ESTIMATION OF SEISMIC HAZARD USING NON-POISSONIAN MODELS

Abstract

Seismic hazard is defined as the probable level of ground shaking associated with the occurrence of earthquakes. Two basic approaches used for the seismic hazard assessment generally used are "deterministic" and "probabilistic" seismic hazard analysis. The Probabilistic Seismic Hazard Assessment (PSHA) approach integrates the effects of all the earthquakes expected to occur at different locations during a specified life period, with the associated uncertainties and randomness taken into account. The spurt in the seismic activity in the recent times and the damage inflicted by their occurrence has lead to the development of new methodologies for seismic hazard assessment. The seismic hazard has been mostly estimated using statistical or probabilistic methodology where the occurrence of earthquake is treated most of the time with Poissonian model. The Poisson model has an exponential recurrence time distribution and thus a constant hazard function, this leads to time-independent seismic hazard estimates. Poisson process occurs randomly, with ho "memory" of the time, size, or location of any preceding event. This assumptions of Poissonian model namely the occurrence of event does not depend on the occurrence of the past event, i.e., it is memory less, makes this model contradictory to the basic theory of occurrence of earthquakes as given by Reid (1910) which implies that the occurrence of earthquakes will relieve stresses along the portion of a fault on which rupture occurs, and that subsequent rupture will not occur on that segment until the stresses have had time to build up again. The chances of an earthquake occurring on a particular fault segment should therefore be related in some way to the time that has elapsed since the last earthquake and, perhaps, to the amount of energy that was released. Thus individual earthquakes on a particular fault segment should not be considered as random, independent events. Poissonian model is unable to appreciate clear understanding of physical process of earthquake generation. Therefore, the methods recognizing the physical process i.e. Non-Poisssonian methods should be used to estimate seismic hazard. The advantage of the Extreme value statistics method lies in the fact that it uses extreme events of definite time interval (i.e. say yearly) instead of using entire catalogue data for a particular region thus overcoming the uncertainties involved into the seismic hazard assessment due to magnitude of completeness. Such extreme events are often widespread felt and well catalogued and this approach reduces the uncertainty involved in assessing the seismicity parameters based on small or moderate magnitude events occurred in pre-instrumental era. By taking into account this advantage of extreme value statistics, in this re,port=ve=pr-esent~the=seism-ic-hazard-assessment based on Gumbel's maximum earthquake magnitude (Gl, G2, G3, and 012, G13, G23) models using graphical method and maximum likelihood method for parameter estimation. Application of this methods to Northern India region considering centre at Delhi.(Latitude 24° — 31.5 ° N and Longitude 74 ° — 81.5°E) shows that Gumble's G13 model is best fitted one for study area. Estimation return period and probability of exceedance has also been carried out using Poissonian distribution along with G-R relationship for the study area.__- --- The objective of the present work is not only to assess seismic hazard but also to evaluate —__A_ _. the best approach for seismic hazard assessment. Thus the hazard assessment has been carried out by Poissonian Distribution and Extreme value statistics and the results which are presented in return period and probability of occurrence of various magnitude ranges are compared". Far difference in the values of return period shows that method of estimation of hazard plays an important role in hazard assessment. Range-of mean values and the higher standard deviations of return periods reveal the need of logic tree approach which is also presented in this present work