Amplitude inversion analysis to estimate the soil permittivity from UAV-based GPR data sets

This paper concerns the reconstruction of the subsurface permittivity distribution from drone-based Ground Penetrating Radar (GPR) data sets by applying an amplitude inversion algorithm, which was initially developed in the context of ground-based surveys. The inversion procedure iteratively recovers the thickness and electromagnetic (EM) velocity in each identified layer, by automatically reconstructing the travel paths of the reflected signals in the individually-analyzed GPR traces. The subsurface permittivity distribution is subsequently obtained from the estimated EM velocities. The inversion algorithm requires the EM velocity in the shallowest layer as input. Therefore, drone-based GPR profiles reduce the uncertainty in the reconstructed soil permittivity models, when compared to ground-based data sets of a similar quality, by eliminating the uncertainty in the input EM velocity. This is because the shallowest layer in drone-based surveys is made of air, for which the EM velocity is precisely known. The presented robustness analysis involves three synthetic GPR profiles that are obtained from the same subsurface model by using antennas with different central frequencies and following different flight paths. This preliminary analysis shows that the inversion algorithm performs well overall, indicating its applicability to drone-based GPR data sets. Indeed, the inversion results are consistent with the original model in all three cases. The presented analysis also critically discusses the advantages and limitations of the procedure.