Next: SMM advantages and disadvantages
Up: Vlad: Irregular data migration
Previous: Vlad: Irregular data migration
Common sense and basic physics suggest that in order to continue a
field (gravitational, magnetic, wavefield) into a direction, it is
necessary to know the values of the field on one or more surfaces
nonparallel to the continuation direction, and the law that governs
the field (approximations can be made or field laws inferred so
that we need only one surface). Because field quantities are usually
invariants, nothing mandates that field
continuations be done on Cartesian grids (although in many cases it is
numerically convenient to do so). In particular, a look at the
downward continuation equation Claerbout (1999) shows that it can be written as:
| ![\begin{displaymath}
\frac{{2i\omega }}{v}\frac{{\partial Q}}{{\partial z}} + \le...
...{2i\omega }}\frac{\partial }{{\partial z}}} \right)\Delta Q = 0\end{displaymath}](img4.gif) |
(1) |
where
Q, the Laplacian of Q, is an invariant.
It can also be computed on an unstructured spatial mesh.
In principle, this means that semistructured mesh migration (SMM)
is feasible.
I called the mesh semistructured because it is regular in time,
but unstructured in space.
Next: SMM advantages and disadvantages
Up: Vlad: Irregular data migration
Previous: Vlad: Irregular data migration
Stanford Exploration Project
6/8/2002