Lateral Variations in the Earth’s Crust and Their Effect on Seismic Wave Propagation

Abstract

A review of the results of a number of crustal seismic experiments which were carried out over the Canadian Shield over the past fifteen years in general shows that precise values of crustal velocities are not well determined and are very sensitive to how the data set was sampled. When the seismic data is combined with gravity data, the only reasonable solutions to the inverse problem are ones which have large lateral velocity variations in structure (1). There is also very strong evidence from both the seismic and gravity data of many experiments that indicates that large velocity gradients exist with depth in the crust such that the velocity contrast at the Moho is not large (2). In recent years considerable progress has been made in using random media theory to explain precursors to various earthquake phases and the problem of amplitude and travel-time fluctuations across seismic arrays (3, 4, 5, 6).

In this paper the results of a number of numerical experiments performed by Mereu and Ojo (7) are presented in which seismic waves are propagated through a crust modelled as a random medium. The starting model is a single-layered crust with a vertical velocity gradient. A two-dimensional set of small smooth random velocity deviations is then superimposed on the vertical gradient model. The resulting models show short reflectors at various depths in agreement with many deep seismic reflection experiments (8, 9). Ray-tracing experiments using the method of Gebrande (10) showed that the effect of the lateral and vertical anomalies is to scatter the energy and break up the continuous travel-time curve from a vertical gradient model into segments of Different slope similar to those observed in many long-range refraction experiments. Many of the numerical experiments produced a Pg segment and a P* segment with an ‘apparent’ Conrad discontinuity at a depth of 10 to 20 km. As the correlation distance was increased, the apparent depth tended to increase. For small correlation distances there was a tendency for the travel-time curve to break up into three segments indicating that the PnP branch should be observable above the noise but the position of the associated cusp is very sensitive to the effects of scattering.