User Discussion

Go to the list of seismic processes. Go to SIOSEIS introduction.

QUESTION:

(1) The documentation on "segdin" states that the command looks for a file called "in" (or "IN", does case matter?) from which it reads a "tape unit number" after a tape is done to change tapes. Since there is only one tape drive then this number could always be "43" (for high density). Correct? Therefore, after sioseis reads the tape it will eject it (an option I think I will turn on). The user should then put in a new tape and then create a file called "in" with "9" in it. This is true for both of the directories, e.g. "/ldata/realtime/brute_stack" and "/ldata/realtime/reformat".

(2) The scripts "/reformat/copy" and "/brute_stack/bstack" will be running at the same time and each running separate segdin commands. How is this done with only one tape drive? I assume that the first job to see the tape or the file "in" (e.g. copy) will snatch the tape drive and the other job (e.g. bstack) will sit and wait in back ground. Then when tapecopy is done then the tape will go "OFFLINE" and be ejected. The user should then put the tape back in (without creating "in" in /reformat), at which point the other script will run (e.g. bstack). When bstack is done it will eject the tape and the user puts in a new data tape. Which ever directory has the "in" first will determine the job that starts up first. Thus the presents or absense of "in" controls when each actually starts working.

Is this correct?

ANSWER:

1) The filename can be "in" or "IN". Don't use "In" or "iN" though. 2) The contents of "in" should be 43. (FYI the tape drive senses the tape density on reads. e.g. The density setting is only used when writing a tape.)

"43". not "9" (unless 9 and 43 are the same drive and the density setting is ignored)

3) Make sure the reformat/copy is running in a different directory from the brute stack. Then the file "in" doesn't conflict. Each job then waits for it's own file "in". Pretty cute huh. Unix/Sioseis allow the tape drive to be shared! So the procedure goes something like:

Insert tape 1
Start copy, which ejects the tape when done.
Insert tape 1 again. Change windows or directories to the brute stack window.
Start brute stack
Go to another window on grampus and start the Atlantek. You now have 3 windows and 3 processes going. The copy window and stack window have active sioseis jobs, so you need 2 more windows; one for controling file in for the copy job and one for controlling file "in" for the stack job.
Put tape 1 away in box to send home. Insert tape 2. Change directories to the copy directory. Create file "in" with a "43"
When 3490 is ejected, reinsert. Change directories/windows to the stack directory. Create file "in" with a "43" in it.
When 3490 is ejected, put it in box to send home.
Insert tape 3. Change directories to the copy directory. Create file "in" with a "43"
When tape 3 is ejected, reinsert it. Change directories to the stack directory. Create file "in" with a "43"
When 3490 is ejected, put it in box to send home.
Insert tape 4.
etc. etc.

When a seismic line ends, terminate the stack job only with a "-1" in file in. Then control-c the plot job on grampus. Then you can continue to copy without doing a stack. Stack and copy are independent this way. Sometimes the stack may be screwed up and you want to start it again - you don't want to copy the tape again. Also, you want to copy all traces and not decimate in the copy, but you may want to use the short end of the streamer and decimate when doing the stack. I set heezen so that there were 5 windows. Using CDE, you could put one CDE layer for the copy and one for the stack. The copy layer would have two windows; one for the copy job and one that controls file "in". The stack CDE layer has 3 windows; 1 running the job, another controlling file "in", and another on grampus for the plot. I had a file called "inn" that had "43" in it in each the copy directory and the stack directory. Then the watch could just cp inn in or even !! The key is using different directories for the stack and the copy.

QUESTION:

Makes good sense to me. Is each shot, one file on tape?

ANSWER:

In SEG-D, each shot is terminated with a file mark. In the SIOSEIS parameters you use in the copy script, each 3490 tape is terminated by a file mark. As an aside, I think the file marks in the SEG-D format also cause the Sun utility "tcopy" to be VERY slow on Heezen. I heard that someone tcopy on heezen and each tape took the same amount of time to copy as it did to collect. SIOSEIS does the reformat in about 1/3 that time. I heard third hand that the person insisted on SEG-D because "SEG-Y drops some information contained in SEG-D". That's not true for the SIOSEIS script you are using because sioseis copies the SEG-D external header into a SEG-Y "trace 0", which is a concept Digicon and LDGO used on the Ewing back in the early '90s. Quite cute. SIOSEIS uses the LDEO clock for the shot time rather than the Syntron PC clock. The LDEO clock (known as Joe's clock), is the same clock that the navigation system uses and is a GPS clock. The streamer depth and compasses and gun information are in trace 0 for future use. but can be put into the SEG-Y header during the copy stage if you can describe the special bird locations in process geom. Turns out nobody really uses that information anyway, but you are retaining it in trace 0 just in case you every do need something. As another aside, there's talk of a national seismic data archive. This issue comes into play. You see where I stand. SEG-D and those file marks really are nasty especially for a disk file.

