Polynomial fitting + narrow reject-band f-k filtering for seismic coherent noise elimination

Abstract

The paper presents a f?k filtering method to reduce coherent seismic noise. The method uses an adaptive strategy that the f?k filtering automatically adapts to lateral variations in apparent dip and amplitude of coherent noise events. We first use orthogonal polynomials to fit apparent dip and amplitude of coherent noise events. We then incorporate the instantaneous variations in apparent dip and amplitude into narrow reject-band (NRB) f?k filters which are applied in a spatially and temporally varying way across seismic traces. The method is only applied to the area where coherent noises are located and will not affect the signal that is located outside the area in the same seismic record but has the same regularity as the noise. That is, in a given window, the method removes coherent noises using the differences between noise and signal in three aspects: apparent velocity, frequency, and amplitude. The data outside the window will keep intact. The method can remove any coherent noise in prestack and poststack seismic data, which will favor following seismic processing and modification of seismic sections. The novel method has been applied to field seismic data and shows good attenuation characteristics with minimal distortion of signal.