Using CAM as the imaging operator in my least-squares inversion means that my geophysical regularization operator will only be acting along the inline ph axis. Therefore, the most significant changes between the CAM result and the geophysical RIP result will be seen in unstacked results. To study these, I have selected several unstacked inline volumes from crossline=20.9 km. These volumes can be seen in Figures through , where (if you rotate the pages 90o clockwise) I am displaying the depth slice on the top of the figure, the common inline ph section on the left, and the common image gather to the right of the common inline ph section. The migration result is shown first, then the geophysical RIP result for each (common inline ph section - common image gather) pair.
Figures and show a common inline ph section from ph=.1875 and a common image gather from inline position 20.375 km. Comparing Figure and Figure , the RIP result is considerably cleaner. The effects of the regularization are clear in the common image gather (right part of the figures), where the unlabeled oval indicates gaps in the events that are filled by RIP. In the common inline ph section, the ovals indicate particular areas of the shadow zones that are being filled in. Oval ``A'' highlights a reflector that is discontinuous and has inconsistent amplitudes where it does exist in the migration result. In the RIP result, this reflector is continuous with strong amplitudes along its full extent. Oval ``B'' extends across one of the shadow zones. The shadow zone is considerably cleaner in the RIP result, with almost none of the up-sweeping artifacts seen in the migration result. Also, the reflectors themselves extend farther into the shadow zone, particularly on the right side of the oval. The events also extend farther into the shadow zone indicated by oval ``C''.
Moving further under the salt and to a larger offset ray parameter, Figures and show a common inline ph section from ph=.2325 and a common image gather from inline position 21.65 km. The unlabeled oval in the common image gather shows events that have been strengthen and are more horizontal in the RIP result (Fig. ) than the migration result (Fig. ). The ovals marked ``A'', ``B'', and ``C'' on the common ph sections show the same type of improvement seen in the previous comparison (Figs. and ). The reflectors at the right side of oval ``B'' in Figure extend much farther into the shadow zone than those from the migration result.
Another interesting comparison can be made at common inline ph section from ph=.15 and a common image gather from inline position 20.9 km (Figures and ). In this migration result (Fig. ), the common image gather once again has events with gaps caused by poor illumination, indicated by the unlabeled oval. These gaps are largely filled by 7 iterations of RIP, as seen in Figure . We see the same improvements in ovals ``A'', ``B'', and ``C'' as discussed for the previous two comparisons. In this comparison, there is an additional oval ``D'' that indicated reflectors under the salt that are less affected by artifacts and more likely to be accurate in the RIP result than in the migration result.
Finally, looking at the same common image gather from inline position 20.9 km but moving the common inline ph section to ph=.2925, we compare Figures and . In this common ph section, the effect of the critical angle and user-defined mutes described in the previous chapter can be seen in and below the salt body. The oval in the common image gather shows the same events with gaps in the migration result (Fig. ) that are filled by RIP (Fig. ) that was seen in the previous comparison. It is interesting to see that the improvements seen in the ``A'', ``B'', and ``C'' ovals in the previous examples are also seen in this example, at a much larger ph. RIP has a positive effect for a large range of ph.