Investigating Noise Attenuation by Anisotropic Diffusion
Event name: 80th EAGE Conference and Exhibition 2018
Session: Student Poster: Geophysics Other, Energy, CO2
Publication date: 11 June 2018
Info: Extended abstract, PDF ( 720.47Kb )
Price: € 20
Attenuating noise from seismic recordings is one of the main objectives during processing, and if executed successfully it improves the interpreter’s ability to characterize the geology. Anisotropic diffusion is a noise attenuation method that focuses on reducing various forms of incoherent noise that obscures the signal by smoothing in predefined directions following the texture. Constraints can be applied to the smoothing by the dip field itself or corresponding attributes to ensure preservation of meaningful edges in the seismic section, such as variations in the temporal reflectivity strength related to reflectors or the spatial terminations of said reflectors. Such constraints can include magnitude and reliability of the local dip, and degree on anisotropy or coherency. The amount of smoothing can also be controlled by the number of diffusion steps, meaning the same diffusion applied several times for a clearer result, but with caution to avoid over smoothing. In this work, anisotropic diffusion was formulated as a line integral convolution and results were investigated when applied on post-stack seismic data, along a dip field estimated by a gradient structure method. The results showed a decrease in random noise with characteristic salt and pepper features, causing more distinct reflectors.