Next: CONCLUSIONS
Up: Mora, Clapp & Biondi:
Previous: Building the 3-D Grid
Several additions are still necessary for a complete link between GOCAD and
AVS:
- Completion of the Edit function.
- Updating to GOCAD++, which will soon be available at SEP.
- Ability to display well log data (e.g. velocity, dip meter, SP, etc).
- Improved gridding capabilities.
The final item poses the greatest problem. The current gridding
algorithm is extremely limited in the range of models it can handle. In
addition the velocity characterization is limited by the surface approach
taken by GOCAD.
The documentation of GOCAD++ states that it
can grid constant velocity, compaction functions,
and arbitrary x, y, and z functions within a layer. This is a significant
improvement, but still not ideal. Ideally, we would like to be able
at one stage to describe the velocity function within the
domains at a very coarse level, and as our knowledge
improves, to refine our definition. In addition we would like to have the
connectivity between surfaces be recognized, eliminating the high
frequency component of the velocity model that is common when characterizing
domains.
As a result SEP is considering two
alternative approaches.
The first approach is to use tools already available in GOCAD. Specifically
to characterize the velocity function within the
domains by tetrahedra.
As time goes along these tetrahedra become more numerous and
get smaller. This approach
would lean heavily on GOCAD's DSI algorithm to give reasonable values within
the tetrahedra and for going from one level of model complexity to another.
The boundary between domains would be smoothed by recognizing
tetrahedra along the boundary and forcing a degree of continuity across the
boundary.
The second alternative is to characterize the velocity
function within the domains as
a series of 3-D splines. This method is commonly used in 2-D problems and is
effective in progressively refining the model (by adding additional coefficients). How to create continuity along the border of regions is unclear. One possible
alternative is to define new domains along boundary regions that also have to
be characterized by a limited number of spline coefficients.
Next: CONCLUSIONS
Up: Mora, Clapp & Biondi:
Previous: Building the 3-D Grid
Stanford Exploration Project
11/12/1997