Recent Advancements in Quantitative Full-Wavefield Electromagnetic Induction and Ground Penetrating Radar Inversion for Shallow Subsurface Characterization

Abstract:

Ray-based or approximate forward modeling techniques have been often used to reduce the computational demands for inversion purposes. Due to increasing computational power and possible parallelization of inversion algorithms, accurate forward modeling can be included in advanced inversion approaches such that the full-wavefield content can be exploited. Here, recent developments of large-scale quantitative electromagnetic induction (EMI) inversion and full-waveform ground penetrating radar (GPR) inversions are discussed that yield higher resolution of quantitative medium properties compared to conventional approaches due to the use of accurate modeling tools that are based on Maxwell’s equations. For a limited number of parameters, a combined global and local search using the simplex search algorithm or the shuffled complex evolution (SCE) can be used for inversion. Examples will be shown where calibrated large-scale multi-configuration EMI data measured with new generation multi-offset EMI systems are inverted for a layered electrical conductivity earth, and quantitative permittivity and conductivity values of a layered subsurface can be obtained using on-ground GPR full-waveform inversion that includes the estimation of the unknown source wavelet. For a large number of unknowns, gradient-based optimization methods are commonly used that need a good start model to prevent it from being trapped in a local minimum. Examples will be shown where the non-linearity invoked by the presence of high contrast media can be tamed by using a novel combined frequency-time-domain full-waveform inversion, and a low-velocity waveguide layer can be imaged by using crosshole GPR full-waveform inversion, after adapting the starting model using waveguide identification in the measured data. Synthetic data calculated using the inverted permittivity and conductivity models show similar amplitudes and phases as observed in the measured data, which indicates the reliability of the obtained models.