De-noising Method for 3D Field Data Based on the MSSA

Successful noise attenuation results in an enhanced image of the subsurface geology, which is an important subject in seismic data processing. In this paper, we introduce a rank reduction algorithm to attenuate random noise, which is based on multichannel singular spectrum analysis (MSSA). This algorithm is implemented for 3D seismic data by utilizing randomized singular value decomposition (R-SVD) into the work of the rank reduction. Numerical tests of 3D synthetic and field data show that this technique is powerful in attenuating random noise as well as preserving the amplitudes of useful signals.