NAME
eign - module used to suppress coherent noise from seismic
data using covariance analysis of the noise and eigendecom-
position.
SYNOPSIS
eign [ -Nntap ] [ -Ootap ] [ -Cparmfile ] [ -Llistfile ]
[ -Eevecfile ] [ -V ] [ -H ] [ -? ]
DESCRIPTION
eign can be used to remove a coherent noise pattern from
seismic data. A training region consisting of the noise to
be suppressed is specified by giving four vertices surround-
ing that convex region. The training region is then divided
into data vectors, the size of which are determined by the
user. A covariance analysis of the training zone produces
an average covariance matrix for the data vectors. The
eigenvalues and eigenvectors corresponding to this covari-
ance matrix are computed. Selecting the dominant eigen-
values and their corresponding eigenvectors, these are used
to encode the training region, rebuilding the dominant
covariance structure of that data. A target region is
specified and the dominant training region eigenvectors are
used to encode the target data. This step reconstructs that
part of the target data which has the same dominant covari-
ance structure as the training region. The encoded target
data is then subtracted from the original data. The result-
ing residual is the desired result.
In the production mode, eign will process selected input
records, producing one output record for each input record.
The training zone can be changed as a function of the input
record number. There is also a mode in which only one
specified input record is processed, producing five or six
output records corresponding to: 1) the training region,
2) the encoded training region, 3) the target region, 4) the
encoded target region, 5) the residual, and 6) the encoded
residual (if desired). This second mode is used initially
to establish processing parameter values for the later pro-
duction step. Provision is also made in this mode to write
the eigenvalues and eigenvectors determined from the train-
ing region to a disk file. This eigenstructure from a sin-
gle record can then be used in the production mode to pro-
cess any number of input records.
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. The program will give an
error message if the input file is read protected. If
no input file is given, and there is no standard input,
the program will hang. (default is standard input )
-O otap
Enter the full path name for the output file to receive
the output data. If not specified, the output data
will be written to the standard output. The program
will give an error message if the output is being writ-
ten to a directory which has write protection.
(default is standard output )
-C parmfile
File from which processing parameters necessary for
executing eign are specified. An error message will be
issued if parmfile does not exist. If -C is present on
the command line with no argument, parameters are
assumed to be in-line in the job script marked with the
flag -eign.crd. By using this method (described in the
STARTJOB section of this manual), the user is allowed
to specify the eign parameters in-line with the pro-
cessing sequence. This option is also achieved by
omitting the -C flag from the command line. The param-
eter cards required for eign are:
Card 1: data column entries
Column Variable
1-5 card number and name (1EIGN)
6-10 data block size, trace direction, (e.g., 6)
11-15 data block size, time direction, (e.g., 4)
Note: the product of data block size in trace
and time cannot exceed 25
16-20 number of samples skipped in trace direction between
adjacent elements of data vector, (usually 1)
21-25 number of samples skipped in time direction between
adjacent elements of data vector, (usually 1)
26-30 number of samples overlapped in trace direction between
adjacent data vectors, (recommend at least 1)
31-35 number of samples overlapped in time direction between
adjacent data vectors, (recommend at least 1)
36-40 number of dominant eigenvectors for encoding target data
(usually 1 to 3)
45 robust encoding, target region: bi-weight coefficient
0) no robust encoding
6) to 9) value of bi-weight coefficient
(default = 0)
46-50 number of dominant eigenvectors for encoding residual
(usually 0, which prevents this step)
55 robust encoding, residual data: bi-weight coefficient
0) no robust encoding
6) to 9) value of bi-weight coefficient
(default = 0)
60 random data block location
F) do not use random data block location
T) use random data block location to stagger
the starting position of first data block
along each time interval
(default = F)
Card 2: data column entries
Column Variable
1-5 card number and name (2EIGN)
10 processing mode:
0) multiple records, train on each record
