NAME
opstf - create forward optical Tp scans and semblance
panels
SYNOPSIS
opstf [ -Nntap ] [ -Ootap ] [ -v0v0 ] [ -pmpmin ] [ -pxpmax
] [ -npnp ] [ -ststexp ] [ -rsnrst ] [ -renred ] [ -dmindmin
] [ -dmaxdmax ] [ -W ] [ -S ] [ -live ] [ -L ] [ -V ] [ -? ]
DESCRIPTION
opstf (OPtical STack Forward) creates optical stack panels
(Tp scans) and semblance panels, with optional range limit-
ing, from CDP-sorted data through the optical stacking pro-
cess described by E. de Bazalaire in the February, 1988
issue of Geophysics. This program is one of six programs in
a suite of programs for creating, processing, and/or analyz-
ing optical stack results. The other programs in the suite,
and their functions, are:
opstk - Extracts the stack and the velocity field from the
optical stack panels.
opstd - Demultiplexes the OPSTF output to create separate
optical stack and semblance panels for analysis.
opstcv - Resamples semblance panels created by program OPSTF
from constant Tp traces to constant (stacking) velocity
traces for analysis.
opstr - do the inverse of opstf
XOS - Provides graphical analysis of optical stack or
semblance panels created by OPSTF.
FLOWS
The optical stack tools may be assembled in a number of ways
depending on the goal of the user. In any event an analysis
step is very important, in particular the determination of
the range of Tp's necessary to adequately span the ranges of
moveouts in the data and the mute functions (either picked
curves or min & max velocities) required to isolate the pri-
maries. We assume that the analysis steps have been done
(usually through the interactive program opstws):
1. opstf --> opstk
This is probably the most important flow. Here the CMP-
sorted data are pushed through the forward Tp transform and
straight into the stacker where muting, if any, is done and
the output stacked section along with an optional velocity
field is generated.
2. opstf --> opstd --> opstcv
This flow uses the enhanced filtering capabilities of the
optical stack domain to allow better extraction of tradi-
tional continuous velocity analysis. Again muting is
allowed in the Tp domain to fairway the data around the zone
of primaries.
3. opstf --> [filtering] --> opstr
This flow allows both 1 & 2-d filtering to be done on the
optical stack transformed data in order to enhance primary
energy before transforming. This technique relies on the Tp
domain to gain better leverage over events. The filtered
data is then transformed back to X-T space for further pro-
cessing.
Command line arguments
-N ntap
Enter the input data set name or file immediately after
typing -N unless the input is from a pipe, in which
case the -N entry must be omitted. The input to the
Optical Stack procedure opstf is CDP-sorted data
(currently requires SIS/USP format) data which has NOT
been corrected for normal moveout but has been properly
corrected for recording gain and has had any other
desirable pre-stack processing (such as trace editting,
deconvolution, DMO) applied. First breaks should have
been muted from the data.
-O otap
Enter the output data set name or file immediately
after typing -O. This output file is not required when
piping the output to another process. The output data
set also requires the full path name (see above). The
output from program OPSTF contains optical stack and
semblance panels multiplexed in trace order. The out-
put may be input directly into program OPSTK for
extraction of the stacked data and stacking velocity
field or input to program XOS for analysis after pro-
cessing by program OPSTD, which demultiplexes the data
into separate optical stack panels and semblance
panels.
-v0 v0
The velocity, in ft/sec or m/sec, of the recording
(observing) or replacement medium. For marine data,
this is typically not greater than an approximation of
the local water velocity. For land data, this is typi-
cally not greater than an approximation of the near-
surface velocity in the area of the survey. Use of
velocities slower than these may be appropriate. Use
of faster velocities, however, is probably not
appropriate. Analysis of stacking results using vari-
ous velocities is strongly recommended. Default: None.
-pm pmin
Initial (minimum) Tp value, in seconds, used in the
creation of the stacked data and semblance panels. The
value assigned to this parameter determines the slowest
velocity scanned for in the data. Default: 0 sec.
-px pmax
Final (maximum) Tp value, in seconds, used in the
creation of the stacked data and semblance panels.
This parameter defines the fastest velocity scanned for
in the data and is typically large (e.g. > 50 seconds),
as can be seen from the description following.
