SUMMARY Train-induced vibrations act as potential powerful high frequency source for imaging subsurface with higher resolution than typical ambient noise interferometry. In this study, we present results of seismic interferometry applied on three days of railroad traffic data recorded by an array of seismographs along a railway in Dublin, Ireland. Our virtual shot-gathers show significant surface- and body-wave energy that could be used for seismic interferometry. Reflection sections obtained with our interfrometry approaches applied on selected time windows of train-induced vibrations is consistent with nearby borehole data and an active seismic profile. The consistency of the results given by these approaches confirms that train-generated vibrations represent a valuable source of signal for high resolution subsurface imaging. Furthermore, our results show spurious arrivals that are due to the train geometry and also the cross-correlation approach that needs consideration for body wave interferometry studies.

<p>The behaviour of geological slopes during seasonal weather patterns represents one of the challenges for assessing the geotechnical state of health of the ageing infrastructures. In the presence of man-made soil infrastructure slopes, rainfall and prolonged dry periods can cause cycles of swelling and shrinking of the ground that could potentially compromise their structural integrity. Recent research has found that time-lapse velocity monitoring, has the potential to provide information on climate-related deterioration of geotechnical infrastructure. Variations of the ground conditions could manifest as changes in seismic velocity, detectable through the seasons and after extreme weather events.</p> <p>In this work, we perform seismic imaging and velocity-monitoring of a critical railway embankment in the United Kingdom using fibre optic distributed sensing (DAS). The study area is a 6 m tall, and 350 m long embankment slope built more than 100 years ago in the outskirts of London (Surrey). The railway is currently utilised mostly by commuter trains. Since August 2022, a passive DAS dataset rich in train signals has been acquired. data acquisition will continue until July 2023. Furthermore, periodic active surveys have been conducted along the slope.</p> <p>Firstly, to validate the seismic response of the fibre (i.e., maximum usable frequencies based on the gauge length), we calculate and compare surface wave dispersion curves derived from both DAS and geophones using passive ambient noise, train signals and active sledgehammer shots. As a result, we obtain consistent and comparable dispersion curves ranging from ~200 m/s at 10 Hz to ~140m/s at 40 Hz. </p> <p>Secondly, we invert, using global search algorithms, DAS-derived dispersion curves for 1D depth-velocity models to identify and clarify the trend of the near-surface (top 10 m) seismic structures. </p> <p>Thirdly, we apply seismic interferometry and moving window cross-spectral techniques to measure changes in seismic velocity at the embankment using the 6-month passive DAS data acquired so far. </p> <p>The ultimate goal of this project is to develop a geophysical tool diagnostic of geotechnical deterioration of critical infrastructures by linking together DAS-based seismic observations, temporal seismic velocity changes, weather data and laboratory-based soil sample tests.</p>

We investigate the shallow geological features and previously unknown depths of the main geological interfaces in the highly urbanized Dublin City area by using a set of complementary ambient noise seismic methods. A sparse seismic array, composed of 19 broadband sensors, was installed in a 5 × 5 km area across the city centre. The dataset was acquired over a 5‐month‐long deployment and used to perform horizontal/vertical spectral ratios and frequency–wave number analysis cross‐correlation interferometry. From the horizontal/vertical high‐frequency resonance peaks, we estimate the depth to bedrock. Then we use a subset of six seismic stations to obtain the frequency–wavenumber Rayleigh wave fundamental mode in the 0.8–3 Hz frequency band. Next, the ambient noise dataset is cross‐correlated in order to extract the empirical Green's functions before measuring surface‐wave phase velocities by undertaking a dispersion analysis in the 0.5–9 Hz frequency band. Then, a Monte Carlo global optimization algorithm is used to invert the phase velocity dispersion measurements. We generate a reference one‐dimensional depth S velocity profile along with a set of localized one‐dimensional S velocity profiles. Finally, a smooth three‐dimensional shear wave velocity model is derived for the top 1000 m in the sedimentary Dublin Basin. From the one‐dimensional velocity profiles and three‐dimensional shear velocity model, with some sensitivity down to approximately 1.2 km, we observe velocity changes with depth that highlight the presence of three consistent interfaces. We discuss interpretative scenarios that correlate the velocity features to potential stratigraphic boundaries occurring within the sedimentary Dublin Basin, suggesting that basement rocks could be located at a depth significantly greater than the top kilometre. The results of this study may indicate that future geothermal studies should be directed at structures deeper than 1 km, towards the bottom of the sedimentary basin beneath the city.