3D geometry loading

3D geometry loading in ProMAX, a practical crPERL example

Kevin W. Hall, David C. Henley and Hanxing Lu
ABSTRACT
Last year, the radial component data from the Manitou Lake 3D-3C dataset were
processed with geometry copied from vertical component data (geometry in the trace
headers from the original processor), which have 2.3% fewer traces than the radial
component data. This was done in an attempt to expedite the process. Although the
difference in the total number of traces seems small, poor radial component results were
obtained, reflecting a mismatch of trace headers to trace data. To quickly and accurately
load geometry information into ProMAX trace headers from observers notes and shot log
files, a Perl script was developed for use with the CREWES Perl Process module. After
merging survey data and binning, a visual quality control method was used to spot any
problems. A comparison of selected results reaffirms that it is crucial to have geometry
correct, in order to obtain good final processing results.
INTRODUCTION
The Manitou Lake 3C-3D data was received as four separate SEG-Y files: vertical
component shot gathers with geometry in the trace headers, vertical, inline and crossline
shot gathers with no geometry in the trace headers. The vertical component shots with
geometry were processed first, before realizing that this dataset had 2.3% fewer traces
than the others, likely due to trace kills by the original processor. In the interests of
getting a result quickly, the geometry was copied to the radial component anyway. The
results were: 1) a consistent horizon could not be interpreted for hand-statics on the radial
component receiver stack, 2) imaging resulted in migrated sections with discontinuous
reflections, and 3) time slices through the migrated volume contained high-frequency
amplitude variations, resulting in a wormy appearance (Lu et al., 2006).
In an attempt to improve on the results presented last year, the vertical, inline and
crossline datasets were re-processed without reference to the vertical-with-geometry
dataset. To do this, we needed a way to quickly and accurately combine and load
information from multiple ASCII files on disk to ProMAX trace headers. It was decided
that the fastest way to do this was to utilize the CREWES Perl Process module (Bland,
2002).
Geometry loading
We have four sources of geometry information for this survey. The SEG-Y files have
field file ID (FFID) in the standard SEG-Y trace headers, as well as source line (S_LINE)
and source station (SOU_SLOC) in the optional SEG-Y trace headers. The observers
notes have FFID, Record Number (S_LINE*1000+SOU_SLOC; referred to as script in
the shot log file), ShotID, S_LINE, SOU_SLOC and max/min R_LINE and max/min
SRF_SLOC. Alone, the observers notes do not contain enough information to set
R_LINE and SRF_SLOC in the ProMAX trace headers, because the receiver patches did
not have a consistent number of traces from shot to shot, or from receiver line to receiver
line. Fortunately, we also have an I/O shot log report (Appendix A), which has all the

CREWES Research Report Volume 19 (2007)

Hall, Henley, and Lu

information we need except for FFID. Finally, we have a SEGP file with all the survey
information indexed to line and station numbers.
FFID and ShotID were saved as a CSV (comma separated values) file from the
observers notes. The ProMAX flow then consists of:

Disk Data Input (sorted by FFID, excluding bad shots)

Trace Header Math (create and initialize new trace headers)
Trace Header Math (fix existing trace headers based on comments in the
observers notes)
CREWES Perl process (runs perlcode.pl; Appendix B)
Extract Database Files (final run only)

Perlcode.pl (Appendix B) first reads the entire CSV file into an array. For each ensemble
from Disk Data Input, it finds the ShotID based on FFID, scans the shot log file until it
obtains a match for ShotID, reads all R_LINE and SRF_SLOC information from the
corresponding Active Seis Map Packet, and sets the ProMAX ShotID, R_LINE and
SRF_SLOC trace headers. Debugging consists of printing warnings to debug.csv when 1)
the ProMAX and shot log script numbers do not match and 2) when the number of
traces in the ProMAX ensemble does not match trace count in the shot log (Figure 1).
Note that this process does not require any editing of the shot log file.
This ProMAX flow was run iteratively until all issues were resolved. Most errors were
due to repeated shots, and could be avoided by editing the FFID list in Disk Data Input.
Finally, the SEGP survey information was merged into the ProMAX database using the
3D land geometry spreadsheet.
QUALITY CONTROL
Accuracy in matching geometry information to seismic traces is of primary
importance for the subsequent processing of the traces into useful images, especially for
3D surveys. As shown in the next section, seemingly small errors can lead to severe
imaging problems. For this reason, a quality control step applied to the raw seismic data
just after loading the geometry is highly recommended. Considering the sheer volume of
data handled by many processors, there will always be a temptation to either skip this
step, or minimize the time devoted to it.
Many processors have a favorite way of doing quality control on raw data; and
hopefully most of these methods are effective. One rudimentary way of doing QC on 3D
data, which displays all the traces in a survey, and yet allows the QC step to be done in a
reasonable period of time, is to simply read and display the entire data set a shot gather at
a time. The key to effectiveness in this procedure is to plot no more than 3 or 4 seconds
of data and to plot 3 or 4 key trace headers, in colour, along the top of the trace plot using
a tool like Trace Display in ProMAX. Figure 2 shows a display panel from the Manitou
Lake 3D seismic survey with source-receiver offset, source-receiver azimuth and receiver
line plotted above the seismic.

