Figure 32 Correlation of local plane-wave estimate and original image. The fault is resolved sharply laterally, but is blurred vertically. The sharp lateral resolution leads to gaps in the fault line where the original adjacent layers lack contrast.
The first seismic test case in Figure 33 is a success. The wavefield character of the image due to the sedimentary layers vanished. The major faults are delineated, particular the large fault in the south-west corner. The faults are sharply resolved, and contrast well with the otherwise eventless background.
Unfortunately, the image shows a few shortcomings. In some locations, the image shows a distinct ringing that seems to be due to insufficiently suppressed, steeply dipping layers. The salt body and its internal chaotic reflections obscure the center of the image. The pentagonoid region is not delineated, nor are the major truncating faults within the salt body. In places, the image shows some distinct north-south streaks, which I believe are due to the acquisition footprint that is also visible in the original image.
The loss of resolution and lack of faults in the vertical sections of Figure 33 is due to the correlation over time. The distinct border between the upper and lower data part indicates that the patch size should possibly be reduced.
Overall, the quality of the discontinuity image seems comparable to the quality of commercial discontinuity maps (); Bahorich and Farmer (1995). My salt dome image is a subimage of one of Bahorich's examples and the attributes can be compared directly.
The discontinuity image of Figure 34 suppresses the wavefield character of the original image, which was not excessive in the first place. The discontinuity map reveals a complex pattern of numerous faults and discontinuities. The discontinuities are well resolved. The discontinuity map is slightly contaminated with noise, which reduces the sharpness and contrast of the individual events and which is most-visible in the image's center. The noise might be due to events that only partially fall within the correlation's temporal averaging. The vertical section indicate that a few strong sedimentary layers at the bottom were not suppressed well.
Are the image's discontinuities truthfully delineated? The faults that I can identify in the original image all correspond to events of the discontinuity map. Some of the events - e.g., the fault marked F in the original image 12 - deviate from the rather continuous and smooth event that I detect in the original seismic image. Other equally strong events of the discontinuity image seem to relate to what amounts to small, hardly visible discontinuities in the original image. If my observations are correct, one may argue that the discontinuity attribute fails to distinguish between major and minor discontinuities. On the other hand, the method deserves credit for detecting such minor discontinuities.