 (MS), occurring at 60 km along KF on 10th of January 1869 [Kayal et al., 2010;Bora et al., 2017 a].This wrecked a huge portion of areas between Silchar and Nagaon.On the same note, another event of 7.3 (MS) (27.5°N: 93.5°E) seemed to originate along KF on the year 1943.Recently, Kumar et al., [2016] reported that the entire region originated during the transition from Mesozoic to Tertinary period.Within this transition, rivers Kopili and Kalang had been channeling through this part, causing broad topographic depression (Figure 1).This KF possesses approximately width and length of 50 km and 300 km, respectively, covering Manipur, Arunachal Pradesh, Bhutan and Assam.According to  Biswas et al., 2013a,b;Bora and Biswas, 2019].The red circles symbolize the used events for this study, while triangles represents the temporarily deployed stations.The Great and major earthquakes occurred in and around this region was represented by the blue star symbol.(In inset the India map was shown).Kayal et al., [2006] and Kayal et al., [2010], KF is characterized by high seismic activity with events at a depth ~50 km.Bhattacharya et al., [2008] showed the depth of Kopili fault to range from 20 to 30 km.A high VP value at a depth of 40 km beneath Mikir Hills was also assessed by them.Meanwhile, Kundu and Gahalaut [2013] computed the active deformation of this fault as 2.9±1.5 mm per year (Slip rate).As reported by Kayal et al., [2006 and2010], this fault was a normal strike slip fault, having a dip towards the N-E side.

DATASET
In this study, we have utilized 300 digital seismograms generated by events with magnitude ranging from 1.9 to 3.9 in ML.We have selected 6 stations as shown in Figure 1.All these stations are maintained by various Government agencies and institutions [National Geophysical Research Institute, India Meteorological department and NEIST-Jorhat].This network of six stations possessed GPS-synchronized Broadband Seismographs.The data was acquired at a sampling frequency of 100Hz.The scattering of epicenters pertinent to this study is portrayed in Figure 1.Out of these 300 digital seismograms, only 192 seismograms have been chosen in order to study the attenuation mechanism in the study region within an epicentral distance up to 180 km and having a depth between 8.3 to 42 km.The Figure 2, represents an example of signal to noise ratios of different parts of the seismogram.We selected the events such that they must have a good signal to noise ratio.The selected events are randomly distributed, so that the radiation pattern can be least significant [Negi et al., 2014].
The onset of P and S wave are considered with the aim of deriving the wave spectra.Since we have selected events with small epicentral distances, the difference between S and P wave arrivals has been relatively trivial.This has led us to choose small sample lengths.The reason behind this step is to rule out P-wave contamination as much as possible.Prior to the selection of sampling length, we implemented baseline correction and eventually removal of trend line.A customized Matlab FFT code has been applied to compute the wave spectra by incorporating instrument correction.

