Time-Lapse Seismic Monitoring of a Railway Embankment Using Train-Induced Distributed Acoustic Sensing

Aged railway embankments, constructed in the late 19th and early 20th centuries without modern engineering standards, are increasingly vulnerable to failure due to ageing, climate change, and rising transportation demands. Extreme weather events, particularly periods of prolonged wetting and drying, pose a significant risk to the structural resilience and serviceability of these earthworks. This study explores a novel passive monitoring framework at the Withy Bed site near London, UK, utilising Distributed Acoustic Sensing (DAS) to capture the long-term seismic response of a live railway embankment. By leveraging train-induced vibrations as a continuous ambient seismic source, we provide a non-invasive and high-resolution assessment of the slope's condition. The site is further equipped with a suite of geotechnical sensors, including volumetric water content (VWC) and suction sensors, to provide direct measurement of the embankment's internal state. We present time-lapse results derived from a 350m fibre-optic cable buried within the embankment, with processed data spanning every month of the year. The findings demonstrate a clear correlation between shear wave velocity (Vs) and geotechnical sensor data, specifically, Vs decreases during periods of high water content and low suction, reflecting a reduction in soil stiffness during wet seasons. Conversely, during dry periods, the data indicate a significant increase in Vs as the water content decreases and soil suction increases, resulting in a measurable rise in the overall stiffness of the embankment. The results show clear month-to-month changes in dispersion trends and Vs, with significant percentage decreases in Vs during wetter months and a progressive recovery of stiffness during drier periods. These temporal changes are spatially coherent along the embankment and repeatable across successive train events, demonstrating the robustness of the passive approach. The time-lapse analysis confirms that train-induced seismic waves provide sufficient energy and consistency to resolve seasonal variations in near-surface stiffness without repeated active surveys. This work demonstrates that passive DAS provides a practical, non-intrusive, and scalable solution for continuous monitoring of railway embankments, supporting the early identification of condition changes and enhancing infrastructure asset management.