NAME
groupv - perform multiple filter analysis of each trace
SYNOPSIS
groupv [ -Nntap ] [ -rere ] [ -rsrs ] [ -nsns ] [ -nene ] [
-flin ] [ -vlin ] [ -vminvmin ] [ -vmaxvmax ] [ -V ] [ -o ]
[ -tmintmin ] [ -tmaxtmax ] [ -npernper ] [ -aaaa ] [ -bbbb
] [ -cccc ] [ -AAalpha0 ] [ -BBalpha1 ] [ -plt ] [ -? ] >
[plot file name]
DESCRIPTION
groupv takes a single trace and performs multiple filter
analysis on the trace. Each trace is Fourier transformed,
narrow bandpass filtered in the frequency domain using a
Gaussian filter, and inverse transformed. The Maxima of the
filtered traces correspond to the group velocity arrivals
and the corresponding peak amplitudes are proportional to
the Fourier amplitude spectrum of each mode. The Gaussian
filter is given by
H(f,f0) = exp[ -alpha(f0) [ (f/f0) -1 ]**2 ]
where the filter parameter alpha is given by
alpha(f0) = alpha0 + alpha1 * f0 * f0
If the input time series is g(t,r) and its Fourier transform
is G(f,r), and G(f,r) = A(f,r) * exp( - j omega r / c(f) ),
then
the filtered signal is
g'(t,r) = f0 sqrt(pi/alpha) A(f)exp[j omega(t - r/c)]
exp[ - (t - r/U)**2 * pi*pi*f0**2 / alpha ]
where c is the phase velocity , f0 is the filter center fre-
quency, omega = 2 pi f0, and U is the group velocity at the
frequency f0. (Herrmann, BSSA, 1973, 663-671, among others)
The parameter alpha governs the filter bandwidth in the fre-
quency domain and in the filtered time trace. The reason for
a frequency dependent version of alpha, is that some
authors, being interested in group velocity dispersion pri-
marily, have noted that a form alpha ~ f0**2 yields uniform
resolution in the time domain. For exploration data, alpha0
=0.0 , alpha1 = 5.0 works well for 1 -20 Hz, but is very
poor at higher frequencies.
The output consists of two options: plots, if -plt is set,
and an optional listing of spectral amplitudes, if -o is
set. The standard output is used for the plot stream, and
the file GROUPV.o is used for the spectral amplitude-groupv
velocity summary. One of the plots is of spectral amplitude
versus frequency, while the other is of group velocity
versus frequency. Logarithmic or linear velocity or fre-
quency axes may be specified.
To display the plots simply enter
plot_xview < [plot file name]
Then use the right mouse button inside the plot window and
choose next page to get the first plot and for every other
plot.
Command line arguments
-N ntap
Enter the full path of the file continuing the data
set. This input can be a pipe, i.e. stdin
-plt If set a plot file is generated, which is sent to the
standard output for use by the plot_xview
-o Enable output of dispersion numbers into file called
GROUPV.o. Summary results of the groupv processing are
provided. For each filter center frequency, the fil-
tered traces are searched for maxima, the frequencies,
source receiver distance, groupv velocity and spectral
amplitude are listed.
-rs rs
(integer) Give the first record for processing (default
= 1)
-re re
(integer) Give the last record for processing (default
= last)
-ns ns
(integer) Define the starting trace for analysis
(default = 1)
-ne ne
(integer) Define the last trace for analysis in each
record (default = ns) The use of the ns and ne pair is
useful to be able to skip dead traces or to select just
a portion of the data set for analysis
-flin
Flag to yield a linear frequency scale for the contur
plot (default = false) which is logarithmic of 2
cycles.
-vlin
Flag to yield a linear velocity scale for the contur
plot (default = false) which is logarithmic scale of 2
cycle length.
-V Flag for verbose output on the standard output. This
lists salient trace header information as well as all
command values used for processing, the default ones
given if not overridden.
-vmin vmin
Minimum velocity value for contour plot (default =
100.0)
-vmax vmax
Maximum velocity value for contour plot. (default =
10,000)
-AA alpha0
Parameter to specify alpha0 (default = 50.27 =
16*3.1415927)
-BB alpha1
Parameter to specify alpha1 (default = 0.0)
-aa aa
-bb bb
-cc cc
These parameters control the selection of filter center
frequencies according to
If cc.eq.0.0 centfreq = aa + bb*I I=1,nper
cc.ne.0.0 centfreq = 10.0**[aa+bb*I] I=1,nper
(default aa = 1.5, bb=0.5, cc = 0.0)
-nper nper
Maximum number of center frequencies to process
(default = 80)
-tmin tmin
Starting time in milliseconds (default = 0, beginning
of trace)
-tmax tmax
End time in each trace for analysis in milliseconds
(default = end of trace)
-dmul dmul
Multiply each distance, itr(117), by this factor
(default=1.0). This is useful when the header has
different units than expected.
-tmul tmul
Multiply the sampling interval, dt, by this factor
(default = 1.0). This is useful if the sampling inter-
val is in different units than expected.
-? Query mode. With this flag, groupv will give a
description of the command line arguments and stop the
program.
BUGS
The hard limits in the program are 8192 point maximum trace
length for the FFT. The FFT is computed for powers of two
with zero fill where necessary. There is no tapering or
smoothing at the endpoints of the trace. In addition, there
is a hard limit of 80 periods.
Some checking is performed of the input to prevent incon-
sistencies, a little work would make it more fool proof.
The program was written expecting units of ft and mil-
liseconds in the SIS header. Thus the plots are labeled in
units of Hz and ft/sec. For well logging work, if the data
are placed into a SIS format with the sampling interval in
microseconds occupying the millisecond NSI place and if the
distances are given in millimeters occupying the feet posi-
tion, then the computed velocities will be in meters/second
and the frequencies will be in KHz. The trace labels will
thus be incorrect. Perhaps there is a well defined flag in
the SIS format to be used.
One non-standard feature is used for use with reduced travel
time representation of the time series. The itr(121) is
expected to be the offset of the first data point, e.g., the
first sample has a travel time of itr(121)*dt seconds.
itr(121) is set by the program vred.
SEE ALSO
calplot(1), choose(n)
AUTHOR
R. B. Herrmann, Saint Louis University, July 1985.
COPYRIGHT
copyright 2001, Amoco Production Company
All Rights Reserved
an affiliate of BP America Inc.
Man(1) output converted with
man2html