We are now going to use the recipe with the augmented functional
that Plessix (2006) provides to derive the image-space DSO gradient.
First, let us form the Lagrangian augmented functional,
:
(24)
Then the adjoint state equations are obtained by taking the derivative of
with respect to state variables
,
and
:
(25)
(26)
(27)
Equation 25, 26, 27 are the adjoint-state equations.
Variables
and
are the adjoint-state fields and the
solution of the adjoint-state equations 25 and 26. Variable
is the scaled image slice at the subsurface offset
.
Now the gradient of the objective function 15 with respect to velocity is:
(28)
If we combine equations 25, 26, and 27 with equation 28,
we will arrive at the same solution as in the previous section.