|
|
|
| Wave-equation migration velocity analysis by residual moveout fitting | |
|
Next: Computational cost
Up: Gradient of the objective
Previous: Derivatives of objective function
The evaluation of the derivatives of the moveout parameters
with respect to slowness
takes advantage of the fact that we need to evaluate the derivatives
only at maxima for the objective function
in equation 5.
At the maxima, the objective function
is stationary and thus its derivatives with respect to the moveout
parameters are zero, and
we can write:
|
(A-12) |
As discussed above,
the derivatives in the second term (II)
of equation 9
are different from zero
only when the moveout coefficient
and the image element share the same
spatial coordinate.
Consequently, for each
there is only one
for which the inner products above are different from zero.
Equation 12 can thus be simplified into:
|
(A-13) |
Using the rule for differentiating implicit functions,
and taking
advantage of the fact that the fitting problems are all independent
from each other (i.e. the cross derivatives with respect to the moveout
parameters are all zero), we can formally write:
|
(A-14) |
From equation 13 and 14,
the derivative of the local moveout parameters with respect to slowness
is:
|
(A-15) |
Appendix A presents the development for the expressions
to compute the terms
(A-3),
and
(A-5).
Combining the derivatives
in equation 15
with the derivatives in
equations 10-11
we can easily compute the gradient of the objective
function 4 with respect to slowness
that can be written, when
,
as:
|
(A-16) |
I will now examine the effects of each of the terms
in equation 16
starting from the rightmost one.
The third term (III) produces a scalar
for each local curvature parameter
.
This scalar multiplies the image for each physical location
after it has been differentiated in depth and scaled by
,
as described by the second term (II).
Notice that the phase introduced by the depth derivative
of the image in (II)
is crucial for the successful backprojection into the slowness model
that is accomplished by the first term (I).
In this term, first
transforms the image
from the aperture-angle domain into the subsurface-offset domain.
The transformed image is scaled, horizontally shifted, and spread
across the shot axis by the application of
and
.
Finally,
the adjoint of operators
and
backproject the image perturbations
into the slowness model along the source wavepaths and the receiver wavepaths,
respectively.
|
|
|
| Wave-equation migration velocity analysis by residual moveout fitting | |
|
Next: Computational cost
Up: Gradient of the objective
Previous: Derivatives of objective function
2010-11-26