CREWES Research Report Volume 19 (2007)

3D geometry loading

Start Flow

FFID, ShotID

Observers Notes

FFID, script,
Ntraces

ProMAX

FFID == FFID
ShotID == ShotID
Found Stack

ShotLog

Y
Found ActiveSeis

FFID, ShotID, script, script,

Ntraces, Ntraces,
Ntraces-Ntraces, comment

ProMAX
R_LINE, SRF_SLOC

debug.csv
script != script
Ntraces != Ntraces

End of dataset

Y
End

FIG. 1. Flow chart for perlcode.pl (see Appendix B).

CREWES Research Report Volume 19 (2007)

Hall, Henley, and Lu

Source gather from Manitou Lake 3D survey showing three trace headers
plotted above the data. Shown are OFFSET, R_LINE, and SR_AZIM.
FIG. 2. Quality control plot of a shot gather from the Manitou Lake 3D survey. Three trace
headers are plotted here, Source-Receiver offset in black, Source-Receiver azimuth in blue and
Receiver Line (R_LINE) in red. Any irregularities in these header plots within receiver lines, and
any mismatches at receiver line boundaries indicate geometry problems in the data set that must
be corrected before further processing.

There are several things to look for on such a display:

Patterns of first arrivals should be smooth, with no large discontinuities in

them except at receiver line boundaries.

Plots of receiver line number should be flat (constant value) within each
receiver line and should show steps only at receiver line ends (red line; Figure
2).

Plots of source-receiver offset should be smooth curves within receiver lines,

unless the offsets have been modified to signed offset (by rad3d for
application of radial filtering), in which case, they may have discontinuities at
the nearest offsets where the sign changes (black line; Figure 2).

Plots of source-receiver azimuth should be smoothly varying within each

receiver line and discontinuous at receiver line boundaries (blue line; Figure
2).

CREWES Research Report Volume 19 (2007)

3D geometry loading

Plots of other headers, such as X or Y coordinates should be smoothly varying

within receiver lines.

The compelling feature of a display like that in Figure 2 is that, with a little practice, it
only takes a few seconds to quickly scan such a display and spot problems. Using
contrasting colors in the header plots is part of the key to this. With a little practice, a
moderately sized 3D survey (say 1000 shot gathers) can be scanned in an hour with a
high probability of spotting serious problems. Typically, a problem spotted on one gather
will be manifested on more than one gather; so if it isnt spotted on the first gather on
which it occurs, it will be seen on subsequent gathers. A problem which occurs only on a
single gather is unlikely to create a serious problem for imaging, even if missed by the
QC scan.
CONSEQUENCES
Shear statics are notoriously difficult to calculate. One way to deal with the problem is
to interpret horizons on a P-S receiver stack generated using P-shot statics and P-receiver
statics scaled by VP/VS. Time differences required to flatten this horizon can be used as an
additional correction term (hand-statics) for the S-receiver statics. Figure 3 shows a
comparison between the Manitou Lake P-S receiver stack with an incorrect geometry (as
discussed above), and one with a correct geometry. When the geometry is correct, the
reflections are higher frequency, noticeably more continuous, and therefore easier to
interpret.
An immediate consequence of improved S-receiver statics can be seen in time-slices
through the migrated volume at the target time. The time-slice through data with a correct
geometry (Figure 4b; Lu et al., 2007) looks more geologically reasonable than our initial
result (Figure 4a; Lu et al., 2006).

CREWES Research Report Volume 19 (2007)

Hall, Henley, and Lu

FIG. 3a. P-S receiver stack obtained using P-P shot statics and P-P receiver statics scaled by
VP/VS = 2.3. (Incorrect geometry; Lu et al., 2006).

FIG. 3b. P-S receiver stack obtained using P-P shot statics and P-P receiver statics scaled by
VP/VS = 2.3. (Correct geometry; Lu et al., 2007).