Default: None. This parameter is required. It may be
computed from the fastest velocity expected in the data
(Vmax, in ft/sec or m/sec), the initial velocity V0
(described below) and the maximum trace time (Tmax, in
seconds) according to the equation Tp = (Vmax/V0)**2 *
Tmax For example, for a max trace time of 4 seconds, a
Vmax of 4000 m/sec, and a V0 of 1000, the maximum Tp
value would be 64 seconds.
-np np
Number of scans to be created. This parameter deter-
mines the increment between Tp values and the resolu-
tion of the stack and velocity calculations. The run-
time of the program is directly proportional to this
parameter, so it should be chosen carefully. The
"best" value to use for this parameter will vary with
the data being processed, so you will probably want to
try different values in the analysis mode before pro-
cessing the entire data set. Generally, a value which
results in a Tp increment from about 0.5 to 1.5 seconds
should be chosen. Default: None.
-rs nrst
Enter start record number. Default value is the first
record.
-re nred
Enter end record number. Default value is last record.
-st stexp
Power to which the stack divisor vector (number of live
samples being summed for each output sample) is raised
prior to normalization of the stack for each Tp scan.
Default: 0.7
-dmin dmin
Minimum offset (in ft or m) used to construct the scan
and semblance data. Default: Near
-dmin dmin
Maximum offset (in ft or m) used to construct the scan
and semblance data. Default: Far
-W Enter the "-W" parameter to cause the Tp scan data to
multiplied by the corresponding semblance data on out-
put. This option should probably NOT be applied in
this program, since it is also available in both the
stack extraction program OPSTK and the demultiplex pro-
gram OPSTD and no program option to remove the weight-
ing exists. Default: No.
-S Enter the "-S" parameter to cause the trace header
statics to be applied to the CDP data when computing
the Tp scan data. The method used for statics appli-
caion is described under PROGRAM EXPLANATION (Opera-
tion). Default: No.
-live
If present, normalize stack by the number of live
traces in the stack rather than the (default) normali-
zation by number of non-zero samples in each stacked
sample. This normalization method is recommended for
data which will be inverse transformed in opstr.
-V Enter the command line argument '-V' to get additional
printout.
-? Enter the command line argument '-?' to get online
help. The program terminates after the help screen is
printed.
DISCUSSION
The optical stack procedure of de Bazelaire (1988) is a
fast and efficient method for automatically extracting
stacked data and stacking velocities from CDP-sorted seismic
data. The stacked data and velocity information are
extracted from panels containing sums (stacks) of the CDP
data which, when viewed graphically, are not dissimilar to
the spectra computed for conventional velocity analysis. In
either the conventional case or the optical stack case, the
CDP data are summed along various hyperbolas. The optical
stack technique differs from the conventional velocity
analysis technique both in the definition of and application
of corrections for normal moveout. It is these differences
which allow the method to be faster and more efficient than
the conventional technique.
The optical stack technique is based upon a reformulation of
the normal moveout equation in terms of geometrical optics,
from which the name optical stack is derived. In this for-
mulation, the conventional (Dix) equation
T**2 = T0**2 + (X/V)**2 (1)
(where T is the two-way travel time for offset X, T0 is
the zero- offset time, and V is the velocity at time T0, and
the notation **2 means square) is rewritten as (the optical
stack equation)
(T + Tr )**2 = Tp**2 + (X/V0)**2 (2)
where Tr is a delay time of the apex of the hyperbola rela-
tive to T0 (i.e, the time difference between the "true" sub-
surface reflection point and its image in the constant velo-
city medium), Tp is the total zero-offset time, (T0 + Tr),
V0 is the velocity of the input or recording/observing
medium, and the notation "**2" means squared. Fermat's
Principle from geometrical optics theory is invoked to
stipulate that ray paths described by these equations are
equivalent, as long as stigmatism (focusing) exists.
Since the ray paths described by the equations (1) and (2)
above are equivalent, the normal moveout hyperbola for the
two points must be the same. By equating the derivatives of
the Dix equation and the optical stack equation at any
zero-offset time T0, the relation between V, the Dix stack-
ing (RMS) velocity, and V0, the initial velocity, is found
to be approximately
V = V0 * sqrt(Tp/T0), (3)
where sqrt signifies square root. This approximation breaks
down for small T0, but since there is typically little (spa-
tial) sampling of the data where the equation breaks down,
the approximation is generally valid.
Operations
NMO correction and Tp scan panel creation.
