For the maintenance of railways on soft soils, accurate knowledge of the subsoil conditions is essential. Soft soils at shallow depths have high variability; thus, high spatial resolution is required. Spare telecommunication fiber-optic cables, known as dark fiber, can be used as an array of sensors to measure waves generated by running trains, which offers a unique opportunity to characterize shallow soils at high spatial resolution. We used dark fiber to measure seismic waves generated by running trains and implemented a seismic interferometry technique to retrieve surface waves. We evaluated the reliability of selected parts of the recorded signals split as bow waves (the train approaching the fiber), train waves (the train passing alongside the fiber), and tail waves (the train leaving the fiber) to retrieve broad-band surface waves. The analysis was performed in two distinctive zones. Zone I consists of a thick-soft (2.0-6.0 m thickness) layer, and Zone II consists of a thin-soft (less than 2.0 m thickness) layer, both overlaying a "stiffer" sand layer. At Zone I, train waves yielded the best results in revealing the thick-soft layer. At Zone II, the bow waves yielded clear high-frequency energy, revealing the overall soil structure but without identifying the shallow thin-soft layer.
Keywords: critical train speed; dark fiber; fiber optics; surface waves.