Soil permittivity and conductivity characterization by full-wave inversion of near-field GPR data

Full-wave inverse modeling of low-frequency, nearfield ground-penetrating radar (GPR) data was used for simultaneously reconstructing both the electric permittivity and conductivity of the soil. Low GPR frequencies provide a significant sensitivity of the reflection coefficient to electrical conductivity and are less affected by soil roughness and local heterogeneities. Based on the near-field model, several numerical experiments were conducted to simultaneously retrieve ground electrical conductivities and dielectrical permittivities in the range 10-180 MHz for different water contents. We calibrated a time-domain GPR system equipped with transmitting and receiving 80 MHz unshielded dipoles antennas using measurements collected at different heights over a water layer of known electrical conductivity. Then, the GPR model was validated for measurements collected over water subject to a range of electrical conductivities. A good agreement was obtained between the radar data and the fullwave electromagnetic model for the different antenna heights but the water layer properties were not accurately retrieved. These differences were attributed to errors in the transfer functions of the antenna mainly due to the instability of the radar system. The future challenge in this research will focus on an accurate determination of the antenna transfer functions of a stable radar system for improved medium reconstruction.