Prestack land data multiple removal example

The above methodology has also been tested on a shot record from a land data survey. The preprocessing and multiple prediction is described by Kelamis et al. (1999). I display in Figure 14a the selected shot record. Figure 15a shows the predicted multiples. Note that these multiples are the ones that are directly generated by the shot record based convolutions Berkhout and Verschuur (1997) and no adaptation has been applied yet. Both $\ell^2$ and $\ell^1$ adaptive subtraction has been carried out for this data. The resulting records are displayed in Figures 14b and c, respectively. The removed multiples are shown in Figures 15b and c.

Figures 15b and 15c demonstrate that the multiples are better attenuated with the $\ell^1$-norm for long offsets. Although the truth cannot be revealed from these results, it appears that the $\ell^1$ results are more reliable.

 

comp.s.4
Figure 14 Example of adaptive multiple subtraction for land data. a) One selected shot record from a land survey. b) Estimated signal after $\ell^2$ adaptive subtraction. c) Estimated signal after $\ell^1$ adaptive subtraction.

 

comp.n.4
Figure 15 a) Predicted multiples for Figure 14a. b) Removed multiples with $\ell^2$ adaptive subtraction. c) Removed multiples with $\ell^1$adaptive subtraction. Far offset multiples are better attenuated with the $\ell^1$-norm between 1.2 and 1.8 seconds.