1) multiple records, read training eigenstructure from
disk
2) process one record
(default = 0)
15 not currently used
20 output option:
0) if processing only one record, output all
test records (5 or 6, depending on whether
or not the encoded residual is computed)
1) output training data
2) output encoded training data
3) output target data
4) output encoded target data
5) output residual data
6) output encoded residual data
(default = 5)
25 embed option:
T) embed output training/target/residual data
within original input data
F) show output training/target residual data
in a field of zeros
(default = T)
26-30 initial record to process (default = first)
31-35 last record to process (default = last)
Card 3: data column entries
Specification of training region vertices:
Column Variable
1-5 card number and name (3EIGN)
11-15 vertex 1: trace number
16-20 vertex 1: time sample number
21-25 vertex 2: trace number
26-30 vertex 2: time sample number
31-35 vertex 3: trace number
36-40 vertex 3: time sample number
41-45 vertex 4: trace number
46-50 vertex 4: time sample number
51-55 last input data record number for which this training
region applies (default = 99999)
Card 4: data column entries
Specification of target region vertices:
Column Variable
1-5 card number and name (4EIGN)
11-15 vertex 1: trace number
16-20 vertex 1: time sample number
21-25 vertex 2: trace number
26-30 vertex 2: time sample number
31-35 vertex 3: trace number
36-40 vertex 3: time sample number
41-45 vertex 4: trace number
46-50 vertex 4: time sample number
Card 5: data column entries
The eigenvectors which are to be used to encode the
target data are specified on this card. A value of 1
refers to the order 1 eigenvector, which corresponds to
the largest eigenvalue. These values are entered in 10
adjacent fields, format I5, with the first field in
columns 6-10. The total number of entries here must
match the number given in columns 36-40 on the 1EIGN
card above. Each order number must be unique. Use as
many as these cards as necessary.
Column Variable
1-5 card number and name (5EIGN)
6-10 order of first eigenvector to use for encoding
11-15 order of second eigenvector to use for encoding
etc.
Card 6: data column entries
If the residual is to be encoded (a positive entry in
columns 46-50 on the 1EIGN card and output options = 0
or 6), this is the specification of which eigenvectors
are to be used to encode the residual data. A value of
1 refers to the order 1 eigenvector, which corresponds
to the largest eigenvalue. These values are entered in
10 adjacent fields, format I5, with the first field in
columns 6-10. The total number of entries here must
match the number given in columns 46-50 on the 1EIGN
card above. Each order number must be unique. Use as
many as these cards as necessary. If the residual is
not encoded, this card is not needed.
Column Variable
1-5 card number and name (6EIGN)
6-10 order of first eigenvector to use for encoding
11-15 order of second eigenvector to use for encoding
etc.
-L listfile
File to which the run time listing (i.e., sysout) for
this program is sent. If option -L is present on the
command line with no argument, the name of the file to
receive the run time information is generated automati-
cally as EIGN.XXX.YYY where XXX and YYY are derived
from the least significant digits of the process i.d.
(PID) and parent process i.d. (PPID) of this process.
If option -L is not specified on the command line, the
name of the file to receive the run time information is
generated automatically as EI.XXXXX.YYYYY where XXXXX
and YYYYY are derived from the process i.d. (PID) and
parent process i.d. (PPID) of this job. This is compa-
tible with the naming convention required by the
STARTJOB system, so to use that system the -L flag is
not used.
-E evecfile
If operating in the one record processing mode (vari-
able ipropt = 0 on card 2 above), presence of this
option indicates that the eigenvalues and eigenvectors
determined from the training region are to be written
to file evecfile. If processing multiple records with
option ipropt = 1 on card 2, the presence of option -E
indicates that the training eigenstructure is to be
read from disk file evecfile. In either case the
default file name is eign.vec if -E is present without
an argument. If -E is not included in the command
line, no eigenstructure read or write is done.
-V Verbose mode. Detailed information about processing is
sent to the output listing device.
-? or -H
Query mode. With this flag, eign will send a descrip-
tion 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
W. J. Done (1987)
COPYRIGHT
copyright 2001, Amoco Production Company
All Rights Reserved
an affiliate of BP America Inc.
Man(1) output converted with
man2html