Extraction of diving body waves from a dense network of seismometers at kilometric offsets: a case study from the Paris Basin

We reconstruct the body and surface waves from the seismic ambient wave field, recorded by a dense seismic array deployment in the Paris Basin, France, with a final objective of performing a 3D seismic tomography by inversion of the retrieved P-phase arrivals. The array was installed in November 2010 and consists of around 100 stations. The stations were shifted to different locations every day, yielding around 580 recording locations with an interstation distance of about 400 m. Each station has continuously recorded around 3-4 days of the seismic ambient wavefield. We calculate the cross-coherency between each station couple in the frequency band of 1.0-4.5 Hz and estimate the empirical Green’s functions. We use the polarization properties of the cross-correlation tensors to separate the P- and Rayleigh wavefields. The results show the reconstruction of the fundamental and higher modes of Rayleigh and Love waves, as well as of diving P- waves. We observe the apparent group velocity of the fundamental and first higher mode of the Rayleigh wave around 0.5 and 1.5 km/s, respectively, and the apparent group velocity of around 0.8 km/s for the fundamental mode of the Love wave. The extraction of the P waves is challenging because of a high amplitude coherent artefact that can cause misinterpretation of the P- wave moveout. We propose a new approach to filter out the extracted P- wave in presence of this artifact and reconstruct the P-wave with a correct apparent velocity of around 2-3 km/s, validated against available active seismic data. This approach is based on template-matching and can be regarded as the most crucial step in P-wave retrieval from our dataset. Future steps will consist of using the extracted P-wave arrival time for a 3D tomography of the anticline structure located beneath the array.