CRUSTAL STRUCTURE OF INDIA AND ADJOINING REGIONS USING RAYLEIGH WAVE DISPERSION

Abstract

Among the different methods which have been used to deduce crustal structure from seismic observations, the use of surface wave dispersion holds a prominent place. This method has been used since early 1950's. The development of the method of computing theoretical Rayleigh wave dispersion curves by Haskell and its subsequent improvement by other workers, establishment of long period seismographs by Ewing's group at Colombia, development of the improved techniques to compute group velocity dispersion curves from seismograms and establishment of global digital seismograph network, which yields high quality data have been the important landmarks in the history of surface wave dispersion studies. A Wide Band Seismological Observatory, established at the Department of Earth Sciences, University of Roorkee, Roorkee in October, 1982, has yielded a unique data set of earthquake records. The seismographs installed at this observatory have a flat velocity response from 0.2 to 20.0 seconds, which enables these to generate analog records of local as well as distant seismic events with equal fidelity. The Rayleigh waves recorded by these seismographs have been found to span a period range of 3.0 to greater than 20.0 seconds. This provides a valuable opportunity to use short period Rayleigh waves to get refined models of the crustal structure at shallow depths. The crustal structure of India and surrounding regions has been investigated by a number of workers. A variety of methods has been used to compute group velocity dispersion curves from Rayleigh wave data recorded by various types of seismographs at different locations. In the present study Rayleigh waves recorded at Roorkee have been processed using multiple filter technique to get group velocity dispersion curves and to deduce crustal structure. For this purpose 18 earthquakes have been selected from seven source zones in and around India. All of which have produced well dispersed wave trains of Rayleigh waves. Computer programs have been written for implementation of multiple filter technique, which includes computation of response of seismographs and its use in applying instrumental correction in frequency domain. The validity of this technique was tested by applying it to synthetic seismograms generated for given dispersion curves of prescribed shape. Another set of computer programs were written to generate theoretical dispersion curves of Rayleigh waves for a layered earth model using Knopoff's method. In this program roots of the dispersion function have been located by pattern search method, and group velocity has been obtained by numerical differentiation. Use of Rayleigh wave records at a single station to deduce the seismic velocity structures in different crustal regions is one of the important features of the present work. Earthquakes used in this study have occurred in Iran, Hindukush, Afghanistan- Tadzhakistan Border, Kirghizikstan, Tadzhakistan, Tibet, China and India-Bangladesh Border. The analog records have been digitized and the digital data so obtained were analysed to obtain Rayleigh wave group velocity dispersion curves. The observed dispersion curves have been matched with theoretical dispersion curves generated using Knopoff's method. No formal inversion of dispersion curves has been attempted. Rayleigh waves in the period range 3.0-50.0 seconds have been used. Short period Rayleigh waves have yielded better control on the thickness and velocity of top layer. The interpretation of dispersion is in terms of average crustal structure over the entire path between the respective source region of earthquakes and the recording stations. The paths cross diverse tectonic provinces.For the purposes of interpreting observed dispersion curves all plausible crustal models have been examined and their theoretical group velocity dispersion curves derived for comparison with observed curves. The total absolute errors in the matching of observed and computed dispersion curves have been found to be below 10%. In the following paragraphs some important resuls have been described. In the East Ganga Basin this study has revealed a crustal thickness of 40.0 km. which agrees with that found by earlier workers. The thickness of top most sedimentary layer has been found to be 5.0 km which contrasts with the value of 3.0 km found earlier. The path of Rayleigh waves from the epicentre of the only event used is confined entirely to the East Ganga Basin. The Rayleigh wave paths from three events in east China cross part of the Tibet plateau and Himalayan ranges. Over this path the average crustal thickness is about 61.0 km with a low speed layer at the top and another low speed layer in the depth range of 21.0- 36.0 km. The shear wave velocity in the low velocity layer in the crust is only 3.71% less than that in the layer above. The three events in the Tibet region have their epicentre almost north of Roorkee. The paths of Rayleigh waves have a substantial fraction in Tibetan Plateau and also cross Himalayan ranges. The dispersion data for Tibet region suggest the presence of low velocity layer at a depth of 20.0-40.0 km. The presence of the low speed layer has been suggested by a few other workers. The shear wave velocity in this layer has been found to be 3.71% less than that in the layer above. The average crustal thickness turns out to be 65.0 km. The dispersion curves from four earthquakes from the Afghanistan region suggest a three layer crust with a total thickness of about 66.0 km. For the Hindukush region from which five events have been analysed, a four layer crust with total thickness of about 71.0 km is found. A thin velocity layer of 6.0 km thickness at a depth of 30.0 km has also been found, the presence of this has also been suggested on the basis of body wave studies. Athin layer at the top is found in both these cases. An average crustal thickness of 46.0 km. is suggested in the region which includes part of Iran, Pakistan and Western Ganga Basin. No previous study exists with which these results could be compared. Some geophysical studies in the Western Ganga Basin have revealed an average crustal thickness in the range of 38.0- 45-0 km. The present study has demonstrated the usefulness of using Rayleigh waves from Wide Band Observatory to deduce crustal structure in that the details of near surface layers have become more refined. Non availability of sufficient number of events with well developed dispersed Rayleigh wave trains nas been one of the main limitations of this study.