METHODOLOGY
In delineating attenuation characteristics, we implemented coda-normalization method, proposed by Aki [1980] and extended by Yoshimoto et al., [1993].Considering coda-waves to be constituted by back scattered S-waves from heterogeneities, the amplitude of S-waves is normalized by coda amplitude, measured at identical frequency bands.
Choosing the lapse time (t C ) to be more than twice of S-wave travel time (t S ) which is measured from source origin time, the coda-spectral amplitude is found to vary proportionately with source spectral amplitude (e.g., A S (f) or A P (f)) Equation 1 is the same as Equation 7 [Yoshimoto, 1993].The expression indicates that we can apply the spectra of coda wave to normalize the spectra of body waves.This is the premise for the extension of Aki [1980] formulations for S-waves.This method assumes that the scattering coefficient in the area is constant.As seen in Figure 1, the events are distributed in random direction nearby the receiver sites that is under consideration.Negi et al., [2014] opined that such kind of spreading supports in the elimination of the impact of source radiation effect to the optimal degree.Furthermore, the adoption of this method nullifies the instrument response, source effects and site amplification terms since they are inherent to both direct and coda-waves [Molnar et al., 1973;Rautian et al., 1978, Kumar et al., 2014, Bora and Biswas, 2017].
If Ap(f, r) and A S (f, r) represents the maximum amplitudes of direct P-and S-wave, respectively, and A C (f, t c ) be the rms value of coda spectral amplitude, then from taking the average of above-mentioned coda normalization method, Q P and Q S can be obtained by using the following equations: (2) (3) where f is frequency, r is the average hypocentral distance, V P is the average P-wave velocity, and V S is the average S-wave velocity.As reported by Aki [1980], if we use several P and S waves of the recorded events having diverse azimuth values, then we can neglect the influence of source radiation pattern which is one vital assumption in the inference of the above 2 equations.Yoshimoto et al., [1993] suggested that the geometrical spreading Z(r) acts as a significant factor as it influences the amplitude decline rate of seismic wave.Z(r) is proportional to r −α (where α is the geometrical spreading factor).In our study region, we assumed α to be unity.
It is obvious from the above equations (2 and 3) that the values of A S , A P are functions of frequency and hypocentral distance (r).As soon as we computed A S , A P and A C from the recorded earthquake data, the values of Q P and Q S can also be resolved at different central frequencies from the slopes of regression analysis, corresponding to the Equation 2 and Equation 3. It has been observed that Q increases with frequency (>1 Hz).This behavior can be described by a power law (4) where Q 0 is value of Q at 1 Hz, f is frequency and n is the frequency-dependent coefficient.
Taking logarithm of Equation 4 it becomes, (5) Once the value of Q(f) is known for a wide range of frequencies (f), the values of Q 0 and n can be determined for P and S waves in a medium.

DATA ANALYSIS
We used vertical-component seismograms for the Pwave and the north-south component for the S-wave investigation, since the maximum amplitudes of S wave are nearly equal in both north-south and east-west components.Based on the seismograms, we calculated the direct P and S wave's amplitudes in a 1.28-s time window.
Five different Butterworth band pass filters were applied to the seismograms (Table 2) that filter the separated ground motion.This is followed by determining the peak to peak amplitude and half of that value is assigned to be A P (,) and   (,) for P and S waves respectively.Coda-spectral amplitude A C (f, t c ) is calculated from north-south component within a 1.28-s time window.We choose the peak values, so that we can have the highest fluctuation of the body wave.The time window

P-AND S-WAVE ATTENUATION
for Coda-spectral amplitude was calculated from the twice of the onset of S wave arrival.Following Gupta et al., [1984], the P and S wave velocities are affixed at 6 km/s and 3.5 km/s, respectively.

