** Next:** Conclusion
** Up:** Normalization by operator fold
** Previous:** Normalization of Kirchhoff operators

Calculating operator fold for finite-frequency migration operators is
not as simple as modeling and migrating (or just migrating) a vector
full of ones.
A vector that is full of ones in the time domain has
zero temporal frequency, and will not propagate in the same way as the
frequencies of interest.
Similarly, a vector that is full of ones in
the frequency domain cause problems because it will be localized in
time. Randomizing the phase may solve this particular problem, but it
will cause others.
Duquet et al.'s approach 2000 is also not
appropriate for normalizing wave-equation migrations because they are
necessarily full-volume methods that cannot be implemented in a
target-oriented manner. A naive implementation would require full
modeling and migration for every point in model-space.

Fortunately the general formulation in Chapter does
not have these limitations, and is applicable to any linear operator
provided an appropriate choice of reference model exists.

** Next:** Conclusion
** Up:** Normalization by operator fold
** Previous:** Normalization of Kirchhoff operators
Stanford Exploration Project

5/27/2001