CONCLUSIONS

The accuracy of the McClellan transformation filter can be improved by rotating the 9-point McClellan filter by 45 degrees, and then averaging the rotated filter with the original filter. The spectrum of the averaged filter matches the spectrum of the ideal transformation filter more accurately than the higher-order 17-point McClellan filter. The averaged filter can be applied by alternately convolving the data with the two filters; this procedure is about as computationally intensive as extrapolating the wavefield using the 17-point McClellan filter.

The application of the proposed transformation filter to 3-D poststack migration yields a migration algorithm that correctly images steep dips without dispersion of the wavefield.