SOURCE PARAMETERS OF UTTARKASHl EARTHQUAKE OF OCTOBER 19, 1991 BY WAVEFORM MODELLING

Abstract

A detailed analysis of strong motion data recorded during an earthquake makes it possible to infer the physics of the earthquake source which is quite complicated and can be difficult to work out. The faulting process (Mikumo and Miyatake, 1987) during an earthquake may be highly inhomogeneous. The distribution of slip (Wald et al. , 1991) over the fault can vary both in amount and direction. There are regions on the fault where large slip might occur which corresponds to asperities. Regions of small slip are found to be close to regions of aftershock activity. Some details of this variability of slip distribution over the fault plane can be inferred from modelling of strong motion data. In the present work the strong motion data recorded during the 1991 Uttarkashi earthquake have been modelled. The source parameters of the earthquake as given by USGS were chosen as the starting point. The objective was to determine whether these parameters need any modification to generate synthetic accelerograms. The details of the slip distribution and the orientation and location of fault,plane in relation to recording stations were found to be important factors. All these aspects were considered in generating the strong motion synthetics. The Uttarkashi Earthquake of October 19, 1991(Ms = 7.0) occurred in Garhwal Himalayas. Its epicentre was located to the north of the Main Central Thrust (MCT) in the Higher Himalayan sub-province which is composed of crystalline rocks. The epicentre of this earthquake was located about 58 km from the site of the 268 m high Tehri dam which is under construction in the lesser Himalayas. A detailed study of the strong motion data, at seven of the thirteen stations (Chandrasekaran and Das, 1992) where it was recorded, has been used to learn about the details of the source of this earthquake. For the purposes of modelling the earthquake source has been represented by a number of point shear dislocations distributed over the model fault plane buried in an elastic ,homogeneous and isotropic half space. The ground motion at a near distance has been calculated by adopting a simplified source radiation process and making use of simple theory (Aki and Richard, 1980). The method adopted consists in integrating the far field contributions of Green's function for a number of point sources distributed at the observation point. The radiation begins at the focus where the rupture initiates and moved outwards. The fault is divided into number of sub-faults, all of them of the same size (1km X 1km) . The computational process has been so designed that it represents radiation from a circular rupture front , i.e., all sub faults where rupture front reaches at the same time , radiate simultaneously. Radiation pattern terms have been evaluated for each subfault separately and effect of the free surface has also been taken into account in an adequate manner. The rupture propagates at a constant speed and the direction of slip vector remains the same during the entire rupture process. The source pulse is taken as a step function with a finite rise time and rounded shoulders (Ben Menahem and Singh, 1981). The location of the point of initiation of rupture, rise time and rupture velocity along with slip amplitude distribution on the fault plane are of critical importance in obtaining a good match of synthetic accelerograms with the observed ones. A number of numerical experiments were carried out to arrive at the proper values of the above mentioned parameters. Synthetic accelerograms have been generated at seven recording stations and compared with the oberved ones forllowing the arrival of S waves. The present study has demonstrated that simple theory applied for an elastic half space earth model ,taking a simple source pulse and a fault plane with varying slip amplitude are able to reproduce most of the important features of recorded accelerograms. A pattern of slip distribution has been obtained on the fault plane. The results obtained in the present study are comparable to the other results obtained for this earthquake, e.g., Khattri et al. (1994). The importance of present work lies in the fact that the technique used for generating strong motion synthetics is very simple and has been applied for the case of a simple earth model (half space). This technique needs specification of a limited set of input parameters which are easy to specify yet yields good results.