A Study of the Magnitude Difference Ms--mb for Earthquakes

Abstract

Summary The amplitude ratio of surface and body waves generated by a sample of over 1000 earthquakes is studied on the basis of the values of R = M, - nz,,. By considering first of all, only those events with inb between 5 and 6, the dependence of R on magnitude is practically eliminated. It is found that R depends on depth of focus to a smaller degree than expected but that it varies significantly from one region to another (e.g. from Southern to Northern Pacific). It appears that this variation has little to do with propagation path and is more likely reflecting a difference in conditions at the source. The correlation between R and ‘positive influence’ has also been investigated. This leads to the conclusion that high values of R are associated with heterogeneous source regions, low stress drop, and a slow speed of fracture process; thus giving rise to preferential generation of long periods. The comparison of amplitudes of surface and body waves is now one of the principal methods of discriminating at teleseismic distances between earthquakes and underground explosions, (UKAEA 1965). It is based on empirical and theoretical evidence that differences in source mechanism, size and duration, produce different relative values for the amplitudes of surface and body waves. But if such differences in the source are observable in this way then one should expect that the difference in size and duration of different earthquakes will have a similar effect. Of course, the magnitude of the effect is likely to be smaller, because the difference between two earthquakes is not in general so large as that between an earthquake and an explosion. However, a swarm of shocks in the Gulf of California and an isolated shock in Northern Baja California of about the same local magnitude were found by Thatcher & Brune (1971) to have very different ratios of body to surface wave amplitudes. In general the ratio fluctuates very much from one earthquake to another, so for the reliable detection of a systematic trend it is necessary to use statistical methods, treating earthquakes with the same general source conditions in big samples. One way of doing this is to separate all earthquakes into groups by the regions of their occurrence. Different source conditions can also be expected between initial and subsequent shocks in a chain of interdependent earthquakes. A simple example of such a chain is that of aftershocks following a strong earthquake. Statistical methods for the identification of interdependence in earthquake groups has been given by Keilis-Borok, Podgaetskaya & Prozorov (1972) and by Prozorov & Rantsman (1972).