Both the Dix (Equation 1) and optical stack (Equation 2)
equations describe a family of hyperbolas. In the case of
Dix's equation, the hyperbolas vary as a function of both V
and T0. In the case of the optical stack equation, the
hyperbolas vary as a function of Tp only, since the velocity
V0 is constant. To correct for hyperbolic moveout using
Dix's equation, compute-intensive time-variant interpolation
techniques must be employed. To correct for hyperbolic
moveout using the optical stack equation, only a static
shift of the trace is required. To see that this is true,
we make the substitution
Tr = Tp - T0
in Equation 2 to find
(T - T0 + Tp)**2 = Tp**2 + (X/V0)**2.
Since T - T0 is just the moveout, dT, it is easy to see that
for each X the movout is constant (a static shift), given by
dT = sqrt(Tp**2 + (X/V0)**2) - Tp.
To create a Tp scan trace, all traces in the CDP are shifted
according to the above equation and stacked. A panel of Tp
scan traces is formed by repeating this shifting and summing
procedure for the number of Tp's you define. The full set
of Tp panels is created by repeating this entire procedure
for all CDP gathers input.
Moveout correction by static shift is not as computationally
intensive as time-variant interpolation, so is faster to
apply. It also does not result in stretching of the far
offset traces, since all samples on a trace are shifted by
the same amount. Additionally, with the optical stack
equation it is a simple matter to correct for inverted
(upward curvature) hyperbolas, which can occur for large
velocity contrasts at a concave (synclinal) interface.
Note that for such an hyperbola, Tr becomes negative.
Parameter Analysis
Application of the optical stack process to large amounts of
data, though faster than conventional velocity analysis, can
be very costly, in terms of machine time. It is, therefore,
strongly recommended that parameters be analyzed for effi-
ciency using a few selected CDP gathers s before processing
the entire data set. Generally, once parameters are
selected for a data set in an area, the same parameters can
be used for all other data sets in the same area.
Semblance and Tp Scan Panels
The results of the application of moveout correction and
stacking for a single CDP gather for several Tp's produces
a Tp scan panel which is similar to the well known velocity
spectrum, except that it is composed of actual stacked data
samples. Semblance can be computed for the scan panels pro-
duce a semblance panel which very much resembles the conven-
tional velocity spectrum. Both of these panels are output
from the program for subsequent analysis to determine the
effectiveness of the parameters chosen (See documentation
for programs OPSTD and XOS for description of this
analysis).
Tp Increment
Adequate sampling of the family of hyperbolas described by
Equation (2) is critical to the success of the optical
stacking process. For this reason, the First and Last Tp
parameters and the Number of Tp's parameter should be chosen
carefully. Use the analysis program XOS to verify that your
choice of parameters are adequate before processing the
entire data set.
Semblance Weighting
Multiplying (weighting) the Tp scan data by the correspond-
ing semblance data has the effect of accentuating the larger
stacked amplitudes and improving the overall signal-to-noise
ratio of the extracted stack or the Tp scan panels used for
parameter analysis. The effects of this weighting should be
carefully examined before choosing it for application to the
entire data set.
Statics Application
Statics application in program OPSTF is a two-phase process.
The statics value in the headers for each input CDP trace
are extracted and summed to compute an average static for
the entire CDP gather. The difference between this average
static and each original static is applied to each input
trace before computing the Tp scan data. The average static
is then applied to the entire Tp scan data. This is
equivalent to the application of a "residual" static before
NMO and the application of a "bulk" static after NMO.
Semblance Computation
The semblance computed and output by program OPSTF is
"point-wise" semblance, which means that a semblance value
is computed for each (summed) output sample. This semblance
is computed according to the conventional equation
s = (1/N)*(SUM(x)**2)/SUM(x**2),
where s is the sample semblance, x is the shifted CDP data
being summed, N is the number of non-zero values summed, the
notation "SUM" means summation, and the notation "**2" means
squared.
REFERENCES
de Bazelarie, E., 1988, Normal moveout revisited:
Inhomogeneous media and curved interfaces, Geophysics, Vol.
53, 143-157.
Arnold, Richard H. and Semaan, Mars E., 1990, Implementation
of the Optical Stack Method, SEG Expanded Abstracts, Vol II,
San Francisco.
BUGS
unknown
SEE ALSO
opstr, opstk, opstd, opstcv
AUTHOR
Richard Crider, ES&S
COPYRIGHT
copyright 2001, Amoco Production Company
All Rights Reserved
an affiliate of BP America Inc.
Man(1) output converted with
man2html