From this initial migrated image I calculated the dip (Figure 6) and coherence (Figure 7). To calculate the initial back projection points I selected points above 30th percentile in image amplitude, 45th percentile in dip coherence, and at least four model points away from all other selected points. I then calculated semblance at each point and parsed from the list points with semblance below .16 and whose maximum was less than 40% below the average semblance. Figure 8 shows the final location of back projection points. Note how there are few points near the salt body and in the lower portion of the image.
dipxz.vel0
Figure 6 The calculated dip field for the initial migration image. |
coher.vel0
Figure 7 The calculated dip coherence for the initial migration image. |
pts.vel0
Figure 8 The selected back projection points. Note how the points generally follow reflectors and avoid the salt structure. |
As image quality improves, the automatic selector slowly uses more and more model points. Figure 9 shows the points selected after the five iteration of tomography. Note how there are now many more points in the lower portion of the model and closer to the salt. The velocity model after five iterations can be seen in Figure 10 and the resulting migrated image in Figure 11. Figure 12 shows the CRP gathers of the fifth migration. Note how they are significantly flatter than in Figure 12.
pts.final
Figure 9 The selected back projection points after five iterations. Note the difference from Figure 8. |