STUDY OF EARTHQUAKE SOURCE PARAMETERS USING MICROEARTHQUAKE AND STRONG MOTION DATA

Abstract

The parameters characterizing the source of an earthquake are broadly classified into kinematic parameters and the dynamic parameters (e.g., Duda, 1978). Several techniques have been developed to estimate these parameters; in the time domain (e.g., Frankel and Kanamori, 1983) and in the frequency domain (e.g., Brune, 1970). Brune (1970) and Aki (1967) laid a strong foundation for estimating source parameters from the spectra of seismic waves and developing scaling law. Earthquake source parameters provide a great deal of information about the properties of earthquake sources, and have a large number of applications in Seismology and Earthquake Engineering. Some of the important applications include: studying spatial and temporal variation of tectonic stress from the stress drops of earthquakes, developing scaling laws, understanding earthquake rupture processes, quantifying the excitation of high frequency strong ground motion produced by large earthquakes, and discriminating nuclear explosions from earthquakes. The present study is based on observational data and is devoted to the estimation and interpretation of source parameters of microearthquakes and small and moderate earthquakes for the two regions in the Himalaya, namely the Kameng region located in the Arunachal Lesser Himalaya and the region covering parts of the Garhwal and Kumaon Himalaya. This study has been carried out with the following objectives: • To investigate the attributes of source parameters of local earthquakes. • To develop scaling laws. • To estimate/nax and infer its relationship with source size and site effects. ii For the purpose of carrying out the study, the digital data of local earthquakes collected employing seismological and strong motion networks have been used. For the Kameng region, a data set of 79 local earthquakes collected using three digital seismographs operated in the region from December 2002 to May 2003 have been analyzed and interpreted. For the Garhwal and Kumaon Himalaya three different data sets have been used. These data sets are: i) strong motion data of the two moderate earthquakes - the Uttarkashi earthquake and the Chamoli earthquake, ii) strong motion data of 14 local events recorded on strong motion network deployed in the region under Mission Mode Project, and iii) digital data of same 14 local events collected from 12-station seismological network deployed in the Garhwal Himalaya around Tehri region. Hypocenter parameters of 79 local events, collected from the Kameng region, have been estimated employing HYP071 program (Lee and Lahr, 1975). The hypocenter parameters of 14 local events occurred in the Garhwal and Kumaon Himalaya have been estimated using HYPOCENTER program (Lienert et al., 1986) given in SEISAN software (Havskov and Ottemoller, 2005). The hypocenter parameters of the Uttarkashi earthquake and the Chamoli earthquake are taken as reported by India Meteorological Department and USGS. Three component time histories of recorded events were corrected for baseline and instrument response and rotated along the direction of ray path to permit separation of SH and SV components of the ground motion. SH-components of the time series were smoothened by a tapered cosine window function, and velocity/acceleration spectra have been computed employing FFT. A frequency dependent path correction has been applied to the observed spectra to estimate source spectra. Brune's model along with a high frequency diminution factor represented by a Butterworth high-cut filter to allow estimation off^ in (Boore, 1983) have been adopted and fitted to acceleration and displacement source spectra. The spectral parameters, viz., Clo,fc,fmwc and N, estimated from fitted model have been used to compute source parameters namely, seismic moments, source radii, stress drops, displacements along the fault, and dimensions of the cohesive zones. For the Kameng region, from the source parameters of 79 local events (0.7<Mw< 3.7), it has been brought out that 20 events (2.2<Mw<3.7) depict a linear trend between corner frequency (fc) and seismic moment (Mo) and follow the scaling law, Mo (dyne-cm) = 2.0x1022/c"334. Fifty nine events exhibit low stress drops less than 1 bar and majority of these occur at shallow depths less than 10 km, and twenty events have stress drops between 1 bar and 21 bars. The stress drops increase with magnitudes for very small events (0.7<Mw<2.0) and becomes almost constant for events above magnitude 2.0. This shows a failure of selfsimilarity for events below magnitude around 2.0 for Kameng region. The estimated values of/nax by and large conform to the worldwide observations and show a weak dependence on source size. The average values offmax at Kimi, Bichom and Tenga sites are about 25 Hz, 30 Hz and 20 Hz respectively. A relation between seismic moment and duration magnitude Log Mo {dyne —cm) = 2.173 MD + 17.6, has been developed for the