.nr HM 1.5i
.nr FM 0.5i
.nr PO 0.8i
.nr LL 6.7i
.nr PS 8
.nr VS 9
.na
.nh
.EQ
delim $$
gsize 8
gfont roman
.EN
.LP
.TS
center, tab(:);
cbp12 s s s s
l l l l l
liw(1i) lw(1.6i) lw(1.5i) lw(1.7i) lw(0.8i).
Seismic Processing Steps
.sp 0.3i
Process:Purpose:When Applied:Pitfalls:Cost (flops)
_
.sp
.T&
cbp10 s s s s
liw(1i) lw(1.6i) lw(1.5i) lw(1.7i) lw(0.8i).
Data Reduction
_
.sp
Demultiplex:T{
Put data in trace sequential (SEG-Y,D) format, from multiplexed (SEG-A,B) field
tapes.  Label headers with geometry information.
T}:T{
First.  Do in field if possible, as most engineering seismographs do.
T}:T{
Incorrect\: geometry, observer's reports, time breaks, sample rate, record
length, etc.
T}:$10 sup 9$
.sp
Gain Recovery:T{
Multiply data by binary gain codes from gain ranging.
T}:Second, if needed.:T{
Locating correct gain trace.
T}:$10 sup 9$
.sp
Editing:T{
Remove bad records, misfired shots, open channels, noisy traces.
T}:T{
Third, and at other times during processing if needed.  Best done in field
during acquisition.
T}:T{
Must scrutinize plots of all of raw data.
T}:$10 sup 9$
.sp
T{
Summing (Vertical Stack)
T}:T{
Reduce source and random noise by adding multiple impacts, shots, or sweeps
at same location.
T}:T{
After editing.  Often done during acquisition, irreversible.
T}:T{
Noisy or unbalanced shots.  False triggers.  Strong 60 Hz noise will sum to
harmonics.  Large moveup arrays attenuate steep dips and blur statics.
T}:$10 sup 9$
.sp
Correlation:T{
Compress vibrator sweeps into small wavelets.
T}:T{
After summing to save computer time.  Best after despiking and editing.  Often
done during acquisition, irreversible.
T}:T{
Incorrect sweeps, harmonics, spikes produce ghosts.  Acts as a bandpass filter.
Very expensive.
T}:$10 sup 11$
.sp
Gain Function:T{
Remove effect of geometric spreading, amplifying deep events relative to
shallow.
T}:T{
Last step in data reduction above.
T}:T{
Can destroy true amplitude information.  Use a reversible function or save
unequalized dataset.
T}:$10 sup 7$
.sp
.T&
cbp10 s s s s
liw(1i) lw(1.6i) lw(1.5i) lw(1.7i) lw(0.8i).
Geometric Corrections
_
.sp
CMP Sort:T{
Arrange traces by common midpoint.
T}:T{
After data reduction but before velocity analysis or NMO correction.
T}:T{
Incorrect stacking diagram, crooked seismic lines.  Uses little CPU time but
very expensive in terms of storage media.  Gathers no longer correspond to
physical experiments.
T}:$10 sup 7$
.sp
T{
Elevation (Datum) Static
T}:T{
Time correction for elevation differences.
T}:T{
Correct to at least a CMP-variable datum before NMO or velocity analysis.  May
correct to final datum after stack.
T}:T{
Assumed velocities above datum, long offsets.
T}:$10 sup 4$
.sp
Uphole Static:T{
Time correction for lateral velocity variation in weathering layer.
T}:T{
Before NMO or velocity analysis.
T}:T{
Assumed depth of weathered layer; long offsets.
T}:$10 sup 4$
.sp
Velocity Analysis:T{
Estimate $V sub NMO$, $V sub interval$
T}:T{
After determined time corrections and sorting, before final NMO and stacking
or any migration.
T}:T{
Assumes zero dip, slow lateral velocity changes, strong reflectors at velocity
changes, no multiples.  Requires time-consuming human interaction.
T}:$10 sup 9$
.sp
NMO Correction:T{
Correct time on offset traces to zero-offset time.
T}:T{
After sorting and statics, before stacking, part of velocity analysis.
T}:T{
Assumes zero dip, slow lateral velocity changes, no multiples, short offsets.
T}:$10 sup 8$
.sp
Residual Static:T{
Correct any remaining time shifts to straighten out NMO-corrected events.
T}:T{
After NMO, before stacking.
