Inversion of dispersive multioffset GPR data for thin-layer characterization : application to fracture properties estimation - BRGM - Bureau de recherches géologiques et minières Access content directly
Conference Papers Year : 2011

Inversion of dispersive multioffset GPR data for thin-layer characterization : application to fracture properties estimation

Stéphane Garambois
  • Function : Author
  • PersonId : 922496
Francois Lavoué
  • Function : Author
  • PersonId : 892142
Romain Brossier
Jean Virieux
  • Function : Author
  • PersonId : 857422


In the presence of a thin-bed, i.e., when a layer is thin compared to the wavelength of the seismic or electromagnetic signal propagating through it, it is impossible to detect and individualize reflections coming from the two surfaces forming the thin layer. In this case, multiple reflections coming from the two sides of the bed create interferences and generate complex reflection patterns. When multioffset data are available, Amplitude Variations of reflectivity according to Offset (AVO) have been increasingly used in seismic interpretation and more recently tested on Ground Penetrating Radar (GPR) data. Phase and frequency sensitivities of the reflected signals are generally not used, although they contain useful information. The present study proposes an original inversion methodology designed for 1D media and denoted DAPVO (Dispersion of Amplitude and Phases Versus Offset curves). It combines all reflectivity properties (amplitude, phase and dispersion) of a reflected GPR signal generated by a thin-bed embedded within a homogeneous material, a situation often encountered in fractured media, for example. Dispersive properties of the dielectric permittivity of investigated materials (homogeneous formation, thin bed) are described using a Jonscher parameterization, which permitted study of the sensitivity of amplitude and phase variations with offset (APVO) curves on frequency and thin-bed properties (filling nature, aperture). We discuss and illustrate using synthetics the simplifying assumptions and careful corrections which are necessary to convert raw Common Mid-Point (CMP) reflected data into dispersive APVO curves. In particular, the propagation and radiation pattern corrections are made within the inversion process for all tested models of the search grid, which is based on the neighbourhood algorithm strategy. Within this work we notably show with various analytical data (different apertures and fillings) that the inversion process is efficient to provide not only the properties of the thin-bed, but also its depth and the properties of the surrounding homogeneous soil. This approach was successful thanks to the introduction of the dispersion characteristics, which restrict non-uniqueness problems potentially encountered when only AVO curves are used. Finally, all the methodology is successfully applied to a synthetic and to two real CMP GPR datasets, which were acquired along a vertical limestone cliff. It allowed extracting the characteristics of various subvertical fractures (filling, aperture) with satisfying resolution and confidence. The study motivates interest to combine dispersion dependency of the reflection coefficient variations with classical AVO analyses for any thin-bed characterization. The inversion strategy presented here is restricted on a single reflected event embedded within a homogeneous formation. In order to develop this approach to potential other applications, when various reflected events are present and/or when properties do vary laterally (2D media), a combination of velocity model building by Full-Waveform Inversion and high resolution Reverse-Time-Migration is currently investigated.
No file

Dates and versions

hal-00873260 , version 1 (15-10-2013)


  • HAL Id : hal-00873260 , version 1


Stéphane Garambois, Jacques Deparis, Francois Lavoué, Romain Brossier, Jean Virieux. Inversion of dispersive multioffset GPR data for thin-layer characterization : application to fracture properties estimation. AGU Fall Meeting 2011, Dec 2011, San Francisco, United States. ⟨hal-00873260⟩
220 View
0 Download


Gmail Facebook X LinkedIn More