Next: Combined Data and Model
Up: Regularization of the LeastSquares
Previous: Model Regularization 2: Differencing
NMO (for primaries or multiples) flattens events of a single order, but leaves
events of other orders (crosstalk) remaining with residual curvature. To
compute (for instance) the crosstalk model for firstorder pegleg multiples on
the primary model panel, , the following steps are followed.
 Apply NMO for primaries to data: .
 Zero in the range , where
is the zerooffset traveltime to the multiplegenerating
layer.
 Simulate the kinematics of the data's total firstorder pegleg by
applying inverse HEMNO: .
 Apply NMO for primaries to simulate the kinematics of the firstorder
pegleg crosstalk event in the primary model panel, :
.
 Normalize to [0,1] and clip if desired.
Figure illustrates the application of the crosstalk
weight for first and secondorder peglegs on applied to a NMOed
synthetic CMP gather. Notice how the multiples are ``picked'' cleanly out of the
data, while strong primaries are left largely intact.
crosstalk.hask
Figure 2 Synthetic CMP gather with and
without crosstalk weights for k=0 and j=1,2,3 applied.

 
Denoting the crosstalk weights for each as a vector
, we can write the model residual corresponding to the third model
regularization operator:
 
(7) 
Although the crosstalk weights will likely overlap some primaries, the primaries'
flatness ensures that regularization operators (5) and
(6) ``spread'' redundant information about the primaries from
other and other offsets to compensate for any losses.
Next: Combined Data and Model
Up: Regularization of the LeastSquares
Previous: Model Regularization 2: Differencing
Stanford Exploration Project
7/8/2003