THE DATASET AND ITS PREPROCESSING

 

one.shot
Figure 1 A shot profile from the dataset. Note the refracted waves at large offsets. The data have been gained by t^1.5.

 

co.inner.pre
Figure 2 The near-offset section of the data. Note water-bottom multiples and pegleg multiples after some of the deeper reflectors.

 

inner.co.section.decon.raw
Figure 3 The same near-offset section as in Figure 2, now after gapped decon. Some of the water-bottom multiples have been attenuated and the pegleg multiples are attenuated.

Figure 1 shows a shot profile from the dataset. The profile was plotted with t^1.5 gain. The first processing step after reading the tapes and examining the shot profiles was to retain only the inner 120 out of 180 total offsets. The outermost offsets had poor signal to noise and all but the deepest reflectors were swamped by refracted energy. Next, I sorted the data to constant-offset sections and muted the refractions and other steep angle arrivals. Figure 2 shows the near-offset section of the data. Water bottom and pegleg multiples are evident in Figure 2, so I ran single-trace gapped decon on the data. Figure 3 shows the same section as in Figure 2 after gapped decon. Unfortunately, the gapped decon did not remove the water-bottom multiples especially on the nearest offsets. This may adversely impact my ability to use velocity information in the shallow part of the section. I speculate that near offset reflection and multiples were near, at, or beyond critical angle (depending on how many bounces). The water bottom is shallow (60-70 m) and apparently hard since the velocity increases rapidly. This results in amplitude relations between the different order multiples that are not predictable using a 1-D model for wave propagation.

The gapped decon did do an excellent job removing a lot of the ``ringy" appearance of the near-offset sections particularly for midpoints from 1-12 km. Gapped decon did removed or attenuate the pegleg multiples of the deeper reflectors. Compare the bottom-of-salt reflector on Figures 2 and 3.

After decon the data were bandpassed to an upper frequency of 80 Hz. This relaxed spatial aliasing constraints on the migrated output so a coarser output sampling could be used.