Next: Rytov image perturbation
Up: WEMVA theory
Previous: Linearization
The simplest way of computing image/wavefield perturbations
is by simple subtraction of the wavefields for the
background image
from the wavefield of a better image
:
|  |
(13) |
Equation (13) is only valid for small
perturbations of the wavefields (
).
In practice, this requirement means that the cumulative
phase difference between the two different wavefields
is small at all frequencies.
If this condition is
satisfied, we can compute a slowness perturbation
which corresponds to the Born approximation:
| ![\begin{displaymath}
\Delta s_b= {\bf B}^* \left(\mathcal U_o\right)\left[\Delta \mathcal U_b\right]\;.\end{displaymath}](img29.gif) |
(14) |
In practice, the small perturbation requirement is
hard to meet, since small slowness differences ammount
to large cumulative phase differences.
Thus, with the wavefield perturbation definition
in equation (13), we can only handle small
slowness perturbations.
Next: Rytov image perturbation
Up: WEMVA theory
Previous: Linearization
Stanford Exploration Project
10/14/2003