NAME
myds - applies a suite of band-pass filters, optionally
performs a time shift on each filtered trace, normalizes
each resulting component to the model trace envelope, and
sums the results to produce high-resolution seismic reflec-
tion data without distorting amplitude relationships (i.e.,
for "true amplitude processing"). [The phase spectrum of
each filtered version may optionally have a static shift
applied before summing all the filtered, time-variant nor-
malized versions.]
SYNOPSIS
myds [ -Nntap ] [ -N2ntap2 ] [ -Ootap ] [ -O2otap2 ] [ -
Ccardin ] [ -V ] [ -? ]
DESCRIPTION
1. Operations and Options
myds builds the model trace envelope, applies a suite of
band-pass filters, normalizes each resulting component to
the model trace envelope, and sums the results to produce
high-resolution seismic reflection data without distorting
amplitude relationships. The five separate stages of the
MYDS technique are described below.
a. Building the Model Trace Envelope:
The energy envelope of the model trace is computed to
use in the time-variant normalization step. The model
trace can be specified in three different ways:
(1) The model trace is the input trace itself.
(2) The model trace is the band-limited version of the
input trace. In this mode, all the defining fre-
quencies on the 1FLTR cards are read in order to
determine the minimum low-cut and the maximum
high-cut frequency pairs. The model trace will be
band-limited by the filter whose defining frequen-
cies are f1,f2,f3,f4 where f1,f2 are the minimum
low-cut frequency pair and f3,f4 are the maximum
high-cut pair. The filter type used will be the
same as the type specified on the first 1FLTR
card.
(3) The model trace is the trace from an auxiliary
input data set. In this mode, the record and
trace numbers on the auxiliary data set must match
the record and trace numbers of the input data.
b. Filtering:
The seismic trace is broken into a series of narrow
frequency bands by applying a zero-phase Ross- or
Bessel-weighted filter for each set of design frequen-
cies. If there are 'N' filters, there are 'N' filtered
versions of each input trace. The input design fre-
quencies are the vertices of the trapezoid representing
the ideal amplitude response of each filter. The Ross-
and Bessel-weighted filters are created by applying
different weights to a particular function of the form
sin(x)/x (sinc function). These weights and the length
of the filters are determined by the reject level (in
dB) requested.
The Bessel-weighted filter is shorter than the Ross-
weighted filter for given design criteria to meet the
specified criteria (i.e., design frequencies and reject
level). However, if a filter design is not specified,
the Ross-weighted filter is used.
c. Time-Variant Normalization:
A smoothed envelope of the model trace is computed by
convolving the rectified trace with a symmetric,
triangular-shaped smoothing operator whose coefficients
range from 1.0 in the middle to 1.0/(Operator Length in
Samples/2+1) at either end. A similar envelope is con-
structed for each filtered (band-limited) version of
the input trace. Each of these filtered traces is then
normalized to the model trace by dividing each sample
value by the value of its own envelope at that time.
The length of the smoothing operator is tailored to fit
each frequency band. It is typically chosen to be ten
periods of the dominant frequency component. As the
dominant frequency increases from one filter to the
next, the operator length should decrease by the same
percentage. The minimum length, likewise, should be
ten periods of the maximum dominant frequency component
for any of the filters used.
d. Phase Adjustment
A time shift, in ms, is input for each filtered version
of the trace. Each element of the filtered trace is
then shifted towards the start of the trace by the
amount specified for that passband. If the shift
specified is negative, each element will be shifted
toward the end of the trace. This adjustment is a
means of removing phase from the input wavelet.
e. Compositing:
The output trace is created by summing the filtered,
phase-adjusted, time-variantly normalized versions of
the input trace.
2. Filter Design Modes
The filters used in MYDS are individually designed using the
Ross- or Bessel-weighting technique and applied to the data
using a Fourier transform algorithm. Since the filters are
applied in the frequency domain, much computer time can be
saved by computing one frequency-domain filter and shifting
this spectrum to produce the other desired filters.
The spectrum shifting can only be done if all filters have
the same passband and slope widths, in Hz. The filtering
operation in MYDS is much faster when the spectrum shifting
operation is used.