CREWES Research Report Volume 19 (2007)

3D geometry loading

a)

b)
FIG. 4. Time-slices near the target zone through migrated P-S volumes. Dashed line and circles
(white) show location of wells along a Colony member sand channel. a) Lu et al. (2006);
geometry copied from vertical component shots (2.3% fewer traces). b) Lu et al. (2007); geometry
loaded into ProMAX as described in this report.

CREWES Research Report Volume 19 (2007)

Hall, Henley, and Lu

DISCUSSION
A Perl script was developed to quickly and accurately load geometry information into
ProMAX trace headers from a variety of sources. After binning, a visual quality control
step was applied. Total development/quality control time was about one week. A
seemingly trivial difference in the number of traces between two SEG-Y files (2.3%) had
major consequences for image quality in the final migrated volume.
FUTURE WORK
The Perl script is fairly specific to files available for the Manitou Lake survey, but
should be easily modifiable as needed for other surveys. Because of the relatively small
size of the dataset, no attempt was made to optimize the Perl script for speed. For larger
datasets, a re-write may be in order.
ACKNOWLEDGEMENTS
We would like to Calroc Energy Inc. for providing this dataset to CREWES,
Landmark Graphics Corporation and Hampson-Russell (a CGG Veritas company) for the
use of donated software (ProMAX and GLI3D respectively). We also would like to thank
all CREWES sponsors and NSERC for funding.
REFERENCES
Bland, H.C., 2002, The fusion of ProMAX and Perl: CREWES Research Report, 14, 19 pages.
Lu, H., Hall, K.W., Stewart, R.R., Feuchtwanger, D., and Szatkowski, B., 2006, Searching for sand
reservoirs: Processing 3C-3D data from Manitou Lake, Saskatchewan: CREWES Research
Report, 18.
Lu. H., Hall. K.W., Stewart, R.R., 2007, Re-processing the 3C-3D seismic data from Manitou Lake,
Saskatchewan: CREWES Research Report, 19.

CREWES Research Report Volume 19 (2007)

CREWES Research Report Volume 19 (2007)

Acquisition Instrument Packet

Shot Status: ACQUIRED .

Script: 120114
Shot Point Line: 120.00
Line to line: 100 meters
Shot Point Stat: 114.00
Station to station: 50 meters
SPS-Index: 0
Clock TB(us): Early: 62
Process Type: Before Stack
Confirm TB(us): Late: 0

N E W E N T R Y

Shot Id: 1108163934

Ep: Stack: -1
Ser.Num: -1
Creator: Transcriber
Version: 0x2
Org.Ver: 0x1
Created: 02/12/05 02:36:15 GMT
Packets: 7
Shot Time: 11/08
16:76:96 GMT

================================================================================
I/O TRANSCRIBER 2
DETAILED SHOT LOG REPORT
PROJECT: manitou3d
Sat Feb 19 18:33:11 2005
================================================================================

Packets ignored by perlcode.pl have been deleted. Strings matched by Perl regular expressions are highlighted in blue. Data read
from the ShotLog file are highlighted in red.

APPENDIX A: EXCERPT FROM SHOTLOG FILE

3D geometry loading

10

Vibrator
-1
540
0
Sync
Normal
0
Bypassed
3.0
1ms

Radius Energy Gap:

Radius 12dB gain:
Radius 24dB gain:
Radius 36dB gain:
Anti-alias Freq:
TB Confirmation:
TB Window Left(us):
HPE Power Line:
HPE Fixauto:
Record Length:

0 meters
0 meters
0 meters
0 meters
1/2 Nyquist Minimum Phase
Yes
19937
60 Hz
Special fixed freqs
15.360 sec

Line
-----101
103
105
107
109
111
113
115
117
119

FirstSta
-------107
107
107
107
107
107
107
107
107
107

LastSta
------136
136
166
166
166
166
166
166
166
166

#Stations
--------30
30
60
60
60
60
60
60
60
60

#Traces
------90
90
180
180
180
180
180
180
180
180

Components
---------Av, Ax, Ay
Av, Ax, Ay
Av, Ax, Ay
Av, Ax, Ay
Av, Ax, Ay
Av, Ax, Ay
Av, Ax, Ay
Av, Ax, Ay
Av, Ax, Ay
Av, Ax, Ay

CREWES Research Report Volume 19 (2007)

StaInc
-----1
1
1
1
1
1
1
1
1
1

Active Seis Map Packet

Trace Count: 1620

Source Type:
Active Boxes:
Tot Act Channels:
Aux Channels:
TB Mode:
Record Type:
1st Scan Time(us):
Low Cut Filter:
Low Cut Freq:
Sample Period:

Hall, Henley, and Lu

$flag=0;
$success=0;

CREWES Research Report Volume 19 (2007)

