ESTIMATION OF EARTHQUAKE HAZARD FOR CENTRAL AND EASTERN ASIA, HIMALAYA REGION AND PENINSULAR INDIA

Abstract

Earthquake hazards seriously threaten life and property. Evaluation of seismic hazard and development of means of mapping them, are among the research problems in seismology, that urgently require solutions. The existence of accurate and extensive earthquake catalogues encourages the use of earthquake statistics in fitting mathematical models to the patterns of earthquake occurrence and ground motion data. The probabilistic approach allows the incorporation of uncertainty and frequency analysis of earthquake occurrence. It can accurately reflect the true state of knowledge and lack thereof. The disadvantages of probabilistic seismic hazard analysis is the loss oftransparency, which stems from inclusion / integration of voluminous data, theory and judgement. In this work, we compile four different kind of rigorous probabilistic analyses, with an aim, to estimate various hazard parameters for Central and Eastern Asia (within latitudes 0° - 60°N and longitudes 65° - 125° E) in a general manner, and for Himalaya and Peninsular India in detail. Area undertaken covers great central mountain system surrounding and crossing high table lands of Pamir Plateau and Tibetan Plateau. This mountain system encompasses Himalaya, Kunlun, Tien shau, Hindukush andKirthar-Sulaiman mountain ranges. For sake of our study whole area is divided into eight major division, namely: Region A(Mongolia, Baikal and surrounding areas), Region B (Tien Shan). Region C (Tibetan plateau), Region D(Eastern Tibet, Gausu and Ningxia). Region E (Xianshuilie Fault and Western Burma), Region Gand source zones H, ,H21 (Indian Subcontinent) and Region I (North China). Indian Subcontinent, being our major concern is divided into 22 source zones. First exercise consists inpreparing probabilistic hazard maps in terms of peak ground motion (acceleration, velocity, displacement) and peak magnitude over a projected period. These maps are continuous colour depiction of varioushazard parameters. Extreme value theorems of I and III kind are made use of during this course for ground motion and magnitude data respectively. A high degree non-linear fitting to real data for annual probability of nonexceedance and annual extreme magnitude, is achieved in this analysis. Complete error matrix is obtained which further provides an insight into the problem. li Results are discussed for various major tectonic features and numerous locales within the area. Despite being locales of low seismic hazard potential, peak ground motion for some, is overestimated, because of fictious hypocentral depth (~ 0 km) assignment to some events in the catalogues. Overestimation in peak magnitude for some low potential locales ( in the proximity of high hazard zone) are a result of interpolation. This problem is severe for magnitude and graudally fades from displacement to velocity and accelesation as decay rate ofthese parameters with respect to hypocentral distance ascends alongwith this order. Earthquake history spanning over ninety one years (1900-1991) is taken up. Second exercise, adds another dimension to these exercises of hazard estimation through exploiting physical links between crustal deformation, seismic moment and seismic hazard. In this, moment release rates and average slip rates for nine seismotectonic units (slightly modified from 8 division classification) as discussed earlier. This modification involves picking up of two tectonic units; Region G (Central and Western Burma) and source zones Hn and HJ2 which jointly form Himalaya (Region H) circumvented within Indian Subcontinent, in addition to Regions A, B, C, D, E, F and I. During this exercise, seismic moment release rates are computed through the application of extreme value theory and average slip rates by clubbing fault parameters with those of moment release rates. These moment release rates are compared with observations. This agreement can be used to justify the extrapolation of frequency-magnitude statistics beyond the historical and instrumental era in seismic hazard studies as a test of the stationarity of short term statistics against long term effects. Events occurring within 1900 to 1980 are considered here. Next two techniques operate on a data set, having surface wave magnitudes >7.0 and encompassing 19lh and 20lh century for Himalaya and adjoining areas. In the subsequent analysis, estimates of upperbound magniutdes and corresponding waiting times, relationships among annual mode magnitudes, magniutdes equivalent of annual average rates of energy release and upperbound magnitudes, seismic activity and heterogeneity parameters and uncertainties in associated with each and every of them for different seismotectonic provinces of Himalaya and adjoining areas, namely; North Western Himalaya (NWH), Central Himalaya (CH), Eastern Himalaya (EH), and Burma (BRM). This involves in merging energy-magnitude and linear frequency-magnitude relations and furnishes upperbound magnitude compatible with finite strain energy release rate. Above mentioned relationships among various kind of magnitudes are very important from the veiw point of seismic hazard analysis. Upperbound magnitudes are evaluated using both analytical and graphical methods and they match well. Further, we tackled the problems of mean return periods and ultimate displacements to ruptures for great earthquakes in different seismotectonic provinces of Himalaya and adjoining areas through using Weibull theory. Reliability, failure density and cumulative probability curves (in both, time and displacement domains) are plotted which provide useful insight into the problems of recurrence of great earthquakes. Relative plate velocities are incorporated to convert time intervals between the events to displacement intervals. Also conditional probabilities are estimated which further provides insight into the earthquake occurrence. A thought can be given to seismic gaps from the view point of conditional probabilities. As seismic gaps are zones which, though ruptured in the past but remained seismically quiescent in last few decades, are the probable locales of high potential for rupture. The probability of this happening in the coming years can be best simulated using conditional probabilities. Another feature of the Weibull distribution analysis is that, its cumulative probability serves itself as a measure which governs the commencement of intensification of prediction oriented observation. In this work we also suggest that epochs when intensification of prediction oriented observations (i.e. monitoring of precursory phenomena) should start for North Western. Central, Eastern Himalaya and Burma. In a nutsnell, continuous probabilistic hazard maps in terms of peak magnitude, acceleration, velocity and displacement and computation of various other hazard parameters offer a systematic, picture of estimated seismic hazard in Central and Eastern Asia, Himalaya and Peninsular India