Tomographic determination of interval velocities from picked reflection seismic data

Abstract

Tomography in the context of reflection seismology might be defined as the use of measured traveltime data to determine rock velocities, I propose a method of tomographic inversion that uses not only traveltimes measured between source points and receiver points, but that uses ray parameter information (essentially, the angles at which reflected waves leave the source and arrive at the receiver) as well. A consequence of this approach is that reflecting horizons may be assumed to be only locally continuous, and horizon positions are not determined during the process of inverting for interval velocity. A related consequence is that picking of traveltime + reflection-parameter data may be automated in a straightforward way (this automated picking technique was developed in the USSR as an aspect of the method of controlled directional reception). I invert the picked data by means of an iterative ray-tracing technique to find the velocity model that minimizes a nonlinear least-squares objective function. This objective function is based on the mismatch between the actual paths of rays in the ground, as determined from the picked data, and the computed paths of rays in the velocity model. A penalty term is included in the objective function, to discourage sharp variations in the velocity model. This tomographic method gave good results when I have used it to invert synthetic data from vertically stratified media, and my preliminary attempts to invert synmetic data from laterally inhomogeneous media gave promising results as well.