@SCRIPT = gethdr('SCRIPT'); #New header created with Trace Header Math (S_LINE*1000+SOU_SLOC)
@R_LINE = gethdr('R_LINE'); #New header created with Trace Header Math (all zeros)
@SRF_SLOC = gethdr('SRF_SLOC'); #New header created with Trace Header Math (all zeros)

#Open ShotLog file (see Appendix A) for reading

open(IN,"<ShotLog_MANITOU-3D.txt") or die "Error: $!\n";

#Set ShotID header word.

for ($i = 0; $i<$#SHOT_ID+1; $i++){
$SHOT_ID[$i] = $ShotID[$FFID[0]];
}
puthdr('SHOT_ID',@SHOT_ID);

# Start of subroutine to deal with ProMAX ensemble trace headers

sub onEnsemble {
$comment="";
@FFID = gethdr('FFID'); #From in SEG-Y headers
@SHOT_ID = gethdr('SHOT_ID'); #New header created with Trace Header Math (all zeros).
print OUT "\n$FFID[0],$ShotID[$FFID[0]],"; #Print ProMAX FFID and Observer notes ShotID to debug.csv

#Open output file for debug information

open(OUT,">debug.csv") or die "Error: $!\n";
print OUT "FFID,ShotID,PromaxScript,ShotLogScript,PromaxNtr,ShotLogNtr,dNtr,comment";

#Read observers notes

open(FFID,"<manitou_shotid.csv") or die "Error: $\n";
while(<FFID>) {
/(\d+),(\d+)/;
$ShotID[$1] = $2;
}
close(FFID);

$| = 1;

APPENDIX B: PERLCODE.PL

11

3D geometry loading

12

CREWES Research Report Volume 19 (2007)

while(<IN>) {#loop over every line in ShotLog

chomp();
if(/Shot Id: (\d+)\s+Ep: (\w+):/) {#Find lines containing Shot ID:
if($ShotID[$FFID[0]] == $1) {
if($2 =~ /Stack/) {#and Ep: Stack:
$flag = 1;
} else {
$flag = 0;
}
}
}
if(/Script: (\d+)/ && $flag == 1){#Find lines containing Script: and read the script number
print OUT "$SCRIPT[0],$1,"; #Print ProMAX script and ShotLog script numbers to debug.csv
if($SCRIPT[0] == $1) {
$flag = 1;
} else {
$flag = 1; #Potentially fail by setting flag = 0. Desirable?
# Set $comment if the script numbers dont match and carry on
$comment = $comment."Warning: Script numbers do not match! ";
}
}
if(/Active Seis/ && $flag == 1){#Find lines containing Active Seis
$flag = 2;
$chan = 0;
$r_line = 0;
$srf_sloc =0;
}
if(/Trace Count: (\d+)/ && $flag == 2) {#Find lines containing Trace Count
$TraceCount = $1/3;
print OUT "$crProMAX::NTRACES,"; #Print number of traces in the ProMAX ensemble to debug.csv
print OUT "$TraceCount,"; #Print number of traces in the ShotLog trace count to debug.csv
print OUT $crProMAX::NTRACES-$TraceCount.","; #Print the difference to debug.csv
if($TraceCount != $crProMAX::NTRACES) {#Set $comment of number of traces does not match
$comment = $comment."Warning: Number of traces does not match! ";
#exit; #Potentially fail by uncommenting this line. Desirable?
}
}

Hall, Henley, and Lu

CREWES Research Report Volume 19 (2007)

#Print $comment to debug.csv. Should be blank unless script numbers and/or number of traces mismatch
#Overwrite ProMAX headers with info from ShotLog
print OUT $comment;
puthdr('R_LINE',@R_LINE);
puthdr('SRF_SLOC',@SRF_SLOC);
close(IN);

#Read receiver line, station and station increment information

if(/(\d+)\s+(\d+)\s+(\d+)\s+(\d+)\s+(\d+)\s+(\d+)\s+Av, Ax, Ay/ && $flag == 2) {
$rline = $1;
$FirstSta = $2;
$LastSta = $3;
$StaInc = $4;
$success = 1;
if($dostuff) {#set $dostuff to 0 to skip next for loop. Faster for debugging.
for($i=$FirstSta; $i<$LastSta+1; $i+=$StaInc){#Build trace header arrays
$R_LINE[$chan] = $rline;
$SRF_SLOC[$chan] = $i;
$chan++;
}
}
}
elsif ($success == 1 && $flag == 2) {#Done reading Active Seis Map Packet
$success = 0;
$flag=1;
next;
}
} # end of while(<in>)

13

3D geometry loading