Page 366 - Acquisition and Processing of Marine Seismic Data

6.10 QC IN DECONVOLUTION 357

Surface consistent deconvolution also assumes a i. Picking the autocorrelation gate
minimum phase source wavelet and noise-free ii. Picking the deconvolution design gate
seismograms. It is recommended to be applied iii. Calculation of the autocorrelations of
to the datasets before the AVO analysis. selected shots
Although the method is widely used in onshore iv. Determination of deconvolution parameters
seismic processing, the most reasonable from autocorrelograms
application in marine seismic processing is v. Application of the deconvolution
decomposition only for source location. On the vi. Checking the results to ensure the quality of
receiver and offset functions, imperfect geo- deconvolution output
phone coupling or localized soil conditions
Autocorrelations are generally calculated and
and the variations in near-surface conditions
displayed for a number of shots. It is recom-
are accounted for in the decomposition of the
mended to select shot records at different shot
land survey data. In marine seismic, however, a
locations along the seismic line so that the auto-
poor receiver coupling or the effect of a weath-
correlation characteristics of the seismic data
ered surficial layer is not pronounced; the
from beginning to the end of line can be ana-
most important part is the repeatability of
lyzed together. This enables us to identify
the source or the stability of the source signal.
whether the autocorrelation of the source wave-
Therefore, a decomposition accounting for the
let is stable for all shots in the line, which is par-
source function for marine seismic data can pro-
ticularly important when the geology varies
vide the necessary whitening in the amplitude
laterally in the survey area. Fig. 6.45 shows an
spectrum.
example common offset section of a regional
As an example application, Fig. 6.43 shows a
seismic line and autocorrelograms of its selected
number of stack sections, each of which is shots from four different locations on the line.
decomposed for the convolution effects of dif- Locations are selected from the points where
ferent components as source, receiver, CDP, the geology tends to change laterally, to analyze
and offset locations, as well as the effects of all its effect on the shape of the source wavelet
components together. Although the decomposi- embedded in the seismic data. Indeed, the
tions based on offset, CDP, or receiver functions appearances of the autocorrelograms do change
do not ensure any whitening in the amplitude along the line, including the length of the first
spectra of the deconvolution output, decompo- transient zone. In areas of complex geology,
sition based only on the source locations yields
especially on the continental margins of steep
a significantly whitened amplitude spectrum.
slopes, it may not be easy to reliably determine
Fig. 6.44 compares a small portion of a shot
the operator length from the autocorrelograms.
gather from a conventional marine seismic data-
Although some of the commercial processing
set before and after surface consistent deconvo-
software allows us to use variable deconvolu-
lution, in which the decomposition was
tion operator lengths along the seismic lines,
performed only for source location.
generally one single operator length that opti-
mally represents the whole seismic line is deter-
mined from computed autocorrelograms.
6.10 QC IN DECONVOLUTION Deconvolution tries to whiten the amplitude
spectrum to yield a broader-band data, and this
Application of a Wiener deconvolution pro- widening is also achieved both for extreme
cess consists of the following steps, which are low- and high-frequency components of the spec-
generally applied to the shot gathers: trum, which are sometimes dominated by the