RESULTS AND DISCUSSION
As stated earlier, the extended version of coda normalization method [Yoshimoto et al., 1993], is deployed to estimate Q P and Q S .Least-squares fitting and the slopes estimations are applied to keep the deviation to minimum.Accordingly, we calculated Q P and Q S using the following relations (6) (7) The optimal fitting lines are plotted in Figure 3 for TZR station.We observe that Q P and Q S exhibit an increase proportionately with in frequency, implying exclusive dependence on frequency (Table 3).At frequency of 1.5 Hz, Q P and Q S are found to be 76 and 171.While the estimates plunge to 1097 and 2936 for Q P and Q S , respectively at 12 Hz.Figure 4 (a and b) illustrate the variation of Q P and Q S at all the receiver stations.There is no correlation with tectonic or geological features for Q P and Q S for frequency 1.5 Hz as indicated by Figure 4. Identical results arise for other frequencies.Apart from this, no reasonable differences appear in Q P and Q S at a given frequency among the stations.As such, the average values of Q P and Q S are computed by taking mean over respective values at all station shown in Figure 5, with standard deviation enlisted in Table 3.It can be observed that smaller standard deviations arise for low frequencies than high frequencies.The standard deviations in Q, as shown in Figure 5 and listed in Table 3, increase when the frequency increases.The standard deviations are physically caused by focusing or defocusing of seismic waves as a result of heterogeneities in the lithosphere.As per Frankel and Wennerberg [1987], the scatters in the amplitudes of direct waves are ascribed to other processes rather than scattering and intrinsic attenuation; in spite of smoothing the radiation pattern by considering a huge sets of earthquake events.In the media, where velocity increases with depth, the geometrical spreading is inversely proportional to the hypocentral distance.This produces substantial loss in the amplitude of direct waves.Besides, the variation in azimuth of earthquake locations is also a possible reason for the scatter effect.
The value of Q S /Q P >1 for the frequency range 1.5-12 Hz is assessed for this study.Similar results were also reported by Rautian et al., [1978] and Yoshimoto et al., [1993].They obtained same results at higher frequency levels.Equivalent estimates are also found for South Eastern Korea [Chung and Sato, 2001] for a frequency range of 1.5 to10 Hz, where the ratio is established to be greater than 1.The obtained value of Q S /Q P in our study regions agrees with the outcomes reported by the other studies as mentioned above.We found that Q S > Q P by a factor of 2.44 in 1.5-12 Hz range, which agrees well with the theoretically expected ratio of 2.41 for whole north-eastern part of India [Sato, 1984].
With a power law of the form Q = Q 0 f n , where n is the frequency-dependence coefficient and Q 0 is the quality factor at f=1Hz, we estimated Q 0 and n (Table 4).The value of n is approximately close to 1 and varies because of heterogeneity of the medium from area to area Aki [1981].The power law fitting yields Q P = (45±3)f (1.3±0.07) and Q S = (101±4)f (1.4±0.05) for the study region.The low estimates can be correlated with the seismically active nature of this fault [Bora et al., 2017b].
We observed that within the entire frequency range, the estimated values of Q S is more than Q P .This implies that for this study area, P waves are attenuated more than that of S waves.As reported by Bianco et al., [1999] and Sato and Fehler [1998], the attained for S − wave from Equations (3) Q S /Q P >1 is also observed in the upper crust of various other areas, having a higher amount of lateral heterogeneity.This has direct relation with disparities pertinent to this region.Blending geological aspect, varied level of disparities can also be taken into account.Van Eck [1988] proposed an interrelation among the level of tectonic activity and the degree of frequency dependence.It has been noticed from the reported literature [Roecker et al., 1982;Rodriguez et al., 1983;Havskov et al., 1989] that seismically active provinces are generally characterized by low Q 0 values and high n values; while stable regions are described by high Q 0 and low n values [Pujades et al., 1991;Pulli, 1984 andRhea, 1984].
In order to validate the results, we attempted correlation of them with the available geophysical confirmations, reported by other researchers for this entire region.It can be gleaned from reports of Bhattacharya et al., [2002Bhattacharya et al., [ , 2008] ] that KF could be cited as the next prominent active zone after Shillong Plateau in the NER-India.They detected a considerable seismic activity along the northwestern-south eastern margin of KF.Additionally, Bhattacharya et al., [2002] had done 1D and 2D mapping of whole NE region.They found high b-value and high fractal dimensions (1.65 < D < 1.85) beside the KF.It is already established by Mogi [1962] that high b-value signifies complex heterogeneities or the existence of rock mass with variegated fractures.The higher D-values beside the KF can also be ascribed to the heterogeneous transverse structure of this fault [Bhattacharya et al., 2002[Bhattacharya et al., , 2008]].

COMPARISON WITH GLOBAL DATA
To build an in-depth understanding, we carried out an investigation to discern a global perspective in the context of quality factors related to P and S waves.To this end, we made a plot in Figure 6(a) and (b), respectively comparing our findings with those estimated for other parts of the globe having diverse tectonic settings.Based on attenuation features, the categorization was executed.The pattern of Q P with respect to frequency is identical (Figure 6  in Q values with increasing frequency indicates the frequency-dependent nature of the Q estimates in the region.It suggests that the attenuation characteristics in the KF zone are well comparable to the other seismically active regions of the world.However, the lower estimates of Q P are primarily accredited to complex degree of crustal heterogeneities.Similarly, in Figure 6(b), it is seen that the pattern of Q S exhibits an identical pattern as found in Kummaun, India.
Figure 7 gives an illustrative valuation of average estimates Q S /Q P ratio of our study area to previous studies of various locations at 1 Hz.The disperse nature of Q S /Q P values is ascribed to distinct regional behavior of the attenuation mechanism [Lees and Lindley, 1994].It is evident that our result falls into the region of Q S >Q P .Lees and Lindley [1994] theoretically showed that Q P equals 2.25 times of Q S (i.e.Q S /Q P =4/9) for entire world crust.Keeping this equivalence of Q P with Q S , a bench mark has been plotted in Figure 7, represented by dashed line.The solid line indicates Q S =Q P .However, the ratios so far, reported worldwide considerably diverge from this theoretical value.Yoshimoto et al., [1993] revealed that Q S /Q P >1 for low frequency (0.1 Hz) seismic waves.For the frequencies in the range of 1Hz to 30 Hz, Yoshimoto et al., [1998]   opined that Q S /Q P >1.This has been also obtained worldwide [Castro and Munguia, 1993;Chung and Sato, 2001;Kinoshita, 2008;Padhy, 2009;Abdel-Fattah, 2009].Recently, Kumar et al., [2014] reported Q S /Q P >1 for the Kinnaur Himalaya region for frequencies ranging from 1.5 to 12 Hz.Therefore, the derived Q S /Q P ratio tallies well with other values reported worldwide.We have also plotted a comparison of Q S /Q P which was computed at diverse frequencies with those conducted at other tectonic regions worldwide in Figure 8.In the frequency range of 1.5 to 12 Hz, our estimate corroborates well with dissimilar regions of the sphere.

