Detailed knowledge of the 3D basin structure underlying urban areas is of major importance for improving the assessment of seismic hazard and risk. However, mapping the major features of the shallow geological layers might become expensive where large areas must be covered. In this study, we propose an innovative tool, based mainly on single station noise recordings and the horizontal-to-vertical-spectral ratio (H/V), to identify and locate the depth of major impedance contrasts. The method is based on an identification of fingerprints of the major impedance discontinuities and their migration to depth by means of an analytical procedure. The method is applied to seismic noise recordings collected in the city of Almaty (Kazakhstan). The estimated impedance contrasts versus depth profiles are interpolated in order to derive a three-dimensional (3D) model, which after calibration with some available boreholes data allows the major tectonic features in the subsurface to be identified.