Aiming for a geophysical survey to around 1.5 km depth, we conduct a passive‐source surface‐wave survey along a 70‐km‐long line on the western side of the Junggar basin in northwest China. We use frequency–time analysis and spatial autocorrelation methods to extract surface‐wave dispersion curves from ambient‐noise data of station pairs, as well as ambient‐noise multichannel analysis of surface waves from multistations. Dispersion curves corresponding to the same positions obtained by these three methods independently are comparable, which shows the validity of using these methods. After inverting more than 400 dispersion curves and aligning their results together, we construct a 2D shear S‐wave velocity image around 1.5 km deep and 60 km long. A less than 6% difference between the inverted S‐wave velocity profile and the borehole data in situ shows the relatively high accuracy of our method. The 2D S‐wave velocity image clearly delineates the main geologic structures, including transition of landform, intrusion of rock mass, and shape of bedrock beneath the survey line.