A novel seismic interferometry method for high-speed train-induced vibrations based on deconvolution and weighted stacking

ABSTRACT High-speed trains (HSTs) can generate seismic wavefields with extensive propagation ranges during operation, which can be used for seismic interferometry (SI) imaging. We develop a novel SI method for coherent sources, such as HSTs. In the new SI method, the source power spectral term is eliminated using deconvolution interferometry. We also develop a novel adaptive covariance filter (ACF)-based Nth-root weighted stacking method to suppress other interferences and further improve the signal-to-noise ratio (S/N). The ACF-based Nth-root method aims to improve the quality of virtual shot gathers (VSGs) further by amplifying segments of the noise cross-correlation function with high S/N, thus more effectively extracting valid signals under complex noise source distributions. Through the processing of synthetic and measured data, the empirical Green’s function (EGF) of Rayleigh waves is successfully recovered from HST seismic data. Dispersion curves are extracted from VSGs, and the S-wave velocity structure is inverted. Our results indicate that it is feasible to use HSTs as noise sources for passive surface wave imaging and that deconvolution is more suitable for processing coherent source seismic data than crosscorrelation. Meanwhile, the ACF-based Nth-root method can further improve the S/N of EGFs. Our method provides an efficient and effective option for passive surface wave imaging.