Characterization and trajectory inversion of subway-induced vibration signals based on urban seismological methods

Urban anthropogenic seismicity is no longer a nuisance that interferes with natural seismic observation; it now provides an alternative data stream—a passive, pervasive information carrier for sensing urban dynamics. Subway-induced vibrations are a repetitive and regular anthropogenic seismic source that is frequent in modern cities. However, its signal remains inadequately studied and characterized, which limits its application in the field of urban seismology. In this study, several seismometers were deployed on the campus of Harbin Engineering University in Northeast China to capture seismic signals from typical urban activities, with a particular focus on subway-related ones. The main frequency bands, time-frequency characteristics, and multi-scale periodicities of subway-induced seismic signals were analyzed and characterized. Polarization analysis conducted in both the time and time-frequency domains revealed the polarization patterns and directional characteristics of wave propagation. The results demonstrate that seismic polarization not only contributes to the determination of directions of subway train circulation, but also enables subway trajectory estimation through dual-station observations. Additionally, a comparative analysis between observations at nearby and distant stations reveals that spectral time-frequency characteristics near stations are dominated by train acceleration/deceleration mechanisms. This study establishes a robust "seismic fingerprint" template for characterizing subway signals within multi-source urban wavefields, providing both theoretical and empirical foundations for vibration-based infrastructure monitoring and urban dynamics sensing.