This paper proposes a background noise suppression technique for vibroseis data based on non-stationary signal decomposition. Taking the time-frequency correlation between sweep signals and effective signals as prior constraints, a non-stationary signal decomposition model is constructed. The orthogonal matching pursuit algorithm is adopted to realize the adaptive separation of effective signals and complex background noise, thereby breaking the limitation of traditional methods that assume stationary noise. Tests on synthetic data verify that the proposed method can effectively extract uncorrelated effective vibroseis signals amid complex noise environments. Applications to field seismic data acquired from the Gobi area of the Tarim Basin with low signal-to-noise ratio further demonstrate that this method outperforms the frequency-wavenumber predictive filtering method in background noise attenuation. The processed seismic sections feature improved event continuity and elevated overall signal-to-noise ratio.

vibroseis, high-efficiency acquisition, low signal-to-noise ratio (SNR), uncorrelated, background noise; nonstationary signal decomposition