[OpendTect_Users] Importing sparsely sampled horizon

[Arnaud Huck]

  Dear Marc, OpendTect users,

Yes you can grid your attributes in OpendTect, just like 2D and 3D 
horizons. Nevertheless if you wish to load these attributes as horizons 
then you will need to extend you survey range. The settings we have put 
during import are too strict and we are going to change the logic 
behind. Already now you can visualize and grid horizons that are 
vertically outside the survey Z range. Thus please extend the Z range 
for importing  the horizons, and you can later revert that change.

Let me add some words on gridding:

First of all your data must be along the seismic grid along inlines and 
crosslines. If you follow this criteria then the sparseness of your data 
is not a problem. For instance you cannot load a north-oriented grid of 
points if the seismic survey is not north-oriented too.

Attribute data such as residual source and receiver statics should be 
loaded as grid in the menu Survey --> Import --> Horizon --> Ascii --> 
Attribute 3D. This implies that you have at least one supporting 3D 
horizon to hold the attributes. You could create for instance an 
additional column if your input file with a constant TWT column.  If you 
have attributes and the supporting horizon is the same input file, you 
need to use the menu Survey --> Import --> Horizon --> Ascii --> Horizon 
3D that also allows you to import grids while importing the horizon.

I would advise not to grid the horizon and grids during import, but to 
do this in the scene, where you have more controls of the gridding 
process by visualizing the input and output grids. The following steps 
should be followed:

1- Make sure that the supporting horizon contains the output grid 
definition: The output small inline/crossline steps, and the desired 
ranges. If that is not the case, grid the supporting horizon to this 
geometry. A better alternative however is to scan the horizon during 
import, and change the volume sub-selection by extending the 
inline/crossline ranges, and lower the steps to the survey defaults.
Gridding should be considered as an interactive procedure, containing 
the following steps:

a- First you may want to fill only small holes using inverse distance 
weighting to the convex hull. The parameter size should be about the 
size of your holes, neither much smaller nor much larger.
A much better alternative available in the latest development release 
4.1.1 only is to perform a triangulation with again an interpolation 
limited to the maximum size of your holes.

b- Fill most of the remaining space with a triangulation using a much 
larger limit (2000 - 10000 meters gives better results than no limit). 
The output geometry may then be "to the full survey".

2- Grid your 3D attributes using the same gridding method as for the 
horizon. All values are allowed: negative only, positive only, both, 
integers only, ... Please note that for this last kind you will need to 
toggle on the option "Values are classification" in the gridding window.


Let me also add to finish that several gridding methods from GMT will 
become available in the next major stable release 4.2. This will be at 
least the methods "Continuous curvature" and "Nearest neighbour". This 
continuous curvature is  a very good alternative to the triangulation.

Please note also that you can combine grids using several horizon and 
mathematics attributes, to be output as a grid on any horizon. The 
output grid can be converted into an horizon using ascii export + import.

Best regards,
Arnaud.

-- Senior Geoscientist
-- dGB Earth Sciences
-- Nijverheidstraat 11-2, 7511 JM Enschede, The Netherlands
-- mailto: arnaud.huck at dgbes.com, http://www.dgbes.com
-- Tel: +31 53 4315155 , Fax: +31 53 4315104


On 09/21/2010 11:44 PM, Marc Langlois wrote:
> I am trying to import into opendtect sparsely sampled 3D data points as a horizon for QC purposes. The input data points are in a column-oriented text file with X, Y and time values. The Z values are in this case residual source and receiver statics, but I'd also like to use any other value, e.g. weathering statics, surface topography (converted to time), etc.
>
> I can import the text file and create a horizon but I can't get the "Grid" algorithm to create a new gridded (interpolated) horizon. Is this possible in OD? If so, what is the recommended procedure and gridding parameters to create a horizon from sparsely sampled points?
>
> The static values are in the range -40 to + 40 ms. Do the negative values cause problems when the survey Z range is from 0 - 6000 ms?
>
> Regards,
> Marc Langlois
>
> TBI, Calgary, Canada.
>
>
> _______________________________________________
> Users mailing list
> Users at opendtect.org
> http://lists.opendtect.org/mailman/listinfo/users
>