Automatic velocity analysis of crosswell seismic data

Guoping Li, Robert R. Stewart

A new method for velocity analysis of crosswell seismic data is discussed in this paper. Based on semblance analysis, this method derives velocities from crosswell direct arrivals in an automatic manner and so avoids time-consuming hand picking of traveltimes. To develop the method, an isotropic continuous elastic medium with a linear velocity-depth relationship is assumed. Two steps are involved. In the first step, traveltimes for direct arrivals are calculated for different velocity guesses, using theoretical equations we derive for a linear velocity function. Theoretically calculated traveltime trajectories of direct arrivals have been found to exhibit a quasi-hyperbolic pattern, one characteristic appearing on real crosswell data. A numerical study shows that they agree very well with those measured from synthetic data. In the second step, coherency (semblance) analysis is done for amplitudes of direct arrivals within a time window along each traveltime trajectory calculated for different velocity guesses. The velocity with the largest semblance value is then picked and used as final inversion output.

Examples of inverting velocity information from direct arrivals in synthetic, physical modeling, and field crosswell seismic data gathers, by using the new velocity analysis method, are given. Results prove this method to be a potential velocity inversion technique with efficiency and reliability.