Application

The Mahogany field, located in the Gulf of Mexico, is dominated by a salt body structure. One of the data set acquired on this area consists on a 2-D Ocean Bottom Seismic (OBS) multicomponent line. Rosales and Guitton (2004) present the steps involved in the PZ summation that results in the P component section that is now under study. The PZ summation was successful in eliminating the receiver ghost; however, other surface-related multiples, like the source pegleg, are still present after the combination.

The remaining multiples are a problem when performing any migration-velocity-analysis technique. Therefore, we apply the methodology discussed in the previous section to eliminate the remaining multiples. Figure  shows three characteristic angle-domain common-image gathers, after three processes: the PZ summation described in Rosales and Guitton (2004), the downward continuation of the sources to the receiver label, and the migration with the velocity model presented in Rosales and Guitton (2004).

 

adcigs
Figure 2 ADCIGS after PZ summation at three different locations.

It is possible to observe a similar residual curvature for both primaries and multiples. Additionally, at this stage the primaries and multiples present a very similar moveout, since the migration velocity is still not perfect. Any migration-velocity-analysis technique done with this data will be biased with the multiple reflections still present. Additionally, Figure  presents the same ADCIGs as in Figure  but after Radon transform in the angle domain is not easy to distinguish between primaries and multiples; therefore, another process is required.

 

adcigs.art
Figure 3 ADCIGs in the Radon domain, before any residual curvature process.

A residual curvature process will help to separate primaries and multiples in the Radon domain. The advantage of using residual migration (as discussed on the methodology) over residual moveout is that residual migration reduces the effects of image-point dispersal between events imaged at the same physical location but with different aperture angle Biondi (2004). Performing SRM with a value of $\rho$ different than 1 (e.g., $\rho=2.0$) produces a distinct difference between the residual moveouts of the primaries and multiples. Figure  shows the same ADCIGs as in Figure  after SRM with $\rho=2.0$. It is now easy to distinguish between primaries and multiples through their distinctive curvatures.

 

adcigs.srm
Figure 4 ADCIGs after SRM with $\rho=2.0$.

Applying the Radon transform splits the image into two different curvatures; therefore, it is possible to distinguish between primaries (positive curvature) and multiples (negative curvature); additionally, it is possible to apply principles and techniques similar to those discussed by Sava and Guitton (2003) or Alvarez et al. (2004). Figure  shows the ADCIGs in the Radon domain. After SRM is possible to distinguish between primaries and multiples, compare the results on Figures  and . Figure  presents the result after eliminating the multiples and applying SRM with a $\rho$ value of $\rho=0.75$; this result shows a satisfactory elimination of multiples.

 

adcigs.srm.art
Figure 5 ADCIGs after SRM in the Radon domain. Compare with the ADCIGs in Figure  and note how multiples and primaries are split into two different trends.

 

adcigs.srm.art.isrm
Figure 6 ADCIGS after multiple suppression.