Results and Analysis

In order to increase the signal-to-noise ratio of the data for comparison purposes ``supergathers'' were created, with each one taken as an average of 11 consecutive CMP's. Figure 4 shows some of these supergathers before the multiple suppression. Notice the almost flat primary reflections and the curved multiples. Figure 5 shows the same supergathers for the extracted primaries. Figure 6 shows the remaining multiples (the plot amplitude has been amplified to show the details of the curvature). The extraction of the multiples was successful except on the shallow part where there are not enough traces to discriminate between primaries and multiples. Basically the same results were obtained with all three implementations of the Radon Transform.

Figure 7 shows the stacked section of the primary reflections after the suppression of the multiples. Comparison of this section with Figure 2 shows that many of the reflections in the original stacked section did indeed correspond to multiple reflections. As mentioned before, these multiple reflections would have been virtually impossible to identify in the original stacked section. To stress this point, Figure 8 shows an amplified version of the stacked section for the multiples. Obviously the shallow part is suspect as explained before, but the deep section shows the most prominent multiple reflections. If a stacked section such as this were handed to a seismic interpreter, there is the risk that he could make erroneous inferences about the subsurface. A more extreme case would be that in which the NMO correction was performed with the velocity of the multiples, such as could happen if the multiples (at least some of them) were incorrectly taken to be primaries.

 

HR_mul1.agc.SG
Figure 6 Supergathers after multiple suppression. Residual multiples.

It is important to realize that the suppression of the multiples not only makes interpretation easier by highlighting the primaries, but also improves the resolution of the primaries by allowing a better selection of the primary stacking velocities. Figure 9 shows a typical velocity analysis before the suppression of the multiples, whereas Figure 10 shows the same velocity analysis after the suppression of the multiple energy. It is clear that the presence of the multiples masked the velocity trend of the primaries making it more difficult to select the correct stacking velocity function appropriate for the NMO correction of the primaries. By getting rid of the multiples it becomes clear what the primary stacking velocities should be. The better selection of the primary velocities improves the image of the stratigraphic features of interest such as pinchouts of Paleozoic against the basement.

 

HR_prim_stack1.rot
Figure 7 Stacked section of primaries only.

 

HR_mul_stack1.rot
Figure 8 Stacked section of residual multiples.

 

velan4
Figure 9 Velocity analysis before multiple elimination. Notice the multiples at low velocities.

 

prim_velan2
Figure 10 Velocity analysis after multiple elimination.