Admittance inversion of GPR transmission for crosshole tomography

We investigated a full waveform inversion algorithm in 2D, that was developed without precise knowledge of the source functions and without any 3D compensation (e.g., point sources) of radar wave propagation. To address this, we proposed a new methodology that measures the electric and magnetic fields generated by a common electric current source, and that inverts the admittance data, the ratio of two mutually orthogonal components of these recorded fields in the frequency domain. To alleviate local minima problems, we devised a data weighting method based on data misfits and data resolution. The developed approach was initially demonstrated by inverting 2D finite-difference time-domain (FDTD) simulation data from a crosshole radar survey. Furthermore, it was demonstrated that numerical data simulated in 3D could be inverted by the developed 2D algorithm. The frequency spectra of the source functions used in the inversion experiments did not agree with the true sources; they were noticeably either higher or lower. The results of the numerical experiments showed that the material property distribution can be calculated with high accuracy, without having accurate source information and without any pre-processing needed to compensate for the characteristics of radar wave propagation in 3D space.

► Inverting apparent admittance data of crosshole radar surveys.
► Full waveform inversion independent of source functions.
► 2D waveform inversion without compensation of 3D nature of radar wave propagation.