Time-frequency vector median filtering and its application to noise attenuation

Tim Seher (Spectrum Geo Ltd.) and Fernando Perez Ortega (University of Leeds) evaluate how coherent and incoherent seismic noise complicates the analysis of seismic data and attenuation of this noise is one of the primary objectives of seismic data processing prior to imaging. This  information was presented at EAGE 2018 in Copenhagen, Denmark.

Coherent and incoherent seismic noise complicates the analysis of seismic data and attenuation of this noise is one of the primary objectives of seismic data processing prior to imaging. Seismic noise often varies in space as well as in frequency and time, which can be exploited for noise removal. Here, we present a novel noise attenuation method that makes full use of these  properties. The new method combines a short time Fourier transform, an extension of the vector median filter to complex numbers, an efficient thresholding method, and a fast dip scan in the frequency domain. This novel method allows the exploitation of variability in seismic noise inside a single process and the attenuation of seismic noise all the way to the spatial alias frequency. This paper introduces the new noise attenuation method with
a special focus on the extension of the vector median filter. It then  demonstrates the effectiveness of the new processing algorithm for low and intermediate frequency noise attenuation using a field data example from offshore Namibia. The new noise attenuation method facilitates imaging of deep crustal reflectors crucial to meeting the imaging objectives.

Click to read the full article here:

Time-frequency vector median filtering and its application to noise attenuation