The drawback to this shifting algorithm is the error pro-
duced in shifting a filter specified on integer Hz boun-
daries (e.g., 10-15-20-25 Hz) when the spectrum resides in
the computer on the power-of-two Fourier transform frequen-
cies (e.g.,10.01-15.01-20.01-25.01 Hz). The displacement
error becomes most significant for 1-ms data and 1500-sample
trace lengths. For 2-ms or 4-ms data, the error is negligi-
ble.
Unless otherwise requested, the sliding spectrum method
(fast mode of MYDS) is always used (i.e., only the first
filter is input; all others are generated with the same
bandwidth and interlocking slopes). When requested, or when
a panel data set is to be generated, the individual filter
method is used.
3. Analysis Start TIme Adjustment
An optional analysis start time adjustment is available in
MYDS to control the beginning of the time-variant normaliza-
tion. The trace sample corresponding to the analysis start
time will be opposite the first element of the smoothing
operator for the first computed value of the smoothed
envelope. To use this option, you must define the start
time for zero-distance traces, the adjustment velocity, and
whether or not the start time is to be referenced to the
water bottom. The analysis start time is computed for each
trace as:
Analysis Start Time = t0 + t1 + t2
where
t0 = the start time for a zero distance trace;
t1 = the adjustment to be added to t0 for non-zero
distance traces, such that
t1 = (Trace Distance)/(Adjustment Velocity)
(t1 is set to zero if no adjustment velocity
is supplied); and
t2 = the adjustment to add to t0 for referencing
to the water bottom, such that
t2 = (WaterBottomDepth*2)/(Water Velocity)
(water bottom depths and water velocity must
be filed on the input data set prior to
executing MYDS).
4. Internally Generated Design Frequencies
The design frequencies for each filter may be input indivi-
dually or may be generated internally by one of two options.
Both of the modes allowing the design frequencies to be
internally generated require that the first set of frequency
points (f1, f2, f3 and f4) be input be the user.
Option 1 generates the second set of frequency points by
equating the new f1 and f2 to the previous f3 and f4,
respectively. A new f3 and f4 are then computed as follows:
new f3 = new f2 + (previous f3 - previous f2)
new f4 = new f3 + (previous f4 - previous f3)
In other words, f3 and f4 are computed such that the width
of the passband and slope of each filter's amplitude
response remain constant.
Option 2 generates the new f1 and f2 in the manner of Option
1. However, the new f3 and f4 are computed such that the
slope is the same but the width of the passband is doubled.
That is,
new F3 = new f2 + 2*(previous f3 - previous f2)
new f4 = new f3 + (previous f4 - previous f3)
In both options, the procedure of generating new points from
the previous set of points is continued until the user-
specified number of filters (maximum of 20) is attained.
5. Mute Preservation
Program MYDS automatically restores an early muted zone to
the traces immediately before output (mute zones are not
preserved on panel records). As each trace enters the pro-
gram, MYDS counts the number of continuous zero samples
starting from the first sample on the trace. After
filtering, normalizing and summing the component band-
limited traces, the number of zero samples at the start is
restored to the trace data. A 48-ms ramp is applied follow-
ing the mute zone to minimize transients.
NOTE: When a single spike or band-limited zero-phase
wavelet is filtered, the results may be surprising
because the mute preservation operation restores
zeros up to the location of the first non-zero
amplitude of the original input trace.
6. Filter Panel
If the filter panel option is requested, a filter panel data
set is created in addition to or instead of the normal MYDS
output data set. The total number of records on the filter
panel data set will be twice the number of filters 'N'. The
filter panel data set contains two different sets of fil-
tered records. The number of traces per record on the panel
data set is the same as the number of traces requested for
the analysis.
The order of filter application to the panel data is deter-
mined by requesting either "upsum", in which filters are
applied in the order input, or "downsum", in which filters
are applied in the reverse order of input.
The corner frequencies (f1, f2, f3, f4) for each panel are
stored in the trace header words normally reserved for
time-velocity pairs.
A description of the panel output follows:
a. First Set of Records:
The first set of records (Record 1 to Record N+1) con-
sists of the data with individual filters applied.
