Seismic ambient noise reconstruction based on the multi-taper spectral method: Application to the seismic full waveform inversion

Abstract

The seismic ambient noises contain real seismic waves that transmit through the subsurface media. Using seismic interferometry, the disorganized ambient noises recorded on surface can be reconstructed into the virtual source records that are similar to the conventional active source seismic data. However, the conventional reconstruction methods introduce additional noises into the final result, which adds the difficulty of the applications. In this paper, we propose an optimal length multi-taper spectral method based seismic interferometry to reconstruct the virtual source records, and point out that a suitable taper length is important for noise suppression. After that, we propose the source independent full waveform inversion method for the reconstructed virtual source data, which avoids virtual wavelet estimation. The final results demonstrate that the multi-taper spectral method with optimal taper length can reconstruct high quality virtual source data for the full waveform inversion.