Seismic Signal Denoising Using $f-x$ Variational Mode Decomposition

Ground roll is a common coherent noise in land seismic acquisition, which is usually characterized by high amplitudes, low frequencies, and low propagation velocities. Removing ground roll is a key procedure in seismic data processing. In this letter, we develop a novel ground roll suppression approach termed f-x variational mode decomposition (VMD), which has the characteristic of a frequency-dependent, high-wavenumber filter. This new algorithm first applies VMD on constant frequency slices in the frequency-offset domain within the frequency range of ground roll in order to extract the ground roll, and then the denoised data can be obtained by subtracting them adaptively from the original data. A synthetic example is given to demonstrate that the proposed method is able to suppress ground roll well and preserve seismic reflections effectively compared with other conventional techniques such as frequency filtering and f-k filtering. Application of the method on field data further demonstrates the potential of the f-x VMD algorithm in industrial applications.