Validation of an automated detection and characterization of diffraction curves in the WISDOM/ExoMars radargrams with a Hough transform
Résumé
The WISDOM (CNES/DLR) Ground Penetrating Radar (GPR), onboard the ExoMars 2022 Rosalind Franklin rover (ESA/ROSCOSMOS), is designed to study the structure and composition of the first meters of the Martian subsurface prior to drilling operations, providing key information on the geological context of the samples to be collected (down to a 2 m depth), and the potential hazards the drill may encounter.
In the inversion process of GPR data, diffraction curves corresponding to underground scatterers are commonly used to retrieve the dielectric constant of the surrounding medium. The Hough transform adapted to hyperbolic shapes has already proven its efficiency in detecting diffraction curves in terrestrial GPR acquisitions, without requiring any learning or training dataset. The method has been adapted to WISDOM radargrams.
WISDOM being an air-coupled GPR, the effect of refraction at surface level on the inversion is quantified and has to be taken into account. The Hough transform process is applied to synthetic WISDOM soundings, first generated with a simple scattering model of a subsurface accounting for refraction (an example is presented in the figure), and then with more complex ground models and FDTD simulations to evaluate its performances in terms of detection and dielectric constant estimation. Eventually, the same test is performed on experimental acquisitions with different models of the instrument, in a semi-controlled environment with buried metallic targets, and in natural environments being potential Martian analogs.
The Hough transform adapted to WISDOM acquisitions provides an automated detection of potential underground obstacles with objective criteria, as well as a retrieval of the dielectric constant of the subsurface, allowing an estimation of their depth. It will significantly help the interpretation of future Martian radargrams, and meet the allowed decision time requirements of the Rover Operation Control Center when selecting a safe and relevant drilling location.