Ground Roll Attenuation Applying Matching Pursuit Algorithm

Summary The main characteristics of ground roll are dispersion, low frequency, low velocity and high amplitude. It significantly affects the quality of the seismic data as a kind of noise and is necessary to be subtracted from the original data. Common ground roll attenuation methods include band-pass filtering, f-k filtering and so on, but the effective signal is inevitably harmed in the denoising process. In this paper, ground roll noise is extracted by using matching pursuit (MP) algorithm. Low frequency as the feature of ground roll can be identified as atom, and the time shift parameter is also employed to preserve the effective signal. Synthetic example shows that the ground roll is attenuated and the effective signal is preserved in the events that are polluted seriously by the ground roll noise. Application of field data proves the efficiency and superiority of matching pursuit algorithm compared with band-pass filtering and f-k filtering.