Ground-roll noise extraction and suppression using high-resolution linear Radon transform

Abstract

Ground-roll is a main type of strong noises in petroleum seismic exploration. Suppression of this kind of noise is essential to improve the signal-to-noise ratio of seismic data. In the time–offset (t–x) domain, the ground-roll noise and the effective waves (e.g., direct waves, reflections) overlap with each other in terms of time and frequency, which make it difficult to suppress ground roll noise in exploration seismic data. However, significant different features shown in the frequency–velocity (f–v) domain make it possible to separate ground roll noise and effective waves effectively. We propose a novel method to separate them using high-resolution linear Radon transform (LRT). Amplitude and phase information is preserved during the proposed quasi-reversible transformation. The reversibility and linearity of LRT provide a foundation for ground-roll noise suppression in the f–v domain. We extract the energy of ground-roll noise in the f–v domain, and transform the extracted part back to the t–x domain to obtain the ground-roll noise shot gather. Finally, the extracted ground-roll noise is subtracted from the original data arithmetically. Theoretical tests and a real world example have been implemented to illustrate that the ground-roll noise suppression can be achieved with negligible distortion of the effective signals. When compared with the adaptive ground-roll attenuation method and the K–L transform method, the real world example shows the superiority of our method in suppressing the ground-roll noise and preserving the amplitude and phase information of effective waves.