Page 381 - Acquisition and Processing of Marine Seismic Data

372 7. SUPPRESSION OF MULTIPLE REFLECTIONS

FIG. 7.7 (A) A schematic CDP gather after NMO correction, in which the primary reflections (P) are flattened whereas the
multiple reflection (M) is still hyperbolic. In areas of near offsets, multiple amplitudes are almost in-phase (blue area), and after
stacking, their amplitudes strengthen each other. In the far offset areas, however, multiple amplitudes are out-of-phase (red
area), and after stacking, their amplitudes tend to diminish. (B) A real CDP gather example after NMO correction indicating a
proper inner mute zone (blue area) to remove the in-phase multiple amplitudes.

number of traces involved is N, then the ampli- seismic data involves long offsets. However, it
tude of the multiples will be decreased by an removes a significant amount of multiple energy
order of 1/N (Schoenberger, 1996). Observations only in the stack sections, and no improvement
indicate that the amplitudes of the multiples, by means of multiple removal for prestack data
however, change with the offset in CDP gathers is achieved by stacking. Therefore, in practice,
in such a way that the multiple amplitudes are the effect of stacking could be regarded as an
slightly higher than the primary reflection additional contribution to multiple suppression,
amplitudes for far offset traces, and lower for which removes the residual multiple energy
near offsets. Using this information, one can remaining after the specific multiple removal
analyze the amplitudes in a CDP gather to techniques.
obtain the most suitable weighting coefficients
from near to far offsets by means of a least-
squares approach. Then using these optimum 7.2 PREDICTIVE
weighting coefficients in stacking, a better sup- DECONVOLUTION
pression of multiples can be accomplished. This
process is termed weighted stack. Suppression of multiple reflections by predic-
Stacking is one of the most effective processes tive deconvolution is based on the periodicity of
to suppress multiples, providing that the the multiples in the seismic data. A prediction