QUESTION:

One thing I'd like to clarify is that the only time I need to stop the tape copy job is when I am changing DLT tapes. Correct? Is there any need to stop the tape copy job between survey lines, other than to keep separate lines on different DLT tapes (which will be impractical because we are gonna have lots of lines?

QUESTION:

Here's another question. If/when I do try to copy more 3490's than the DLT tape can hold, sioseis will prompt me to change DLT tapes by asking me to create a file "out" that contains the output tape unit number. correct? If this happens is it better to: a) Ctrl-C and re-start the tape copy script with the same 3490 after I change DLT's or b) create the file "out" and let the tapecopy script do its thing or c) something else.. It seems like (a) would require me to always start with the first DLT tape if I want to extract data from the second tape, using "input". (b) would simply mean that I would have to specify the last complete shot to read in on the first tape. Yes?

ANSWER:

I find it much simpler to control-c the copy script at the end of a 3480 tape and restart the script with a fresh DLT tape. It's no big deal if the last DLT file is a partial SEG-Y file because I won't use the file anyway. Start the new DLT with the 3490 tape that didn't make it. SIOSEIS process INPUT has a file NFILES that allows you to specify the number of files on the DLT tape.

QUESTION:

How did you come up with the offset to the closest group?

ANSWER:

This takes a great deal of effort because everybody wants to get the first shot off after 6 hours of streamer deployment. You have to watch closely AND DO THE LEADER MEASUREMENT YOURSELF. Here's my calculation for EW9914:

147.5  fantail to center of dead section
-37.5  fantail to center of guns
-----
110.0
100.  stretch section
-----
210.0
 +6.25 center of dead section to end of dead section
 +6.25 end of live section to center of group
------
222.5

"shortened leader by 31.9 for lines 2-61"
222.5 - 31.9 = 190.6

QUESTION

I also found that "anntyp" works better than "ann" (which didn't allow me to control annotation increments).

ANSWER

I don't understand the anntyp vs ann problem. I thought/think ann converts the symbolic to a numeric for anntyp - therefore they're the same to the plot routine.

QUESTION

I tried the "ann header" with "hdr 19" or "lhdr 19" and got all zeros.

ANSWER

SEGDIN puts the EWING lat/long in header words 19 and 20, but process GEOM writes over the latitude with the X-coordinate. Then STACK zeroes word 19. There's a comment in stack's source code tha says: "THE SHOT X COORDINATE DOESN'T EXIST ON A STACKED TRACE". I don't recall my exact thinking.

QUESTION

Is there a way to annotate according to navigation?

ANSWER

No. The only way I can think of is annotate the stack section with the RP number and do some math. Or, since the stack section trace spacing is even, you could use fanno and ftag along with anning and taginc. I use GMT, because the watch writes lat/long down every .5 hours and I can go to the watch log to find the location. Graham likes to use the RP number. Each to his own.

QUESTION

The brute stack script is bombing with the error message "Not enough AP to perform filter".

QUESTION

Is there away to get annotations to print out on "atlantek" plots?

ANSWER

Thanks for attaching your script. I think both of these problems are due to your using the plot scalar from EW9914. I suspect the plot is writing outside it's program area and overwriting other program variables. You MUST use a new plot SCALAR. Remove the plot parameter from your script and plot will determmine a new one based on parameter def. PLOT will print the new scalar it calculates and you can use it on subsequent plot jobs if you want every plot to have the relative amplitudes.

QUESTION

How do I make sioseis tell me the scalar it determines?

ANSWER

Plot calculates the scalar when paramater scalar is less than zero. Plot uses the value given when paramater scalar is greater than zero. So, plot job 1 has scalar not given and the scalar is printed out. Cut and paste (or round a little) and use it as the value for parameter scalar in subsequent plot jobs.