Pre-stack seismic noise attenuation based on 3D CWT

Pre-stack seismic dataset contains much more information than post-stack dataset and can be applied in velocity analysis, AVO analysis, and fracture prediction. However, pre-stack dataset suffers from a much lower SNR and more complicated noise types than the post-stack dataset. Thus attenuating the random noise and other coherent noise becomes one crucial step for pre-stack dataset. Some 2D-transform-based noise attenuating methods cannot fully utilize the spatial correlation information in 3D pre-stack volume. In this abstract, we propose to use 3D continuous wavelet transform (CWT) to denoise pre-stack seismic data. We first decompose a 3D pre-stack volume to a six-dimensional domain including the space coordinates b=(bx,by,bz)^T, scalar a, dip θ and azimuth φ, which can access to the 3D spatial information. Then we remove some coefficients in dip domain and azimuth domain through the features of noise to attenuate a part of coherent noise, and attenuate the random noise and other coherent noise by thresholding the rest coefficients. Because 3D CWT can fully utilize the 3D spatial information in 3D prestack volume, our 3D-CWT-based noise attenuation method shows a considerable advantage over other 2D noise attenuating methods on some real 3D common offset vector datasets.