The domain for picking plays a major role in deciding the type of tomography algorithm. Different algorithms use different domains for picking, such as the time or the depth in prestack or poststack, respectively. Among these picking algorithms, I chose the algorithm that picks events in the stacked image after prestack migration, which requires a minimum number of picks and provides reflector information. The advantage of having reflector information in the method based on event picking in the depth domain is achieved by prestack migration. For this research, I used planewave synthesis imaging that synthesizes planewaves at the surface.
As an example, I have generated a synthetic data set with finite-difference acoustic modeling. Figure shows the interval velocity model used to generate the synthetic data. To avoid surface-related multiples, an absorbing boundary is applied at the surface. The near-offset section of the synthetic prestack data set, Figure , shows that the surface-related multiples have not been generated.
As an initial guess for the velocity of the medium, a two-layer model, shown in Figure , is used, with water velocity for the top layer, because we can easily estimate the depth of the water bottom from the seismic data even in a real case.
In order to obtain a reference image to pick dominant reflectors, I apply surface-oriented planewave synthesis imaging. I synthesized thirty-one different incidence angles at the surface; the stacked image of them all is shown in Figure . Figure shows the locations of the reflectors picked, with reflector numbers assigned along them.