Geoscience Reference
In-Depth Information
8.3
Seismic repeatability
If we are thinking of comparing travel-times, amplitudes or seismic velocities between
a baseline and a repeat survey, there are several possible other causes of differences
between the surveys that will interfere with our comparison:
(a) Noise. Ambient noise can easily vary from one survey to another. For example, if
the repeat survey is being carried out over a producing field, there will probably be
many sources of industrial noise that were absent when the pre-production survey
was shot. This need not be a significant problem so long as the signal to noise ratio
at reservoir level remains high.
(b) Accessibility. A common problem is that some areas that were shot in the baseline
survey are not accessible to the repeat survey because of production facilities. For
example, the presence of a production platform offshore will create a hole in the
survey acquisition because it will not be possible for a survey vessel to approach it
very closely for safety reasons. This means that over part of the area of interest, it
will not be possible to acquire a survey that closely replicates the original acquisition
pattern.
(c) Near-surface effects. There may be quite large changes in the near-surface over
time, for example as a result of changes in the depth to the water-table with the
season of the year. Although in principle it is possible to allow for these effects, they
introduce an additional uncertainty when comparing two surveys. The shifts may
be comparable to or greater than the expected time-lapse time shifts, but of course
the latter would be confined to the interval below the reservoir whereas near-surface
effects will cause a shift of the entire trace. In the marine case, similar effects can
be caused by a change in the sound velocity in the sea-water column, owing to
temperature changes caused by large-scale currents. Tidal effects also need to be
allowed for.
(d) Source signature. This may differ markedly from one survey to another. Ideally, seis-
mic data should have been converted to zero-phase during processing
(chapter 2)
.
If this has been successful, it will remove much of this problem.
(e) Acquisition parameters. There may be differences in the geometry of the source
and receiver pattern between the surveys. Jack
(1997)
quotes some examples of
effects that were simulated by taking a well-sampled survey and removing some
of the data, followed by processing of the decimated dataset. For example, when
an 80-fold marine dataset was decimated to 40-fold, then the difference section
between the final 80- and 40-fold sections along a typical dip line had an rms
amplitude 40% of the original section. In such a test, because the decimated data
were produced by merely dropping traces from a 'baseline' survey, positioning
repeatability is not an issue. In real data, two problems arise. One is that, even if
we knew the positions of sources and receivers precisely for the baseline survey,
