Geotechnical site characterization with 3D ambient noise tomography

We develop a new 3D ambient noise tomography (3D ANT) method for geotechnical site characterization. It reof geophones, from which experimental crosscorrelation functions (CCFs) are then extracted and directly inverted to obtain an S-wave velocity (VS) structure. The method consists of a forward simulation using 3D P-SV elastic wave equations to compute the synthetic CCF and an adjoint-state inversion to match the synthetic CCFs to the experimental CCFs for extraction of VS. The main advantage of the presented method, as compared with conventional passivesource seismic methods using characteristics of Green's function (GF), is that it does not require equal energy on both sides of each receiver pair or far-field wavefields to retrieve the true GF. Instead, the source power spectrum density is inverted during the analysis and incorporated into the forward simulation of the synthetic CCFs to account for source energy distribution. After testing on synthetic data, the 3D ANT method Valley Downhole Array (GVDA) site in California, using a surface array of 196 geophones placed on a 14 x 14 grid with 5 m spacing. The inverted 3D VS model is found to be concharacterization efforts at the GVDA site.