We collected SH-wave seismic reflection data over a shallow aquifer in southwestern British Columbia to investigate the use of such data in hydrogeologic applications. We used this data set in developing a methodology that uses cone penetrometer data as an integral part of the inversion and interpretation of the seismic data. A Bayesian inversion technique converts the seismic amplitude variations to velocity variations, honoring the probabilities of the priors and adhering to a geologically reasonable sparseness criterion. Velocity measurements acquired with the cone penetrometer provide velocity profiles and vertical seismic profiling (VSP) data, all of which are valuable in properly constraining the Bayesian inversion. The differentiation of lithologies (in this data set, sand and clay) is accomplished by first using a normalization procedure to remove the impact of effective stress, which dominates the velocity variation in the upper 10 to 20 m. The final transformation of seismic velocities to void ratio for the sand-dominated regions is made using laboratory-derived measurements; it provides an image of the heterogeneity of the near-surface aquifer.