In the presence of multiples

High contrasts between layers in the velocity model cause a very large number of multiples to be generated. The result of migrating this multiple-affected dataset with the background trend from the left panel of Figure is shown in the upper panels of Figure . FEAVA is indicated by a vertical path of high energy in the middle of the image and is clearly outlined by the FEAVA detector. Non-focused multiples depart from Shuey's approximation too, but the resulting FEAVA detector output is one order of magnitude smaller than that caused by actual focusing. The lower panels of Figure show the results of migrating with the correct velocity model, albeit with a single reference velocity. The focusing is no longer visible in the image. The focusing-caused FEAVA detector output has fallen significantly, to the level of power of surrounding multiples. The two FEAVA outputs are displayed in the same intensity scale.

 

bg-refvel1
Figure 5 Top-left: image produced with the linear background velocity trend; Top-right: output of FEAVA detection applied after linear background velocity migration; Bottom-left: image produced with the correct velocity (split-step kernel with one reference velocity); Bottom-right: output of FEAVA detection applied after split-step migration with the correct model, one reference velocity. FEAVA detector outputs in the same color scale, for comparison.

 

refvel8-infill
Figure 6 Top-left: image produced with the correct velocity (split-step kernel, eight reference velocities); Top-right: output of FEAVA detection applied after migration with the correct velocity (eight reference velocities); Bottom-left: image produced with the correct velocity (split-step kernel with one reference velocity) on dense data; Bottom-right: output of FEAVA detection applied after split-step migration with the correct model, one reference velocity on dense data. For detail enhancement, FEAVA detector outputs are not in the same color scale.

FEAVA was clearly reduced by migration, but not entirely eliminated. One natural question is whether significantly increasing the number of reference velocities in migration will improve the outcome. However, a migration with eight reference velocities which produced the upper panels of Figure , show that this is not the case for this type of stratigraphic play. The improvements are incremental, visible only by electronically displaying the two pictures in an animated sequence.

Another potential limitation stems from the fact that, due to the combination of depth/offset sampling Sava and Biondi (2001), the range of angles into which offsets can be reliably transformed was limited to $15^\circ$, while the FEAVO detector works up to $30^\circ$. Would energy from greater angles improve the situation? The bottom panels of Figure are produced with an offset sampling four times smaller than before, resulting in reliable transformations from offset to angle up to $45^\circ$. This does not bring improvements either. On the contrary, multiples, highly curved at large angles, create more noise in the FEAVA detector output. The extra smoothness comes from having decreased the midpoint sampling by a factor of four.