Page 13 - Acquisition and Processing of Marine Seismic Data

4 1. INTRODUCTION

processing steps, such as velocity analysis or and is known as quality control in data
prestack migration. processing (QC).
During the processing, some of the processing
almost
An
data-
conventional
seismic
steps and their orders of application may differ processing flow used to process marine seismic
from one dataset to another, depending on the data is shown in Fig. 1.3. However, since the
selection of the processor. For instance, while a data-processing flows are dependent on the data
strong tail buoy noise in a marine seismic sur-
and the data processor, any particular noise in
vey may require an application of a suitable
the seismic data may require additional proces-
frequency-wavenumber (f-k) filter, this may not
sing steps and/or application of the steps in a
be necessary for another line. While some data
different order. The data is processed as gathers
processors apply deconvolution to the shot
or ensembles, termed prestack data until it is
gathers, the others prefer an application to CDPs.
stacked. Following the stacking process, the data
A strict quality control (QC) system must be
is called poststack data.
applied to the outputs of each processing step
Because the data-processing flow, steps and
to ensure that the specific processing application
the parameters are determined by the proces-
produced acceptable results. These QC analyses
sing specialists, and hence are operator-
are schematically illustrated in Fig. 1.2:
dependent, the final product of the same raw
• At first, processor should determine the most input dataset processed by different operators
appropriate data processing flow for the may have a different appearance. The output
input dataset, which consists of proper images may have discrepancies by means of
processing steps. their frequency content, S/N ratio, resolution
• Then, the suitable parameters for each and trace-by-trace consistency. The factors
processing step is determined, such as cut-off affecting the quality of the output of the seismic
frequency values for band-pass filtering, data-processing sequence are
deconvolution operator length, etc. These • S/N ratio of the input dataset
processing steps with initially determined • environment of the data collected (e.g., land
parameters are then applied to the input data data almost always has a lower resolution
one at a time.
than marine data)
• After each application, the quality and • algorithm and software used
acceptability of the outputs are analyzed in • data-processing flow
different domains and with different tools.
• experience of the processor (especially in
For instance, output of a band-pass filter can
determining the parameters and QC
be analyzed by checking the power spectrum
implications)
of the band-pass filter output, while the
success of a spiking deconvolution can be The data processor should apply a series of
assured by controlling the auto-correlograms parameter tests to the input data for each step
of the deconvolution output. Whenever a to determine the suitable parameters, and the
processing step produces an unacceptable results are carefully analyzed after each test
result, its parameters are updated and the attempt. However, only a specific part of the
step is re-applied to the data with new seismic data is generally extracted for parameter
parameters to adjust the mistakes arising testing, since using the whole seismic line for a
from inappropriate parameter selections. parameter test is time consuming and is not
This recurrence can be done on shots, CDPs, practical. For instance, to determine the cut-off
brute stack sections or a near trace section, frequencies for a band-pass filter, tests can be