Accuracy evaluation of typical, infinite bandwidth, minimum time seismic modeling schemes
I.F. Louis, K.D. Marmarinos and F.I. Louis
Journal name: Journal of the Balkan Geophysical Society
Issue: Vol 8, No 3, August 2005 pp. 113 - 128
Info: Article, PDF ( 4.94Mb )
The performance of a variety of travel time modelling schemes is evaluated, in terms of their accuracy, by using numerical simulations. The methods considered predict minimum travel times in two-dimensional P-wave velocity model parameterizations. They are typical, infinite bandwidth, seismic modeling schemes implemented in many inversion and imaging algorithms The methods considered include: the finite differencing approach of the eikonal equation as described by Vidale, the shortest path method as described by Moser and a variation of it as described by Nakanishi and Yamaguchi for calculation of seismic rays and first arrivals, and the two-point ray tracing technique as presented by Vesnaver. Quantitative comparisons between forward modeling approaches are based on the misfit between true and calculated travel times, number of model parameters used and the smoothness of the velocity field. When the constant velocity distribution is utilized Vidale’s method is giving results within tolerance with respect to travel time accuracy in opposition to the ray perturbation and SPM scheme where the error is lower. For fine model parameterizations, Vidale’s scheme remains the most efficient in computer time but still suffers in the error distribution. In stratified models, Vidale’s scheme gives increasing errors for increasing velocity contrast. In smooth velocity fields a drop in the misfit is observed in the far field and the method is still computationally faster. Shortest path calculations are independent from the model whereas the error remains related to the number of possible routes. The two-point raytracing scheme performs better in smooth velocity fields giving time misfits strongly associated with the distance and smoothly decreasing for far offsets.