T}:T{
Eliminates delay information useful for transmission tomography.  Assumes only
slow lateral velocity changes.  Needs human interaction.
T}:$10 sup 5$
.TE
.bp
.LP
.TS
center, tab(:);
l l l l l
cbp10 s s s s
liw(1i) lw(1.6i) lw(1.5i) lw(1.7i) lw(0.8i).
Process:Purpose:When Applied:Pitfalls:Cost (flops)
_
.sp
Data Enhancement
_
.sp
Mute:T{
Zero out arrivals that are not primary P-wave reflections.
T}:T{
Before stacking and/or migration.
T}:T{
Arrival variations with midpoint must be catalogued through human interaction.
Overly sharp clips cause artifacts in further processing.
T}:$10 sup 7$
.sp
Bandpass Filter:T{
Attenuate noise outside of reflection frequency band
T}:T{
Best before stack, NMO, or velocity analysis; can be after stack.
T}:T{
Often much noise in signal frequency band, or weak signals are filtered out.  Alters true and relative amplitudes
T}:$10 sup 8$
.sp
Notch Filter:T{
Attenuate noise in narrow frequency band, such as 60 Hz AC power
T}:T{
Best before stack, NMO, or velocity analysis; can be after stack.
T}:T{
Too narrow a notch will cause artifacts.  Destroys true amplitude and phase.
T}:$10 sup 8$
.sp
Deconvolve:T{
Compress source wavelet shape and duration, improve resolution, attenuate
reverberations.
T}:T{
Best before stack, NMO, or velocity analysis, can be after stack
T}:T{
Can unwittingly remove evidence of real reflectors; will change true amplitude
and phase.
T}:$10 sup 8$
.sp
2-D (F-K) Filter:T{
Spatial bandpass filter, attenuates or enhances arrivals based on dip, moveout,
or apparent velocity.
T}:T{
Anytime after data reduction, depending on type of events.
T}:T{
Alters amplitudes.  A good way to make data look like anything you might want,
biasing interpretations.
T}:$10 sup {7 - 11}$
.sp
Stack:T{
Mimic zero-offset section, attenuate random and much coherent noise.
T}:T{
After sorting, velocity analysis, muting.
T}:T{
Attenuates dipping structures, accentuates lateral coherence.  Depends on
inferred velocities.  Mislocates dipping structures.
T}:$10 sup 8$
T{
Trace Equalize (AGC)
T}:T{
Amplify weak events or traces relative to strong.  Often best used just for
display purposes.
T}:T{
Anytime, usually just before or after stack.
T}:T{
Lose amplitude information.  Can end up enhancing noise.
T}:$10 sup {5 - 7}$
.sp
.T&
cbp10 s s s s
liw(1i) lw(1.6i) lw(1.5i) lw(1.7i) lw(0.8i).
Imaging
_
.sp
Post-Stack Migrate:T{
Correctly position dipping events horizontally.
T}:T{
After stacking and usually equalization.
T}:T{
Depends on average and/or interval velocities.  Cannot improve on steep or
crossed dipping events that do not stack well.
T}:$10 sup 7$
.sp
Depth Conversion:T{
Correctly position events vertically.
T}:T{
After stack and usually migration.
T}:T{
Depth error proportional to average velocity error.
T}:$10 sup 4$
.sp
Pre-Stack Migrate:T{
Correctly position steeply-dipping and crossing reflectors.  Invert for earth
properties.  NMO correction and stacking are a simplified migration that
assumes zero dip.
T}:T{
Partial migration (dip moveout or DMO) can be done before NMO and stack.  Full
prestack migration done after data reduction and often after filtering,
equalization, and deconvolution; no stacking.  Usually applied only to good data
from well-characterized areas.
T}:T{
Heavily dependent on velocity estimates and susceptible to gross errors when
lateral velocity variations are not correctly accounted for.  Extremely
expensive.
T}:$10 sup 14$
.sp
T{
Database Formation
T}:T{
Organize 3-d seismic depth section to be interactively viewed by interpreter and
related to geology, well measurements, other geophysical data, etc. in on-line
environment.
T}:T{
Final result of processing.  Interpreter should be able to change velocities
based on supplemental data and see result on entire sequence of seismic
processing interactively.
T}:T{
Interpreter could force data to meet his prejudices.  Mis-ties and
inconsistencies between seismic and other data must be accounted for.
T}:$10 sup {5 - 14}$
.TE