Surface waves that reflect or backscatter from sharp lateral changes in the near surface provide information about the location, depth, and amount of these variations. We developed a 2D semianalytic forward model for surface-wave propagation and used it to design a processing flow for determining surface-wave reflection coefficients as a function of spatial location along a survey line and frequency (or depth). Using these reflection coefficients, we updated surface-wave phase-velocities (determined by picking dispersion curves) across the survey line. The result was a sharply defined phase-velocity model in the lateral direction, which can be inverted for an S-wave velocity model. Two synthetic examples of blind faults with data simulated by the spectral-element method were developed for our processing flow. Buried faults were properly located, and the S-wave velocity model was improved in the lateral direction. In addition, our method effectively identified faults for seismic data acquired across the Hockley Fault system near Houston, Texas.