LOCAL WINDOWS

The effectiveness of this method depends on the amount of overlap between the unaliased energy at one dip and the aliased energy of another dip. If all the aliased energy is in the same region as some unaliased energy it will not be rejected. If the unaliased energy has a broad dip range e.g. because of event curvature then overlap with aliased energy is much more likely. One way to reduce the dip-bandwidth of the $\omega-p$ spectrum is to process the data using local windows. The windows should be designed to include only a few distinguishable dips.

There are two tradeoffs to be considered when designing windows. As the windows become smaller in time the number of frequencies in the $\omega-p$ domain is reduced. When the number of frequencies is smaller the effectiveness of the continuity detection is reduced. As the windows become smaller in space the dip resolution of the slant stack is reduced. The real and aliased events become more spread-out in slowness and are more likely to interfere.

Figure  shows the result of reconstructing the original data using least squares transforms applied in windows on the data. The windows were 32msec long and 32 ft wide. Each window overlapped its neighbors on each side by 25%. I used the data patching algorithm described elsewhere in this report (). This reconstruction is slightly superior to the one in figure  where the data was all processed together.

 

real3-win-diff
Figure 22 Results of reconstructing data at the original 1ft spacing from data at a 2ft spacing. The least squares slant stack was applied in windows that were 32msec long and 32 ft wide. Top panel is the reconstructed data, the bottom panel is the difference from the original data. This result should be compared with