A practical strategy for improving GPR images referred to inhomogeneous scenarios

Commercial data processing tools, commonly adopted in Ground Penetrating Radar (GPR) applications, model the investigated subsoil as a homogeneous medium. As consequence, often they provide inaccurate or misleading images when the electric permittivity is not everywhere the same. Therefore, it makes sense to explore the possibility of improving the imaging capabilities of these widely exploited data processing codes by introducing a simple procedure implementable through a homemade software routine. Moving in this direction, we propose a semi-heuristic strategy capable of improving the imaging capabilities in the case of a two-layered soil with a not flat interface between the media. It is worth noting that imaging procedures based on refined models of the signal propagation in inhomogeneous media could be considered. However, their use could be not easy for end-users, especially if they lack for a specific training process. The strategy we propose is practical and not mathematically rigorous, and it aims at exploiting in the best way imperfect (but commonly available) processing tools for GPR data. The strategy is made up of two steps. The focusing step is performed by means of a combination of migration algorithms, and the time-depth conversion step is also performed by combining the estimated properties of the surveyed soil. A preliminary experimental result assessing the advantages offered by the proposed strategy is herein provided.