PROBABILISTIC SEISMIC HAZARD ASSESSMENT FOR NORTHEAST INDIA REGION

Abstract

An attempt has been made in this study to estimate the strong ground motion using Probabilistic Seismic Hazard Assessment (PSHA). PSHA provides a framework in which uncertainties at different steps of seismic hazard assessment can be identified, quantified and combined in a rational manner to yield a more complete picture of the seismic hazard with respect to earthquake catalog incompleteness, earthquake source modelling, methods of estimation, earthquake recurrence models and types of attenuation relationships. PSHA methodology has been applied to Northeast India region as a case study. The study area lies between latitude 200— 300 N and longitude 870 - 980 E for which the geological and tectonic setup was studied and an improved earthquake catalog has been prepared by using regional and global regression relations, wherever required. The earthquake data including historical as well as instrumental records from various agencies, published and unpublished reports has been compiled in the study. The quality, consistency and homogeneity of the earthquake data has been assessed in systematic maimer to find out the uncertainties at each step of calculations. The dependent events have been removed in order to consider only the independent events in the catalog. A new methodology has been developed to deal with the magnitude conversion problem for the application of General Orthogonal Regression (GOR) towards conversion of different magnitude types. Through minimization of the squares of the orthogonal residuals, GOR relation is first obtained in terms of the abscissas of the projected points corresponding to the observed data pairs. This GOR relation is then modified to allow direct substitution of observed magnitude values using a linear relation derived between the projected points and the given data points. Commonly used GOR forms, however, express these relations using observed magnitude instead of abscissa of the points on the GOR line. Based on events data for the whole globe during the period 1976-2007, GOR relations have been derived for conversion of mb to M, mb to M, mb to M and M to M,, following procedure formulated in this study. The superiority of the GOR relations obtained using the proposed procedure over the commonly used GOR forms has been shown by computing the absolute average difference and standard deviation between the observed and the estimated values using independent events data not used in the derivation. This procedure has been further validated for a wide range of error variance ratio () values between 0.1 and 7.0 using global and regional data sets. Four regional datasets (Japan region, Mexico region, Peninsular India region, and Northeast India and adjoining region) are used in this regard. The comparative standard deviations of the errors associated with different magnitude types have been estimated to be 0.09, 0.12 and 0.2 for M M and mb, respectively. The values obtained in this study are of similar order as estimated by Thingbaijam et al. (2008) and Kagan (2003). Standard regression, usual GOR and new GOR relations have been obtained using the Northeast India datasets for magnitude ranges 4.6 5 Mb < 6.6 (ISC), and 4.6 < mb < 6.8 (NEIC), and global data sets for magnitude ranges 2.9 < mb < 6.5 (ISC), and 3.8 < mb < 6.5 (NEIC). It is observed that mb,Isc and mb.NEJC estimates are generally lower compared to M%V values as has been observed also by some other investigators. Further, the GOR relations also have been obtained for regional datasets for conversion of body wave magnitude to momen magnitude using 1366, 462, 147 and 1072 events from the Japan region, Mexico regic Peninsular India region, and Northeast India and adjoining region, respectively. Sin' relations have also been obtained between Ms and M in the magnitude ranges 3.0 < M 6.1 and 6.2 < M 8.4, for global data set and 3.0 < M 6.1 for regional datasets. For deriving an empirical relationship between intensity (Imax) and moment magnitude M,NE1c data corresponding to MMI intensity VI and higher has been used. In all, 126 intensity and M,NE1c data pairs for the whole globe are used in deriving this relationship. The empirical relationship derived in this study is useful for conversion of intensity to M for historical earthquakes for which magnitude information is not available. Three different catalogs of earthquake events have been prepared interms of moment magnitude using standard regression, usual GOR and new GOR methods for magnitude conversion; the catalogs are referred to as Cat 11, Cat 2 and Cat 3, respectively. Catalog 3 has so been further used for seismic hazard estimations. The further treatment for declustering and completeness has been carried out on Cat 3 only. The entire Northeast India is subdivided into nine seismogenic zones. This subdivision is based on tectonic and geological features, focal mechanism solutions (Angelier and Baruah, 2009) and spatial distribution of earthquake events. The magnitudes of completeness, which has direct bearing on estimation of Gutenberg Richter parameters 'a' and 'b' values have been derived for the nine seismogenic zones using Stepp's, method (Stepp, 1972). The seismic hazard parameters namely activity rate )c, seismicity parameter 0 and maximum magnitude Mn .. have been derived for seismic hazard assessment. The ground motion prediction equations (GMPE) developed by Boore and Atkinson (2008) and Gupta (2010) have been used to estimate the strong ground motion using CRISIS (Ordaz et al., 2007). IMA PSHA for Northeast India region has been carried out using the classical Cornell- McGuire approach. Seismic hazard has been computed by performing hazard computations at a - grid interval of 0.10 x 0.10 covering the entire Northeast India region. The PSHA results refer to equal weightage obtained by combining the branches of the logic tree applied in the present study. Contour maps have been prepared for mean values of PGA and spectral accelerations at 0.2 and 1.0 sec spectral periods for 50%. 20%, 10%, 2% and 0.5% probabilities of exceedance in 50 years. The corresponding COV maps represent overall variability in seismic hazard for PGA and spectral accelerations (Sa) at 0.2 sec and 1.0 sec for 100, 225, 475, 2475 and 10,000 years of return periods. Estimated mean strong ground motion and the COV values are calculated for 50%, 20%, 10%, 2% and 0.5% exceedance in 50 years. The mean PGA for 50% exceedance in 50 years varies between 0.01g to 0.19g with COV varying between 0.08 to 0.246. The spectral acceleration at 0.2 sec varies between 0.Olg to 0.4g with COV varying between 0.07 to 0.214. The spectral acceleration at 1.0 sec varies between 0.0lg to Dig with COV varying between 0.09 to 0.3. The mean PGA for 20% exceedance in 50 years varies between 0.01g to 0.251g with COV varying between 0.08 to 0.222. The spectral acceleration at 0.2 sec varies between 0.olg to 0.518g with COV varying between 0.07 to 0.195. The spectral acceleration at 1.0 sec varies between 0.01g to 0.143g with COV varying between 0.07 to 0.27. The mean PGA for return period of 475 years varies between 0.0lg to 0.32g with COV varying between 0.07 to 0.206. The spectral acceleration at 0.2 sec varies between 0.02g to 0.656g with COV varying between 0.07 to 0.177. The spectral acceleration at 1.0 sec varies between 0.02g to 0.178g with COV varying between 0.08 to 0.244. The mean PGA for return period of 2475 years varies between 0.02g to 0.5g with COV varying between 0,065 to 0.168. The spectral acceleration at 0.2 sec varies between 0.lg to 1.lg with COV varying between 0.06 to 0.15. The spectral acceleration at 1.0 sec varies between 0.04g to 0.29g with COV varying between 0.07 to 0.206. The mean PGA for return period of 10,000 years varies between 0.02g to 0.68g with COV varying between 0.06 to 0.15. The spectral acceleration at 0.2 sec varies between 0.05g to 1.48g with COV varying between 0.06 to 0.127. The spectral acceleration value at 1.0 sec varies between 0.16g to 0.42g with COV varying between 0.068 to 0.178 The design spectra for various return periods have been prepared for 10 important cities of Northeast India region which can be directly used by considering the local site conditions.