If the "upsum" option is requested, all traces for the
entire panel are first output unfiltered, with Filter 1
applied, with Filter 2 applied, ..., and finally with
Filter N applied.
If the "downsum" option is requested, all traces are
first output unfiltered, with Filter N applied, with
Filter N-1 applied, ..., and finally with Filter 1
applied.
The number of records output in the first set equals
the number of filters plus one.
b. Second Set of Records:
The second set of records (Record N+2 to Record 2*N) is
a summation filter panel. The "direction" of summation
may be selected as either "upsum" or "downsum".
If the "upsum" option is selected, all traces for the
entire panel are first output as the summation of
traces filtered with Filters 1 and 2 applied; with
Filters 1, 2, and 3 applied; ...; and finally with
Filters 1, 2, 3, ..., N-1, and N applied.
If the "downsum" option is selected, all traces are
first output as the summation of traces filtered with
Filters N and N-1 applied, with Filters N, N-1, and N-2
applied, ...; and finally with Filters N, N-1, N-2,
..., 2, and 1 applied.
The number of records output in this second set equals
the number of filters minus one.
In both sets of records, the filtered output traces will
have had the time-variant normalization applied.
Command line arguments
-N ntap
Enter the full path name of the file containing the
input data set. If not specified, input is expected to
be on the standard input (a pipe into myds). If no
input file is given, and there is no standard input,
the program will abend.
-N2 ntap2
Enter the full path name of the file containing the
auxiliary input data set containing the model traces.
If "Source of Model Trace" (cc 45 on 1MYDS card) is
equal to 2, this file must be specified; no default.
If no auxiliary input data set is used, omit this argu-
ment.
-O otap
Enter the full path name for the output file to receive
the output data. If no output file is specified, the
program will write the output data to the standard out-
put (a pipe out of myds).
-O2 otap2
Enter the full path name for the output file to receive
the filter panels. If no filter panels are requested,
or if filter panels ONLY are requested, this parameter
must be omitted. No default.
-C cardin
File from which processing parameters necessary for
executing myds are specified in card image format. An
error message will be issued if cardin does not exist.
The parameter cards required for myds are:
Card Image: data column entries
Column Variable
*******This card is REQUIRED********
**Only one 1MYDS/MYPL card allowed**
1-5 1MYDS or 1MYPL (required)
1MYDS = output standard data set and, if requested,
the optional filter panels data set.
1MYPL = output filter panels data set ONLY.
6-15 OPERATOR PARAMETERS
6-9 Length, in ms, of operator for computing the envelope
of the input data and the data after application
of first filter. Suggested value to enter is
ten times the dominant period (reciprocal of
dominant frequency.
Minimum = 2*Sample Interval (ms)
blank or 0 = 200.
10-15 REMAINING OPERATORS
10-12 Minimum Length, in ms, of operator to use for
calculating envelope on all data filtered
with Filters 2 through N. Suggested value
to use is ten times the minimum period
(reciprocal of maximum frequency).
Minimum = 2*Sample Interval (ms)
blank or 0 = 100.
13-15 Length of succeeding operators as a percentage of
length of each previous operator. Length
should decrease the same percentage that the
dominant frequency increases.
blank or 0 = 100 (same length as previous
operator).
16 Filter Parameters Code
blank or 0 = input all defining frequencies
on 1FLTR card(s).
1 = input defining frequencies for
first band-pass filter on a
1FLTR card; generate subsequent
frequency points as follows:
f1 = previous f3 frequency;
f2 = previous f4 frequency;
f3 = new f2 plus width, in Hz,
of passband of previous
filter; and
f4 = new f3 plus width, in Hz,
of upper slope of previous
filter.
This method can be used with
"Filter Design Flag"(cc 43) to
request the "fast mode" of MYDS
and improve processing time.
2 = Same as "1", except each passband
width is twice the passband width
of previous filter (slope widths
remain the same).
NOTE: Methods 1 and 2 can only be used with band-pass
filters.
17-18 Number of Filters - REQUIRED - total number of filters to
be designed. If all filters are to be user defined
(cc16 is blank or 0), entry must equal number of
1FLTR cards.