CONCLUSIONS
In conclusion, we have carefully delineated the attenuation features of Kopili Region, emphasizing frequency dependent behavior.
The results establish higher values of Q for S waves (Q S = (101±4)f (1.4±0.05) ) than for P waves (Q P =(45±3)f (1.3±0.07) ) in the 1.5-12 Hz frequency band.It is seen that our findings agree with other published results on tectonically active regions that is characterized by high degree of heterogeneity.Hence, it can be useful for further studies in ground motion prediction in Kopili region.The low value of quality factors corresponding to P and S waves refers to seismically active areas with complex heterogeneity.The estimated value of Q S /Q P and their subsequent association with frequency affirm the dominance of scattering as a major contribution to diminution of seismic amplitudes.This study will act as a basic building to grasp attenuation mechanism, prevailing here.It will also help to restrain the frequency dependence of body wave attenuation in the earth's crust, which is important to modify the velocity dispersion due to attenuation.Furthermore, attenuation study of a region is considered as an integral part for seismic hazard assessment, source parameter studies and ground motion simulation process.There are many quality factors reported for whole northeast India, which is a unique combination of different tectonic units.So assigning a single Q for entire northeast India is not justified.The current work indeed finds a quality factor for Kopili fault zone, so the new quality factors will prove beneficial for above mentioned seismological practices.

FIGURE 1 .
FIGURE 1. Map of Kopili Fault region [afterBiswas et al.,  2013a,b;Bora and Biswas, 2019].The red circles symbolize the used events for this study, while triangles represents the temporarily deployed stations.The Great and major earthquakes occurred in and around this region was represented by the blue star symbol.(In inset the India map was shown).

FIGURE 2 .
FIGURE 2. (a) The plot of typical seismogram.The colors represent the corresponding portions of the seismogram.(b) Signal to Noise ratios of corresponding portions for an recorded event at TZR station.

FIGURE 3 .
FIGURE 3. (a) The plot of typical seismogram.Arrows indicates origin time, P-wave and S-wave arrival, while coda waves are indicated by dashed line.(b) Coda normalized P and S wave amplitude entailing vertical and north-south component of seismogram within each frequency band for TZR station.The estimated Q P and Q S values are also shown.
FIGURE 4. Estimates of (a) Q P and (b) Q S at central frequency 1.5 Hz mentioned besides each stations.

FIGURE 5 .
FIGURE 5. Variation of Q P and Q S versus f for Kopili zone with error bars.

FIGURE 6 .
FIGURE 6. Global comparison of estimates of (a) Q P and (b) Q S with different tectonic settings.

FIGURE 7 .
FIGURE 7. Global plot of Q P versus Q S along with the current study (at 1 Hz).The small square representing the estimated value in our study.

FIGURE 8 .
FIGURE 8. Comparison of estimated Q S /Q P ratio as a function of frequencies with those of different tectonic parts of the world.

TABLE 2 .
Values of band-pass filter used in this study.

TABLE 3 .
Average estimates of Q P and Q S along with their SD.

TABLE 4 .
Values of Q0 for P and S along with their n values at all the six stations.