Geoscience Reference
In-Depth Information
25 m receiver group intervals, with every other trace being discarded. As with the
resampling in time, an anti-alias filter is applied before this step. The resampling further
reduces data volumes, by a second factor of two.
Step 8 applies a time-varying gain to the data to boost up the amplitudes of the later
arrivals compared to the earlier ones. As the wavefront from the source travels deeper
into the earth, it covers a larger area and also suffers amplitude decay due to transmission
losses and attenuation. Spherical divergence correction is applied to remove the loss in
amplitude due to the wavefront expanding with depth. This expansion means that the
same energy in the wavefront is spread over an increasing area as the distance travelled
by the wave increases, and hence the amplitude of the wave is less. One can see the same
effect (in a 2-D sense) on waves caused by dropping pebbles into water. Near where
the stone was dropped in, the perimeter of the circular wavefront is quite small and its
amplitude is large, but as the wavefront expands its amplitude decreases. In addition to
a spherical divergence correction it is common to apply an additional exponential gain
function with time to account for the transmission and attenuation losses.
2.5.2
Deconvolution
Step 9, deconvolution, is a process that sharpens the wavelet and removes any short
period reverberations. The theory is explained in a number of texts; see, for exam-
convolved with the trace to remove unwanted ringiness. The operator is designed auto-
matically based on characteristics of the traces and a few simple parameters supplied
by the processing analyst. These include the operator gap. The idea is that the operator
will not change the wavelet from time zero to the end-time of the gap, but try to remove
periodicity at times beyond the end of the gap. Deconvolution may be predictive or spik-
ing, with the difference between the two being the length of the operator gap. A short,
or no gap, gives maximal wavelet compression (hence the name spiking) while a large
gap (32 ms or more) attempts to remove periodicity caused by short period multiples
whose period is longer than the gap. In practice, both forms actually perform a mixture
of wavelet compression and de-reverberation. Care should be taken with spiking decon-
volution since theoretically it assumes minimum-phase input data. The output of data
that have undergone too severe a spiking deconvolution process is relatively variable in
phase. Because of this, it is becoming rare to apply spiking deconvolution before stack.
Consistency of wavelet shape and amplitude becomes increasingly important as more
attempts are made to infer subsurface information from the amplitudes of the reflection
events.
2.5.3
Removing multiples
One of the consequences of dual source shooting is that the number of traces in a
common midpoint (CMP) gather is half that of single source shooting. Each source
