Pre-selection of seismic rays as a possible method to improve the inverse problem solution

Abstract

SUMMARY The paper is a contribution to the study of the instabilities associated with traveltime data inversion in tomographic seismology. It has been found that there are circumstances in which instabilities may be mitigated by reducing in a proper way the number of seismic paths taken into account for the inversion. In order to choose such a set of rays, it has been found useful to apply a pre-selection algorithm which, starting from the set of all available seismic paths, gives a new configuration of rays by minimizing the configuration score. the score is a scalar whose value depends upon some geometrical characteristics of the ray pattern, such as the local ray density and the azimuthal distribution. the method is presently applied to 2-D models where the region crossed by the seismic rays is partitioned into rectangular cells, which is a feature not unusual in tomographic investigations. Numerical experiments of inversion are carried out in order to show the ‘goodness’ of the new algorithm. Comparison with solutions computed without pre-selecting the seismic rays shows the advantage of applying our algorithm for the considered cases. As a measure of the performance, the magnitude as well as the space distribution of the discrepancy between the computed field and the true solution is used in addition to a number of other suitable indicators.