FreeUSP Line001 processing

[NickCrew]

Hi Bill, Hi Conner,

My contract with TPIC (Turkish Petroleum International Company) is over. Now I can delve into the processing details of cpseis.

I've processed this line using SU. Now it is the time to process using Cpseissoftware.

I attach hereweith the icps file. Is this the preferred method to apply geometry? (The other is preparing jd file and apply using FGD)

I could not find the CDP numbers.(They must be from 1 to 1282. Which Header Word?) 

I plan to obtain the gather for velocity analysis. (SVA)  Then brute stack using cdpsort and velocity field.

Any suggestions?

Regards,

Tahir Helvaci

[NickCrew]

Sorted out:

Pls Use FGD instead of XPS / SPSRPS. cfg -> If it is a 2D line make sure the  FIXED_DIST parameter is set as non zero. (Group Interval)

CDP is in Header 37 and SP is in HW 29.

For GR elimination use MDIP. (Negative and negative Slowness parameters scale = -1)

Sort, Gather and SVA

I'll put complete procedure in Tips and Tricks section when finished processing of Line001 (FreeUSP Line).

Regards,

Tahir Helvaci

[Bill Menger]

Sounds good.  Thanks for working on this.  I just got back in country from 3 weeks in Indonesia and was not quite ready to answer you.

Bill

--
You received this message because you are subscribed to the Google Groups "cpseis-user" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cpseis-user...@googlegroups.com.
To post to this group, send email to cpsei...@googlegroups.com.
Visit this group at https://groups.google.com/group/cpseis-user.
For more options, visit https://groups.google.com/d/optout.

--

Bill Menger

Houston HPC Services, Inc.
(281) 633-6820

[Bill Menger]

A quick way to find out what header number is for which variable, you can click on HDRS on the CFE menu tab.

It will then pop up a window like this:

which will give you the meaning of each 64-bit header word.  CPS headers are double precision floating point, so they don't need to be scaled like SEGY headers.

Hope that is helpful.

Bill

[Conner Rodderick]

On Thursday, January 10, 2019 at 5:06:44 AM UTC+8, NickCrew wrote:

Sorted out:

Pls Use FGD instead of XPS / SPSRPS. cfg -> If it is a 2D line make sure the  FIXED_DIST parameter is set as non zero. (Group Interval)

CDP is in Header 37 and SP is in HW 29.

I had a look at that data. The first breaks change quite abruptly on near traces, indicating that there are large lateral offsets by the vibrators. 

I worked on land crews for a few years, and you only get perfect regular data if you were in a flat desert. The rest of the time going through farms, towns, beaches et cetera,  it gets messy. It is feasible that the geophones are planted within 10 metres of a straight line, but the source (vibrators or shot-holes) will deviate all over that place. If you come across really old data with paper observer logs, you will see scribbled comments like "vibs offset 200 ft north". For an excercise, make a near-trace gather, selecting field trace 141,142 and look at the range of header 6 (offset) with HEADSUM. Or load it into CBYT and plot header 6 at the top - you may learn how good are bad it is. So I think setting up geometry with CFG and FGD would not be accurate.

Try processing the line both ways and compare the stacks. I think the shallow reflections would suffer due to wrong offsets.

On the other hand, when processing 2D marine, defining geometry with MGD is okay because the cable feathering only makes a tiny difference to the offsets.

[NickCrew]

Hi Conner, Bill,

I have applied the geometry using FGD, TRIN (Initial skip 2 traces, 282 traces then 2 traces skipped), then applied Field statics using SHFT, then TPOW, then TROT (Allshots.trc)

Second step is to apply MDIP to get rid of Ground Roll. Do you have an example for setting dips (is it trace/msec ?) I don't want to apply FK filtering without the velocity field.

I have tried some dips but the cdpsort and the obtained semblance velocities are not good, nor the stack.

You are right as the shallow data is very poor due to the shooting geometry they selected. (Polish Company)

What I need is processing work files samples perhaps you have kept.

Regards,

Tahir Helvaci

[Conner Rodderick]

On Thursday, January 17, 2019 at 1:53:53 AM UTC+8, NickCrew wrote:

Hi Conner, Bill,

I have applied the geometry using FGD, TRIN (Initial skip 2 traces, 282 traces then 2 traces skipped), then applied Field statics using SHFT, then TPOW, then TROT (Allshots.trc)

Second step is to apply MDIP to get rid of Ground Roll. Do you have an example for setting dips (is it trace/msec ?) I don't want to apply FK filtering without the velocity field.

 To attenuate ground roll with process MDIP, I split the shots into left and
right halves. As a rule of thumb, one would use the slope of the first breaks
for what you wish to preserve, as the reflections hyperbolae would have dip
less than that. Here the first break dip is 12 ms per trace.
For the right half (traces 142-282), the ground roll below 1 second
has a dip of roughly 22 ms per trace. So I set the dips to preserve as
0 to 15 ms per traces, and the reject zone as 20 to 30 ms per trace.
I only looked at 10 shots along the line, maybe the ground roll apparent
dip does vary.
The parameters are in the attached screen grab. The dips are given in seconds
per trace, because HDR_HOR=4 and increments by 1. For the left half of the
shot, the dips would be the opposite.
Note I set TIM_BEG to 1, as I was trying to get rid of the ground roll
below 1 second. The shallower stuff looks harder to eradicate. I have not
sought to filter that. I have not tried fiddling with the CLIP parameter,
nor the NUM_WIN and NUM_INC.
It seems to work fairly well with AGC (process XP) or true amplitude.

I have not tried f-k filter on this, because I presume that spatial aliasing
will be a problem with 25 m receiver interval. If you come across some
modern data, with 1000 channels at 6 m interval, then f-k filter would
be a contender.

[NickCrew]

Thanks much Conner. 

I have run MDIP with success. Now the issue is how to run it using all the cores of CPU. (Created a new topic.) 

Next step: I ran sva job, created the velocity field, nmo, stack  but the stack section  is not as good as that of SU version. Shall I create a new topic or shall we continue on this topic?

How about the processing steps?

Regards,

Tahir Helvaci

[116E32S]

Does anybody know what happened to this dataset? The SEG Open Data webpage still lists it, but there is no download link.