** Next:** Acknowledgments
** Up:** Vlad: Enhanced random noise
** Previous:** Practical issues

The spurious coherent noise patterns passed by the two spatial
prediction filtering methods are different enough that their
coherence is destroyed when added together. A formal mathematical investigation on why this
holds true in the vast majority of cases may yield useful insights on
optimally choosing the parameters for the two methods so that the
final signal-to-noise ratio is maximized. However, even on a heuristic
basis, Enhanced Random Noise Attenuation works. It is simple, fast,
intuitive, and does not require any coding if *f-x* and *t-x*
decons are already implemented. Without being more unstable than
either of these techniques, it builds on both of them in order to
produce a better result.

** Next:** Acknowledgments
** Up:** Vlad: Enhanced random noise
** Previous:** Practical issues
Stanford Exploration Project

7/8/2003