This thesis examines the use of seismic surface-wave interferometry for near-surface characterization from ambient noise records. An artificial sand body at the CNR experimental site was being monitored using a dense receiver array. By analysing the cross-correlations of ambient noise between receiver pairs, we can extract empirical Green’s functions and Rayleigh-wave dispersion curves.We accessed over 7000 receiver paths and selected 2130 reliable dispersion curves for inversion. Resulting estimates were a Vs distribution to almost 5 m depth. Two types of inversion schemes were tested; unconstrained and constrained. Although a similar data misfit (≈11.7%) was achieved by both models, the Vs model resulting from the constrained inversion was found to be more realistic, as it increased smoothly from 100–130 m/s at the surface to 250–270 m/s at depth.