Estimating dielectric permittivity and electric conductivity from simulated multichannel gpr pulses using aco and quasi-newton inversion techniques

Abstract

Inversion of synthetic ground-penetrating radar data to estimate both dielectric permittivity (ε) and electric conductivity (σ) properties simultaneously is presented in this paper. The synthetic Ground-Penetrating Radar (GPR) data was generated by the propagation of a one-dimensional electromagnetic wave (1-D EM wave) through a given geological model. The simulated EM trace was modeled by Finite Difference Time Domain method (FDTD) for three different frequencies (f): 800, 1000 and 1200 MHz. Random noise was also introduced to evaluate inversion algorithm performance. The inversion of GPR data was performed by Ant Colony Optimization (ACO) and Quasi-Newton (QN) techniques. A modified ACO technique was applied to approximate conductivity for deepest positions, and to increase the accuracy and convergence along lower positions. The inversion techniques were able to estimate simultaneously the dielectric permittivity and electric conductivity from synthetic multi-frequency GPR data. The estimated electrical parameters can be used to derive a set of physical properties and to develop a better understanding of the underground geological or geotechnical media.