The deformation or collapse of tracks, due to sinkhole formation beneath railway platforms, is a natural hazard that can affect train traffic regularity, cause extra maintenance costs, but also lead to crucial safety issues. In this paper, we present an innovative seismic monitoring method that uses train as seismic sources. Thanks to interferometric methods and the reconstruction of virtual seismic traces between pairs of receivers, we use the surface waves generated by trains and recorded on a seismic sensor network to image the near-surface. Combined to an adapted design, a robust acquisition system and a deported processing system, this innovative geophysical methodology allows a near-real-time permanent and continuous measurement of seismic attributes. Hence, some geophysical attributeslike surface (Rayleigh) wave phase velocity dispersion curvescan be computed continuously and inverted to produce periodically S-wave velocity volumes. The fully automated process can provide, on a daily basis, a high-resolution (~2 m) S-wave velocity 3D volume of the shallow near surface (from 0 to about 50 m in depth). Then, analysis of the variation of S-waves velocity along time can be directly interpreted to track the genesis and growth of cavities due to leakage and dissolution phenomena. In this study, we show the results obtained after one year of continuous monitoring obtained with a wired accelerometer system. This allowed to observe S-wave velocity variations related to complex phenomena of leakage and decompression. Alternative acquisition systems (fiber optic using DAS technology for instance), processing methods (analyzing body-waves or back-scattering in addition to surface waves), or even different targets