region. One of the conspicuous features is the occurrence of low stress drop events in Kameng region. Such events have been observed from many other regions of the world and found to occur in the same areas where high stress drop events occur. To explain the occurrence of low stress drop events: partial stress drop model (Brune, 1970) and low effective stress model (Brune et al., 1986) have been postulated. According to partial stress drop model these events occur either due to complex fault geometry or due to asperities or barriers on the fault and the fault locks soon after the rupture passes. This inhibits the average slip over the fault iv to reach the average dynamic stress drop over the whole fault. The low effective stress model postulates that low stress drop events occur because low effective stress is available to accelerate the fault. For the Garhwal and Kumaon Himalaya, source parameters of small and moderate sized earthquakes (3.1<Mw<6.7) including the Uttarkashi earthquake and the Chamoli earthquake have been estimated. For the Uttarkashi earthquake the estimated values of seismic moment, source radius, stress drop and displacement along the fault are: (l.l±0.22)xl026 dyne-cm, 9.7±0.2 km, 52.6±5.9 bars and 1.19±0.13 m respectively. The /max values at thirteen sites vary from 4.0 Hz to 12.5 Hz with average value of 7 Hz that corresponds to about 200 m dimension of the cohesive zone. The stress drop of the order of 60 bars has been estimated for the Uttarkashi earthquake by Sriram and Khatri (1997). For the Chamoli earthquake, the estimated values of seismic moment, source radius, stress drop and displacement along the fault are: (4.7±0.72)xl025 dyne-cm, 7.3±0.11 km, 53.2±6.96 bars, 0.90±0.13 mrespectively. The/max values vary from 2.2 Hz to 10 Hz with average value around 5 Hz. The dimension of the cohesive zone estimated from the average value of/max is about 300 m. Seismic moments of 14 local events (3.1<Mw<4.7) range from 4.0xl020 dynecm to 1.06x1025 dyne-cm and their stress drops vary from 25.9 bars to 83.4 bars. The average values of/max for these sites vary from 6.8 Hz to 13.3 Hz and dimensions of the cohesive zones estimated using these values of/max ranges from 192.5m to 376.5 m. Based on the data set of 16 events (3.1< Mw<6.7), the stress drop for this region is 59.1±19.2 bars, which by and large agrees with the earlier reported stress drop of 56±36 bars for this region (Kumar et al., 2008). Using the data set of 16 events, a scaling law, M0 (dyne-cm) = 3.0x1023/c"298 has been developed for the region. The values of/max have poor dependence on the source size and vary from 4 Hz to 18 Hz at various sites falling in the Garhwal and Kumaon Himalaya. In general the sites located on Phyllites show large scatter in the observed values of/max as compared to those located on Sandstones. A site located on Quartzites has least scatter in observed/max. From the study of stress drops of earthquakes using global data, the average stress drops of 30 bars for interplate, 100 bars for intra-plate, and 60 bars for average earthquakes have been estimated (Kanamori and Anderson, 1975). Comparing the average stress drop of 60 bars obtained in the present study, with the global average of 30 bars for inter-plate earthquakes, it can be inferred that the average stress drop associated with the small and moderate earthquakes is almost double in the Garhwal and Kumaon Himalaya. The study has enhanced the understanding about the characteristics of source parameters estimated from local earthquake data. Both Kameng region and Garhwal and Kumaon region fall in the Lesser Himalaya between the Main Boundary Thrust and the Main Central Thrust but have brought out different characteristics of source parameters. For Kameng region, the breakdown of self-similarity has been observed for events below magnitude 2.0. For the Garhwal and Kumaon region the stress drops of events (3.1<Mw<6.7) are almost constant and independent of magnitudes, thus showing the self-similar behavior of earthquakes. Breakdown of self-similarity has been the subject of many studies (e.g., Prieto et al., 2004). However, some studies have shown that breakdown of self-similarity may occur due to the limited frequency band of the data and an artifact of high near-surface attenuation (e.g., Boore, 1986; Abercrombie, 1995; Hiramatsu et al., 2002). For both regions the high frequency spectral roll-off above fmax is represented by the Butterworth filter of order 6 (jV=6), which is almost equivalent to go3 roll-off above /max in VI source spectra. The/max at various sites have brought out site-dependent behavior. However, a slight decrease offmax with increasing seismic moment shows its weak dependence on the source size. From the stand point of earthquake engineering applications, this study has provided information about two important parameters, namely stress drop and/max. Both the parameters form important ingredients for the estimation of strong ground motion using stochastic source model.