[OpendTect_Users] cube spatial interpolation

[Mike Stone]

Emmanuel et al,

You know, I've been trying to use opendtect to interpolate stack data and
have come up with a workflow that works,
I've attached a attribute set that interpolates from by 4 in each direction.
I've been doing by 2 not by 4, *So this is not tested!*
*
*
The Key is to Null Fill the Data before you load it.

Arnaud's approach is better.


Here's the TL:DR version:
---------

What I'm trying to do is interpolate a increment 2 by increment 2 survey to
1x1

Basically I use the reference_stack attribute, The problem is you can't you
it on traces that aren't there, Shifting the data by 1 when it's an
increment 2 volume just shifts the data by 2.

So my solution is to use another software package(SMT) to null fill the
data, Basically I import it into my target survey, Then just export it out.
On load to Opendtect it's key that you pass null traces.

Then I just create a reference stack with inline +1 and another with Inline
-1 from the source.
Then I use the Mathmatics attribute to with the Equation *
Input?Input:(IL1+IL2)/2*
*
*
Then I do the same thing in the Xline direction.

Then just Process it to a stored volume, Unfortunately at 700 traces per
sec, Doing my whole volume(25GB>100GB) takes like 11 hours.
All thats left is exporting it and importing it where it's needed.

This method is fairly inefficient, But it works.
Doing 2x2 pretty much has to read the data 5 times
Doing 4x4 would have to read the data 25 times.



--------------------

I did try griding using the volume builder without success, Actually I tried
it twice.

The first time on my windows box, I got pretty close I had to set it as the
Default Velocity volume to start, But i got an error message saying I needed
100GB of memory.

So i went to my linux box that had 144GB of memory and tried it, Basically I
couldn't set the data type as other, And when I tried to grid it as Vint or
Vavg it gave me an error saying something like it wasn't a valid velocity
volume, If it's trying to depth convert using this, that might explain it.

I Guess I'll try this again with the beta version, this seems much more
effective.

-----

Just a quick note, It's really nice how well opendtect works dealing with
decimated volumes, It pretty much just opens them up and displays them
correctly. With SMT I have to setup a separate survey for every weird bin
spacing, So i have like 4 surveys based on where in the processing the data
is, The processor interpolated stack data in one, The gathers in another,
And the Inversion data. And then I have to worry about moving horizons back
and forth, It's a real headache.



Mike Stone
832-768-0884
MStone at ACT3Solutions.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opendtect.org/pipermail/users/attachments/20110209/7fc4709d/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Interpolate_By_4.attr
Type: application/octet-stream
Size: 1967 bytes
Desc: not available
URL: <http://lists.opendtect.org/pipermail/users/attachments/20110209/7fc4709d/attachment.obj>