Estimating the size of buried scatterers from both ground-based and drone-based ground-penetrating radar data sets

This paper concerns the automated estimation of the size of buried targets from their respective diffractions in both ground-based and drone-based ground-penetrating radar (GPR) data sets. In ground-based GPR surveys, the radius of the assumed spherical or (infinitely long) cylindrical scatterer is recovered by reversing its backward-moving effect on the arrival times of the diffracted signals. Starting from its first break, the presented algorithm rigidly translates the diffraction hyperbola forwards until it can be linearized through a change of coordinates, as it would be with a point-scatterer. The electromagnetic (EM) velocity in the subsurface is accurately estimated from the linear fit, and it is then combined with the applied time shift to estimate the radius. In drone-based GPR surveys, the refractive focusing effect observed at the air-ground interface violates the small-spread approximation, thus preventing the univocal estimation of the radius through the ground-based method. Nevertheless, the algorithm establishes a one-to-one correspondence between the possible radii and EM velocities, based on the link between the shape of the diffraction and the laterally varying transmitted angle at the air-ground interface. The scatterer radius can subsequently be recovered from independent measurements of the subsurface EM velocity. The proposed procedure can be sensitive to significant distortions of the recorded diffraction, which are potentially caused by elongated scatterers, rapid lateral EM velocity variations, signal interference, erroneous trace positioning, random and coherent noise, and excessively low temporal and spatial sampling rates. We apply the scatterer radius analysis to both synthetic and real GPR data sets, showing an overall good performance, and further discuss both the advantages and the limitations of the procedure.