Techniques to improve imaging of seismic tomographic data

Abstract

Tomographic estimation of near-surface velocity anomalies has direct application in prestack depth migration. Seismic velocity information extracted from traveltime measurements often suggests unrealistic geological settings. The linear system constructed by ray tracing is typically very large, sparse and poorly conditioned. In most cases it is too large to invert using a generalised inversion algorithm. Instead, a least squares method to solve the system is employed. Limited angular ray coverage traveltime measures using three synthetic reflection data sets have been shown to allow reliable velocity image reconstruction of anomalous zones that vary up to 50 percent from the background. Tomographic imaging using four summation expansion methods, namely, the algebraic reconstruction techniques (ART), and the simultaneous iterative reconstruction technique (SIRT), a conjugate gradient (CG) method and a singular value decomposition (SVD) method, were used to image the synthetic data sets, and the result of each of these tomographic algorithms was evaluated based on their rms misfit and residual, solution stability and sensitively to data noise and algorithmic operation time. Modifications to the algorithms that stabilise the inversion are shown to improve the velocity images obtained and enhance solution convergence rate. These modifications include the application of damping parameters, weighting schemes, smoothing filters including average and median, implemented both pre-and post-solution as well as convolutional quelling. The effect of noisy traveltime data on tomographic images, and useful methods for dealing with this, are also studied.