Application of median filtering in Kirchhoff migration of noisy data

Yanpeng Mi, Xinxiang Li, Gary F. Margrave

Random noise and abnormally strong coherent noise have been presenting difficulties in seismic migration and imaging. Although noise level should have been reduced dramatically by signal processing such as CMP stacking and multichannel-based poststack noise reduction, the remaining noise may still be too strong for migration and imaging. A feasible way is to perform noise reduction during migration. Kirchhoff migration by weighted hyperbolic summation is a method on which a multichannel based noise reduction technique can be easily applied, since each sample in a migrated section corresponds to a horizontal amplitude series. Median filtering is often effective in removing random noise and strong abnormal amplitudes within a time series and it is often applied in the spatial direction to reduce spatial inconsistency. We applied a median filter before the summation in Kirchhoff migration, and some preliminary results are presented.