Regularization of the AVO inverse problem by means of a ...
Regularization of the AVO inverse problem by means of a ...
Regularization of the AVO inverse problem by means of a ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
CHAPTER 4. TWO-TERM <strong>AVO</strong> INVERSION 41<br />
or<br />
where Lf = WGf and<br />
d = Lf mf + n, (4.5)<br />
mf =<br />
4.2.1 Likelihood function <strong>of</strong> <strong>the</strong> data<br />
<br />
rα<br />
rβ<br />
<br />
. (4.6)<br />
Theoretically, <strong>the</strong> uncertainty between <strong>the</strong> observed data and syn<strong>the</strong>tic data (data generated<br />
using convolution model) is <strong>the</strong> noise in <strong>the</strong> observed data. Assuming <strong>the</strong> noise terms are in-<br />
dependent and Gaussian, <strong>the</strong> noise can be modeled using Multivariate Gaussian probability<br />
distribution given <strong>by</strong><br />
P (n|Cd) = exp{− 1<br />
2 nT C −1<br />
d n}, (4.7)<br />
where Cd is <strong>the</strong> data covariance matrix having <strong>the</strong> form<br />
⎛<br />
σ<br />
⎜<br />
Cd = ⎜<br />
⎝<br />
2 d1<br />
0<br />
0<br />
σ<br />
. . . 0<br />
2 .<br />
.<br />
.<br />
d2<br />
.<br />
.<br />
.<br />
.<br />
.<br />
.<br />
0<br />
.<br />
.<br />
.<br />
0 0 . . . σ2 dMN<br />
⎞<br />
⎟ . (4.8)<br />
⎟<br />
⎠<br />
Using equation (4.5) and (4.7), <strong>the</strong> likelihood function <strong>of</strong> <strong>the</strong> data can be expressed as<br />
where,<br />
P (d|mf ) = Po exp{− 1<br />
2 (Υ(d − Lf mf )) T Cd −1 (Υ(d − Lf mf ))}, (4.9)<br />
Po =<br />
1<br />
(2π) (NM)/2 . (4.10)<br />
|Cd| 1/2<br />
A diagonal matrix, Υ, is also introduced for muting i.e to protect <strong>the</strong> algorithm from trying<br />
to fit with <strong>the</strong> input data with zero entries.