Investigating the effect of source distribution on the retrieval of body waves and of the Earth’s reflection response by seismic interferometry

Abstract:

Over the last decade, seismic tomography based on the interstation correlation of ambient noise has developed into a standard tool for exploring and monitoring the Earth’s interior. This method was first based on the extraction of surface waves. Another, more recent application of seismic interferometry is to extract body waves and retrieve the earth’s reflection response. Compared to surface-wave extraction, body-wave extraction and reflection retrieval is a much greater challenge because ambient noise is typically dominated by the surface-wave energy and because the source distribution is limited in depth. Nevertheless, opportunistic sources such as road or rail traffic generate enough high-frequency body-wave energy to allow the use this noise for passive reflection seismic imaging and for monitoring.

In theory, seismic interferometry yields the exact Green’s functions of the medium if noise sources are uniformly distributed around the stations. However, sources of high-frequency body waves such as trains are usually not evenly-distributed around the areas of interest, and are mainly limited to the Earth’s surface. Furthermore, the distance between the sources and the array may vary and can affect the retrieval of the reflection image. As a consequence, the resulting cross-correlations do not correspond to exact Green’s functions. Nevertheless, these cross-correlations contain information about the subsurface which we should be able to extract if we understand them in detail.

To investigate and better understand the effect of the source distribution on the retrieved earth’s reflection response, we used both simulated train signals and real train-induced vibrations recorded along a railroad in Dublin, Ireland. From the recorded data, we extracted small time windows corresponding to different distances between the train and the array, and we compared the results with the reflection section obtained with the time window when the train passes the array. Our preliminary results document a strong relation between the source location (the train in this case) and the retrieved seismic reflection image. Results obtained from synthetic simulations are consistent with the data and enable us to investigate the correlation between the source location and the retrieved reflection image.