SEISMIC WAVE ATTENUATION CHARACTERISTICS OF THREE INDIAN REGIONS

Abstract

Study of seismic wave attenuation in high frequency range (1 to 30 Hz) has emerged as one of the promising field for indepth understanding of the seismotectonic characteristics in a region. The knowledge of seismic wave attenuation has wide ranging applications and is primarily required for the estimation of earthquake source parameters and prediction of strong ground motion at a site due to probable earthquakes in the region. The temporal and spatial variation of seismic wave attenuation has also been studied as an earthquake precursor. Seismic wave attenuation is described by the nondimensional parameter, Q, called the quality factor, and is measured from the decay of seismic wave amplitudes with time. The attenuation characteristics have been studied for a number of seismic regions of the world by various investigators (e.g., Aki and Chouet, 1975; Steensma and Biswas, 1988; Dominguez et al, 1997; Martynov et al., 1999). Computation of attenuation has been carried out using Pwaves (Qa), S-waves (Qp), coda waves (Qc) and surface waves (Qsur). In India, very few observational studies have been conducted for the estimation of seismic wave attenuation (e.g., Gupta and Rambabu, 1994; Mandal and Rastogi, 1998). Attempts have also been made to study theoretical aspects of seismic wave attenuation by Tripathi and Ram (1997). This study is an attempt to estimate seismic wave attenuation for three seismic regions of India namely, the Garhwal-Himalaya, the Northeast India and the Koyna region. These regions are located in different geotectonic environments. The study has been carried out with the following objectives : 1. To study the frequency dependent attenuation characteristics of the Garhwal Himalaya from the analysis of coda waves of digitally recorded local earthquakes as well as S-waves and coda waves observed on strong motion records. 2. To study the frequency dependent attenuation characteristics of the Northeast India from the analysis of coda waves observed on strong motion records. 3. To study the frequency dependent attenuation characteristics of the Koyna region from the analysis of coda waves of digitally recorded local earthquakes. 4. To investigate the lapse-time dependence of Qc for the Garhwal Himalaya and the Koyna region with a view to study the variation of Qc with depth. 5. To compare the attenuation characteristics for three regions and to evaluate their seismotectonic implications. Following data sets available from three regions have been analysed to carryout the study : The Garhwal Himalaya: Two data sets have been analysed to study the attenuation characteristics of the region. The first data set consists of seven digitally recorded local earthquakes occurring within 100 km of epicentral distances. This data set was collected during March 1989 to December 1990. The coda waves of thirteen seismograms of these local earthquakes have been analysed to estimate Qc in the region. The second data set comprises of strong motion records obtained at thirteen strong motion stations during the Uttarkashi earthquake of October 20, 1991 (nib=6.5). The coda waves and S-waves observed on these strong motion records have been analysed to estimate Qc and Qp in the region. in The Northeast India : The data set comprises of strong motion records of August 6, 1988 earthquake (mb=6.8) occurred in the Northeast India. Analysis of coda waves observed on thirty three strong motion records has been carried out to estimate Qc in the region. The Koyna region : The data set consists of thirteen digitally recorded local earthquakes occurred in the region within 60 km epicentral distance. The data set was acquired during July-August, 1996. The coda waves of 76 seismograms of these earthquakes have been analysed to estimate Qc in the region. To estimate Qc, the coda waves observed on seismogram of local earthquakes and strong motion records have been analysed using the single backscattering model developed by Aki and Chouet (1975). This model works under assumptions e.g., the coda waves are backscattered S-waves, the scattering is a weak processes and the source and receiver are coincident. To estimate Qp, the Swaves observed on strong motion records have been analysed using the single station ratio method. This method, given by Aki (1980a), makes use of decay of maximum amplitude of S-waves, normalised by the coda waves observed at the single station, with the hypocentral