Suppression of ground roll in seismic reflection data using slowness adaptive f-k filter

One of the common problems in reflection seismic records in the land is the existence of surface waves with high amplitude and low frequency which cause the most important parts of reflective signals to be masked. Therefore, it is necessary to attenuate them by processing and acquisition methods. As it is impossible to attenuate the surface waves completely, and sometimes it is likely to damage reflective event signal content, using the complement processing techniques is vital. These methods are based on the properties and assumptions of surface waves. In this paper, we employed a new method called the slowness adaptive f-k filter to attenuate surface waves instead of using the conventional frequency f-k filter. The advantage of the slowness adaptive f-k filter compared with the f-k filter is that in this method, by selection of a space-variant narrow reject-band f-k which is variant with time and space and applying it on data, the flaws of the f-k filter that consist of smoothing the main signal, distortion of the signal and insufficient attenuation of coherent noise, are removed and data frequency content are less influenced.
Finally, both techniques were applied on real and synthetic data. Comparison between the obtained results from both methods showed that these results are nearly similar in some aspects. But overall, the results from the slowness adaptive f-k filter were more accurate.
Ground roll is usually present on reflection seismograms with velocity values between 100 to 1000 m/sec (Telford et al., 1990). In seismic data acquisition, generated by sources like dynamite and vibrators we usually face coherent noises. These types of waves will mask the reflection signals produced in the deeper part of the earth's layers, due to their inherent scattering and low velocity (Saatcilar and Canitez, 1988). Generally the ground role has to be suppressed during the data acquisition operations.
Suppression of the coherent noises during data acquisition will not be complete, therefore elimination processes are subject to application of signal processing sequence procedures (Coruh and Costain, 1983). To suppress surface wave phenomena we have applied the slowness adaptive filtering and have presented its successful results compared to the conventional f-k filtering.
To reduce distortions of the recorded seismic signals, it is recommended to apply a time and space dependent f-k filter. The filter that has been applied here consists of two steps as follows;
1- Apparent slowness of the coherent noise was calculated from the seismograms.
2- With respect to the obtained apparent slowness values, the filter was applied in time and space domain on the seismograms. To do this, the instantaneous apparent slowness must be calculated.
The obtained results showed that the slowness adaptive f-k filter is capable of automatically adapting itself with lateral variations of the apparent slowness. The filtering operation is based on the depth of the seismic events. Some of the filter's characterizations are as follows;
1- Since the slowness adaptive f-k filter is a band compared to the conventional f-k therefore the shape of the signal will remain intact.
2- The slowness adaptive f-k filter will be applied on the specific deep like velocity of the linear coherent noise for which their energy will be deduced. This will preserve the energy of the main reflection signal and reduce the energy of the coherent noises like ground roll.