Maximum = 20.
20-26 FILTER PANEL
20-23 Start record number (RI) of first input record for the
filter panel.
for 1MYDS card:
blank or 0 = bypass filter panel option.
for 1MYPL card:
blank or 0 = first record on data set.
24-26 Number of traces to include in the filter panel
Maximum = 288; blank or 0 = 288.
27-36 ANALYSIS START TIME
27-30 T0: Start time, in ms, of first analysis window for
zero-distance traces.
blank or 0 = beginning of trace.
31-35 Velocity, in feet or metres per ms, for computing
adjustment value to add to start time of analysis
window for non-zero-distance traces.
Adjustment Value = trace distance/specified velocity.
Decimal value may be entered.
blank or 0 = no adjustment for trace distance.
36 Water Bottom Flag: adjust anaylsis start time using
water bottom depth?
Adjustment Value = [(WBD * 2)/water velocity].
blank or 0 = no.
1 = yes (water velocity and water bottom
depth values must be in appropriate
trace headers words on input data set).
38 Method to use to display filter application on the filter
panel records.
blank or 0 = Upsum (order of input).
1 = Downsum (reverse order of input).
43 Filter Design mode to use when design frequencies are to
be internally generatedl
blank or 0 = Fast Mode (sliding spectrum method).
Only valid if: "Filter Parameters Code"
(cc 16) is 1 and filter panels are NOT
requested; otherwise, default is slow
mode.
1 = Slow Mode (compute each filter)
45 Specify source of model trace.
blank or 0 = input trace.
1 = band-limited version of input trace.
2 = trace from auxiliary input data set.
Column Variable
*******This card is OPTIONAL********
**Maximum of 2 9STAT cards allowed**
**Statics on these cards used to perform a phase adjustment to
the individual filtered components prior to summing*********
1-5 9STAT (required)
6-10 Phase Adjustment Static Shift
11-15 Static shift, in ms, to apply each filtered version
16-20 of the input trace. The first statics shift (cc6-10)
21-25 applies to the trace data filtered with the first
26-30 filter, second (cc11-15) to the trace data filtered
31-35 with the second filter, etc. Entries on the first
36-40 9STAT card correspond to Filters 1 through 10;
41-45 entries on the second 9STAT card are associated
46-50 with Filters 11 through 20. These shifts can be
51-55 negative.
Column Variable
*******This card is REQUIRED*******
**One required; maximum is 20. If ALL defining frequencies are
to be user specified, the number of 1FLTR cards must equal
"Number of Filters" parameter (cc 17-18 on 1MYDS card).
Each 1FLTR card contains the parameters required to define
one filter***************************************************
1-5 1FLTR (required)
6-8 F1: first 0% (reject) amplitude response frequency in Hz.
blank = 0.
9-11 F2: required for band-pass and high-pass filters.
First 100% (pass) amplitude response frequency in Hz.
blank = 0.
12-14 F3: required for band-pass filter. Enter second 100%
amplitude response frequency in Hz.
blank or 0 = 0 for low-pass filter; otherwise
15-17 F4: required for band-pass and low-pass filters. Enter
second 0% amplitude response frequency in Hz.
blank or 0 = Nyquist, for high-pass filter.
19-21 Reject level, in dB, for the area outside the 0%
amplitude response defining frequencies.
Minimum = 23; maximum = 120; blank or 0 = 65.
23 Type of filter desired.
blank, 0 or 1 = Ross-weighted filter.
2 = Bessel-weighted filter.
-V Verbose mode.
Detailed information about the processing parameters is sent to the
output listing device.
-? Query mode. With this flag, myds will
send a description of the command line arguments to the standard
error output and stop.
If operating under the Berkeley shell (csh), the -? must be quoted
with single quotes, i.e., '-?'.
BUGS
No trap for running program without an input file and noth-
ing in the standard input.
AUTHOR
Paul Gutowski (1992) Marilyn Miller (1992)
BUGS
A command line option letter and its corresponding argument
may not be separated by whitespace.
COPYRIGHT
copyright 2001, Amoco Production Company
All Rights Reserved
an affiliate of BP America Inc.
Man(1) output converted with
man2html