SOURCE AND PATH CHARACTERISTICS AROUND BILASPUR REGION OF THE HIMACHAL HIMALAYA

Abstract

Observed ground motion due to an earthquake represents the combined effects of three factors: the earthquake source, the path through which the seismic waves propagate, and the characteristics of the local geological structure beneath the recording site. Precise decomposition of the ground motion representing these factors, and their correct interpretation allows studying the attributes of earthquake sources, the media characteristics, and site effects. In the present study, source and medium characteristics around Bilaspur region of the Himachal Himalaya have been estimated from local earthquake data. The study includes development of frequency dependent attenuation relations in the high frequency range (1- 24 Hz) and development of scaling law for the region, and comparison of attenuation characteristics of Bilaspur the region with other Himalayan regions. The study area is located in the Himachal Himalaya which constitutes the northwestern part of the Himalaya. Several tectonic features mapped in the study area include the Main Boundary thrust (MBT), the Main Central thrust (MCT), the Drang thrust, and the Jawalamukhi thrust. Prominent earthquakes reported to have occurred in the region include: the 1905 Kangra earthquake (M= 8.0), the 1945 Chamba earthquake (M= 6.5), the 1975 Kinnaur earthquake (M= 6.2), and the 1986 Dharmsala earthquake (M = 5.5). The study is based on two data-sets of local earthquakes. The first data-set consist of 94 events collected from May 2008 to April 2011 using one digital and four analog seismographs. The second data-set consists of 41 local events, recorded on a 5-station digital network deployed in the area since May 2013. These events (0.5<ML ≤2.9) were recorded by the network from May 2013 to March 2014. Majority of the local events used in the study occur between the MCT and the MBT. However, few events occurred in the Himachal Sub Himalaya and Ganga foredeep. Coda Q has been estimated using single backscattering model. Qα and Qβ are estimated adopting the extended coda normalization method and Brune model is used to estimate earthquake source parameters. Hypocenter parameters of the events were estimated using HYPOCENTER program. The standard errors in the estimation of hypocenter parameters are ≤ 0.50 s in origin time (RMS), ≤ 5.0 km in epicenter (ERH), and ≤ 5.0 km in focal depth (ERZ). Coda Q estimates were obtained from the two digital data-sets from vertical components of ground motion only. A MATLAB code was developed to estimate Qc. 458 Qc estimates with correlation coefficient > 0.7 ii and signal to noise ratio (SNR) > 3 were obtained from the first data set of 94 local events (0.15<ML<3.03; epicentral distance < 100 km; focal depths 1-41 km). From the data-set of 41 local events (0.5<ML≤2.9) coda-Q (Qc) has been estimated for three LTWs of 20, 30 and 40 sec durations, respectively and for three different distance ranges (R) (R<30 km; 30≤R<60 km; and R≥60 km). Applying the coda normalization method to the second data-set, the quality factors for Pwave (Qα) and S-wave (Qβ), in the frequency range 1.5 Hz to 24 Hz have been estimated from the vertical and horizontal components of ground motion. Two functions are used to model the decay of spectrum at high frequencies. The first function represents a high-cut filter to estimate fmax, and slope of the spectrum above fmax and the second function makes use of a high frequency attenuation parameter Kappa (κ). Using the software EQK_SRC_PARA, the Spectral parameters for each event have been estimated, which include: low frequency spectral level (Ω0), corner frequency (fc), the high-cut frequency (fmax), the decay rate (p) above fmax, and the spectral decay parameter ‘κ’. From these spectral parameters, the source parameters were computed. Average estimates of source parameters, fmax, and κ are obtained and interpreted. From the 458 Qc estimates, the frequency dependent attenuation relation Qc =105f1.14 has been obtained for the region for 30 sec LTW. The variation of coda Q for three distance ranges: nearrange (R<30 km), medium-range (30≤R<60 km) and distant -range (>60 km) showed that the Q0 and n estimates vary from 103 to 119 and 1.10 to 1.16, indicating slight increase of Q0 with distance. This means Qc is depth dependent and level of heterogeneity decreases with depth. Comparison showed that at low frequencies up to 10 Hz, the attenuation characteristics of Bilaspur region are almost similar to the south central Alaska, Kachchh and Garhwal Himalaya. From the second data-set, the attenuation relation Qc (f) = (70.3±20.27) f (1.23±0.05) has been developed for the region for 30 sec LTW. Attenuation relations are also developed for three LTW of 20, 30 and 40 sec durations, and for three different distance-ranges (R<30 km; 30≤R<60 km and R≥60 km). Qc estimates are high at higher frequencies showing that at depth the region is more homogeneous. The Qc is found to increase with increasing frequency and LTW length. The high value of n indicates that the region is seismically active. It is found that the attenuation characteristics of the Bilaspur region are almost similar to the Amazon Craton region of Brazil because of some geological similarity. Qc estimates of Bilaspur region are also found to be almost similar to the Kachchh region. iii From the second data-set, the frequency-dependent relations for Qα and Qβ in the frequency range from 1.5 to 24 Hz are: Qα = (43±4) f1.30±0.04 and Qβ = (79 ±6) f1.25±0.02 respectively. The average estimates of Qα and Qβ are found to vary from 71 and 125 at 1.5 Hz to 2901 and 4243 at 24 Hz, respectively.. Comparison of these estimates with the similar estimates for the Himalaya region showed that the Qα estimates at 1 Hz vary between 22 (for the Kumaon Himalaya) and 97 (for the northwest Himalaya), whereas Qβ estimates range between 63 (for the Garhwal Himalaya) and 127 (for the northwest Himalaya). For the Bilaspur region, Qβ/Qα ratio is greater than unity and varies between 1.84 and 1.45 in the frequency range from 1 to 24 Hz. It is found that Qα and Qβ for the Bilaspur region, in the high frequency range from 6 to 12 Hz and 6 to 18 Hz are almost similar to Sikkim Himalaya. Qα , Qβ and Qc are found to increase with frequency but the Qβ estimates are more closer to Qc estimates. This is consistent with other findings that the coda waves are backscattered S-waves. For 41 local events seismic moments, source radii and the stress drops vary from 4.9×1019 dyn-cm to 7×1021 dyn-cm; 187 to 518 m, and less than 1 bar to 51 bars. Stress drops of 11 events with M0 between 1×1021and 7×1021 dyne-cm, vary from 11 bars to 51 bars with average around 22 bars. The variation of the maximum stress drops with depth seems to indicate that the strength of the upper crust decreases below 20 km. A scaling law Mo = 2×1022fc -3.03 has been developed for the region. It is found that fc is source dependent whereas fmax is not related to the source. fmax values at four sites vary from 14-23, 11-19, 9-23 and 4-11 Hz, respectively. κ varies from 0.01-0.035 sec with an average value of 0.02 sec. This range of variation is large compared to the range of 0.023- 0.07 sec observed for the Garhwal and Kumaon Himalaya. For various parts of the world the κ varies over a broad range from 0.003 to 0.08 secs. The estimates of κ are found to be consistent with other regions of the world. The study has greatly enhanced knowledge about the path and source characteristics around Bilaspur region of the Himachal Himalaya. It is found that the Bilaspur region is the most attenuating and highly heterogeneous in character among all the Himalayan regions. The scaling law for the Bilaspur region almost agrees with the other regions of the Himalaya. The strength of the crust around Bilaspur region increases with depth up to about 14 km and then seems to decrease at deeper levels. A temporal change in Qc is observed around Bilaspur region which needs further investigations because such changes in the Qc have been interpreted as a potential earthquake precursor.