LISE RETAILLEAU1, LEI LI2, PIERRE BOUÉ2, THIBAUD BLONDEL3, JULIEN CHAPUT4, ALEXANDRE AUBRY3, ARNAUD DERODE3, MICHEL CAMPILLO*2
1Stanford University, USA, 2Université Grenoble Alpes, France, 3ESPCI, France, 4University of Texas El Paso, USA
Noise based passive imaging has been very successful with surface wave tomography although body waves have been identified early in the correlations. Body waves reflection are most often weak signals that are difficult to extract. When considering the level of unphysical fluctuations in the correlations, performing body wave imaging requires strategies to improve signal to noise ratios. It is nevertheless possible to perform high-resolution deep imaging with correlations as illustrated in two examples here where arrays with large numbers of seismometers are used.
In the first example (Retailleau et al, 2018), we use P and PcP waves from the secondary microseism frequency band that are propagating between Europe and the Eastern US to image the Core-Mantle Boundary (CMB) and D” structure beneath the North Atlantic. We use continuous records from stations in Europe and US. We extract subarrays on each continent for which we apply double beam forming techniques for pairs of subarrays containing at least 200 stations. When considering the spatial distribution of reflectivity with depth around the CMB for the whole data set, we observe complex patterns of lateral and vertical variations of P-wave reflectivity with a particularly strong anomaly extending upward in the lower mantle up to 2600km deep in a region with an extension of about 200 by 100 km.
We present a second example (Blondel et al., 2018) using a small-scale array for passive imaging in a highly scattering medium, namely a volcano. We used the redundancy of the information given by the dense array and a matricial imaging approach. In heterogeneous areas such as volcanoes, the multiple scattering contribution limits the imaging depth to one scattering mean free path, the mean distance between two successive scattering events. Our matrix approach of passive seismic imaging pushes back this fundamental limit by making an efficient use of the singly scattered body waves drowned into a noisy seismic coda. The response functions between a set of geophones placed on top of the volcano are retrieved by the cross-correlation of coda waves produced by multiple icequakes. This set of impulse responses forms a reflection matrix that is processed to obtain an image of the structures that reveal the plumbing beneath the volcano at depth up to 6 km beneath the dome.