distances. Seismic wave attenuation characteristics of three Indian regions, namely, the Garhwal Himalaya, the Northeast India and the Koyna, have been studied. For the Garhwal Himalaya, Qc and Qp estimates have been obtained using the coda waves of local earthquakes and S-waves and coda waves observed on strong motion records. For the Northeast India, Qc estimates have been obtained using the coda waves observed on strong motion records. For the Koyna region, Qc estimates, have been made using the coda waves of local earthquakes. For the Garhwal Himalaya, Qc estimates vary from 104± 16.09 at 1.0 Hz to 2168±95.02 at 18.0 Hz (using the local earthquake data) and 94+11.71 at 1.0 Hz iv to 1443+113.74 at 16.0 Hz (using strong motion data) whereas, the value of Qp varies from 124 at 1.5 Hz to 1571 at 16.0 Hz. This shows that both Qc and Qp are functions of frequency and follow frequency dependent relationships as 0C=(11O+5.15)/(1O2±OO25) (local earthquakes data), qc =(i09±10.95)/(0"±0057) (strong motion data) and Qp =(90±6.44)/(106±0039) (S-waves of strong motion records). One of the significant outcome of the attenuation study in the region is near similarity of attenuation characteristics determined using local earthquakes data and strong motion data. The near agreement of attenuation characteristics obtained from coda waves and S-waves in the region has also brought out that the origin of coda waves can be attributed to backscattered S-waves. Study of lapse time dependence of Qc, using three window lengths of 20, 30 and 40 sec duration, showed that Qc is a function of window length and increases as window length increases. Frequency dependent Qc relationships for these three window lengths are as Qc =(66±3.67)/(U8±a03) (20sec), Qc =(110 +5.15)/(102±0025) (30sec) and Qc =(149±5.90)/(a95±a021)(40 sec). For the Northeast India, it has been found that Qc estimates from strong motion data are function of frequency and their values vary from 122+10.52 at 1.5 Hz to 1356±74.36 at 16.0 Hz. Frequency dependent Qc relationship, Qc =(86±4.04)/(1-02±0026), provides the attenuation characteristics of the region covering parts of the Shillong massif and the area lying between south of the Mikir hills and north of the Halflong-Disang thrust. For the Koyna region, Qc estimates, using 30 sec coda window length, vary from 148±13.5 at 1.5 Hz to 2703± 38.8 at 24.0 Hz. Lateral variation in Qc estimates from north to northwest-southeast has been observed and it seems that the southeastern part of region is more heterogeneous as compared to the northern part. Lapse time dependent Qc study using five window lengths of 20, 30, 40, 50 and 60 sec duration showed that Qc is a function of lapse time window length and increases as window length increases. The frequency dependent Qc relationships obtained for these window lengths are, Qc =(66±7.06)/(U6±0051) (20 sec), ec=(96±7.18)/(,09±0036) (30 sec), ec=(131±13.42)/(I04±0049) (40 sec), Qc =(148±1857)/(104±0060) (50 sec) and Qc =(182±26.75)/(1-02±a069) (60 sec). Jin and Aki (1986) have interpreted Qc as a tectonic parameter and regions of high tectonic activity are characterised by low Qc values. A strong correlation between, n, the power of the frequency dependence of Qc and the level of tectonic activity in the study region has been observed by several investigators (e.g., Aki, 1980b; Akinci et al., 1994). In general, the region having high n value is characterised by higher tectonic activity as compared to tectonically stable region which show low n value. A comparative study of attenuation characteristics for three regions, having different geologic and tectonic environments, indicates that the Northeast region of India is having relatively higher tectonic activity compared to the Garhwal Himalaya and the Koyna region. Whereas, the level of tectonic activity of the Garhwal Himalaya and the Koyna region seems to be by and large similar in nature. Decay of Qc~^ estimates with frequency follow almost similar trend for all the three regions investigated. This observed trends are also by and large, in agreement with that obtained for other tectonic regions of the world. For the Garhwal Himalaya and the Koyna region, increase in Qc with the lapse time is attributed to increase in Qc with depth as longer the time window the larger will be the sampled area of the earth's crust and mantle. This observation seems to indicate that there is a decrease in the level of heterogeneities with depth in the Garhwal Himalaya and the Koyna region.