Geoscience Reference
In-Depth Information
2.5.6
Post-migration processing
After migration the data are more correctly positioned although they may still not be
perfectly located owing to errors in the velocity field, use of time migration rather than
depth migration, anisotropy, etc. However, at this stage the data should look like a
cross-section through the earth with reflections corresponding to changes in acoustic
properties and unwanted events such as multiples and noise removed. The final step
is to convert the data to zero-phase which centres the peak of the seismic wavelet on
the impedance contrast. Even when an attempt has been made to keep the data zero-
phase throughout the processing sequence, it is likely that there is still considerable
phase uncertainty due to the effects of attenuation. Zero-phasing is best performed
with well data. An operator is determined by least-squares matching the seismic data
to the reflectivity generated from the well data (White (1980) and Walden & White
(1984 ) ). Since there is likely to be some error in the positioning of the seismic data
after the migration, this matching is usually done for a number of traces surrounding
the well location. The fit between the seismic trace and the well log generated synthetic
is compared and the trace location that gives the best match is chosen as the trace
for estimation of the wavelet. The process is explained in more detail in chapter 3
(section 3.1.1) . Figure 2.37 shows the wavelet extraction obtained from one of the
commercially available interpretation packages. Once a wavelet has been extracted an
operator can be designed to convert it, and hence the data, to zero-phase. Note that
such an operator is really only applicable to the time-gate used for the extraction. It is
extremely difficult to confidently zero-phase data over large time windows.
The final three steps in the processing sequence outlined in fig. 2.21 are all concerned
with fine tuning the data prior to loading to the interpretation workstation for detailed
analysis. Step 30 applies an equalisation on a trace by trace basis to ensure the spectral
content of each trace is broadly similar, while step 31 applies time-varying bandpass fil-
ters to reduce the higher frequencies with time to eliminate those that are mostly noise
due to their attenuation on passing through the rocks. Finally, step 32 applies time-
varying trace scaling to ensure a balanced-looking section with time. One approach
is to apply Automatic Gain Control (AGC). This applies a time-varying gain to each
trace individually, with the gain calculated so as to keep the average absolute ampli-
tude constant within a window that slides down the trace. A short time-window AGC
(say 200 ms) is highly undesirable if any use will be made of amplitude information
subsequently, because it tends to destroy lateral amplitude changes that may be impor-
tant. A long gate AGC (say 1000 ms or more) is usually acceptable, however, because
the gain is not much influenced by the amplitude of any single reflector. All these final
steps are important, because they can aid or inhibit the interpretability of the dataset that
is delivered to the interpretation workstation. Remove too many high-frequency data
and subtle detail may be missing; leave too much noise and automatic batch tracking
of horizons may be compromised.
 
Search WWH ::




Custom Search