Return to SIOSEIS examples      Go to the list of SIOSEIS processes.      Go to SIOSEIS introduction.
SIOSEIS was used aboard the R/V Ewing during the EW0207 cruise.
The near realtime tape copying and brute stack with display did
not differ much from previous cruises with the exception of a
catastrophic failure of the processing computer which forced us
to use the old Sparcstation 10 "heezen".  Heezen is unable to
drive the Atlantek plotter, so Sun rasterfile plot files were
generated also in the brute stacks.

This document will describe the efforts to produce the brute stacks
in true realtime - as the seismic data are collected!

LDEO installed a data "splitter" on the Syntron MCS data recorder.
One branch of the spltter goes to the 3490E tape drives and one branch
goes to a PC.  The PC has some software called Seisnet that captures
each shot from the splitter and writes it to various disk drives.
Unfortunately, Seisnet "encapsulates" the SEG-D General Header
by writing it's own header, in PC byte order.  Seisnet also adds
a trailer to each SEG-D trace.

On EW0207, the Seisnet files where written over the net to a small
publically accessible disk directory on the LDEO computer grampus.
grampus:/export/home/public/seisnet_tmp  was used until computer
grampus died and the seisnet operation ceased.
There were several different schemes in keeping the seisnet_tmp
directory with current data without filling the disk.  The most
successful, from sioseis' view, was when several shots were kept
and the oldest deleted.  This scheme enabled sioseis to view the
directory as a push down stack; new shots were added to the top
of the stack and shots at the bottom of the stack were removed.

Sioseis process SEGDDIN was written with parameters FORMAT SEISNET
and STACK FILENAME to account for this situation.  FORMAT SEISNET
strips the Seisnet headers and trailers and STACK NAME is used to
read the SEG-D file stack.  SEGDDIN always uses the second file in
the stack because the write of the first file may not be completed.
A script "set_latest" was built to continuously examine the seisnet
shot directory and list the last two shots in the directory,

#!/bin/csh -f
#  script set_latest lists the last two files in directory
#         /export/home/public/seisnet_tmp
#  The list is written to file "latest" in the current directory
#
set forever = 1
while( $forever )
ls -t /export/home/public/seisnet_tmp | head -n 2 > /tmp/latest
   sed '1,2 s^^/export/home/public/seisnet_tmp/^;w latest' /tmp/latest > /dev/null
   sleep 5
end
end

Script "set_latest" can be running continuously throughout shooting.

The EW0207 test used a directory structure with several directories
for multiple simultaneous sioseis jobs.  (Sioseis creates temporary
files for each job and these can interfere with other so each sioseis
job should run in a different directory).  Directory   "log"  was
used for the "set_latest" script and the realtime file name stack 
"latest".   Directory "brute" contained a script that
continually read and processed the shots through CMP stack and raster
plot.  The Atlantek thermal raster plotter allows the plot of the
stack to be plotted as the data are stacked!

File "in" should be use to stop the stack.

There were a couple of glitches with the data being fed to sioseis.
Over the 3-4 days running realtime, the seisnet computer (a windows
pc) disk filled up and no data were sent out.  Another glitch occurred
when the Spectra operator reset the shots numbers.

The "set_latest" script ran without problems.  Likewise with the
sioseis script just doing segddin prout to make a log file.  The
brute stack script continued to run during these data outages.  The
disk full problem was rectified within 15 minutes so there was only
a small gap in the stack.  The shot number reset did the stack in
though because type 2 geometry was used.  Type 2 assumes that 
shot numbers are strictly monotonically increasing.  One solution
would be to use type 9 geometry that uses the realtime navigation
from the data headers.

Spectra looses communications with Syntron, so the first few shots
don't have the "nav block".  The sioseis copy script will print out
a bunch of messages that can be ignored.  The brute stack needs
some of the information from the nav block, so start the brute
script after the third shot.

If doing brute stack in realtime, start the sioseis script after a
few (say 5) shots have been fired on the new line.  If doing the brute
stack from 3490 tapes, use the segdin parameter nfskip 3.

The Spectra/Syntron communication problem also causes the shot number
of the last line to be the first shot number of the new line when
using seisnet.  This isn't a problem with the sioseis realtime setup
because it ignores the first shot in the seisnet pushdown stack.

Lessons Learned:
1)  Have a single user, with a known passwd, run all jobs including the
screen.  Otherwise permissions becomes tedious.
2)  Use nfskip 3 in segdin on brute stacks from tape.
3)  Do not fill DLT tapes, rather always copy the same number of 3490s.
This simplifies the bookkeeping and the reading back of the DLT tape.


Miscellaneous info:

1)  The old Digicourse streamer string is no longer in the SEG-D
external header which contains the LDEO block.  The information must
be being feed to Spectra however since Spectra is getting compass
and streamer depth.  The only place streamer depth is now available
is in the UKOAA files.

2)  The Gun Depth info in the udp "dg" record is all 0s.  The compasses
appear ok in the udp records.  Streamer depth is not a udp record
(compasses were added by Jeff Turmelle for streamer navigation in ARAD).

3)  The Spectra UKOOA P1/90 files only had 240 channels.  JD used
Sprint to make 480 channel files.  Apparently Sprint is needed anyway
because the Spectra files have too many problems.  

4)  Some navigation analysis (realtime nav block vs ts.n files).

5)  For the Macs:
Date: Tue, 30 Jul 2002 07:23:51 -0400
From: Robert Detrick 
To: henkart@ewing
Subject: Plotting Sun raster sections on Mac
Parts/Attachments:
   1   Shown   ~46 lines  Text
   2.1          75 bytes  Application
   2.2          22 KB     Application
----------------------------------------

Paul,

FYI for Baja:

To set up a PowerBook G4 running OSX for plotting on the Ewing, first 
open the Print Center (in Utilities folder) and choose "Set Up New 
Printer".  Select HP DesignJet 755CM and for a PPD select "Other . . ." 
then locate the attached file HP_Design Jet .  This PPD file has been 
hacked so that if you choose 11x17 in Page Setup you will actually get
a 36"x~24" plot.

Next ftp the *.ras (Sun raster file) to the Mac.  Open with 
GraphicConverter.  Use "zoom" under Picture menu to shrink the plot to 
an appropriate scale (e.g. 12.5-25%) then using "rotate" under Picture 
menu perform a 900 clockwise rotation.  Now open Page Setup and choose 
HP DesignJet 755CM under "format for" option, choose 11x17 for paper 
size, and leave in Portrait mode.

Before printing, select "show information" under Picture menu.  Set 
units for inches.  Now go into "size" under picture menu and set width 
for 36".  The correct dimensions of plot now display in Information box 
(this step may not be required but if works for me).  Now open the print
dialogue box in the File menu and after being sure you have selected HP 
DesignJet 755CM for your printer, print.

It is also possible to print out page-size plots.  Just follow the same 
instructions as above, but in "size" set width to 10.5 in.  Use the  HP 
LaserJet 1200, in this case in portrait mode.

Good luck!

Bob


Return to SIOSEIS examples      Go to the list of SIOSEIS processes.      Go to SIOSEIS introduction.