** Next:** Data residuals
** Up:** Noise separation in field
** Previous:** Noise separation in field

Figure through contain the inverted
model domains associated with the four techniques explored herein.
Clearly the sparsity of the results decrease toward the right. Direct
interpretation of the model domain would be easier and more
accurate with the BC, Cauchy, and *l*_{1} results. Distict trains of
energy in the hyperbolic and linear domains are visible. Without
the clip set so low, the *l*_{2} result is not as bad as these panels
make it seem. A distinct fairway of energy, marginally larger than
that presented in the *l*_{1} result is visible. Velocity analysis of
the gathers would likely be more accurate with the results that impose
sparsity, especially at late times.

**m.08
**

Figure 4 Model-space inversions of
shot-gather 08 (Panel (a) from Figure ). The first half
of the panels is HRT space and the second half is LRT space.
Panel (a) shows the BC, Panel (b) the Cauchy, Panel (c)
the *l*_{1}, and Panel (d) the *l*_{2} inverted model spaces.

**m.14
**

Figure 5 Model-space inversions of
shot-gather 14 (Panel (b) from Figure ). The first half
of the panels is HRT space and the second half is LRT space.
Panel (a) shows the BC, Panel (b) the Cauchy, Panel (c)
the *l*_{1}, and Panel (d) the *l*_{2} inverted model spaces.

**m.25r
**

Figure 6 Model-space inversions of
shot-gather 25 (Panel (c) from Figure ). The first half
of the panels is HRT space and the second half is LRT space.
Panel (a) shows the BC, Panel (b) the Cauchy, Panel (c)
the *l*_{1}, and Panel (d) the *l*_{2} inverted model spaces.

In general the *l*_{1} inversion produced consistantly better model
space realizations. While the differences are often minor, the method
provides consistant, stable, high quality results.

** Next:** Data residuals
** Up:** Noise separation in field
** Previous:** Noise separation in field
Stanford Exploration Project

5/3/2005