Modeling of crosshole ground-penetrating radar data

The ground-penetrating radar (GPR) that is one of the non-invasive electromagnetic methods of applied geophysics is widely used to image shallow subsurface with extremely high resolution. The resolution and depth being two important aspects in a GPR survey are affected by the water, clay, soluble salt contents of soils and the center frequency of antenna. It may be difficult to obtain a good subsurface image at desired resolution and targeted depth in the areas characterized by high electrical conductivity. Therefore, a GPR survey based on the crosshole configuration can be a good alternative approach to achieve more detailed subsurface radar velocity distribution. In this study, firstarrival traveltimes being essential for tomographic inversion of crosshole GPR data sets were calculated by a finite-difference timedomain (FDTD) solutions of Maxwell's equations and finite-difference solution of the Eikonal equation throughout a gridded velocity field. Two theoretical subsurface models were used in modeling. In the first model, the subsurface divided into two layers. The second model includes low-and high-velocity blocks embedded in a homogenous medium. The effect of ground-air interface in modeling and the importance of the ratio between separation and depth of boreholes in a crosshole radar survey were also shown during the test studies. Radargrams consisting of the vertical component of the electric field (Ez) recorded in time at the entire receiver locations were acquired from FDTD modeling. Traveltime contour maps for source locations with different depths were obtained from a fast finite-difference Eikonal solver. Raypaths having the minimum traveltime were then calculated by following the steepest gradient direction from the receiver to the transmitter. As a result, the first-arrival traveltimes obtained from both modeling approaches are quite compatible with each other. FDTD modeling is an important tool to determine and evaluate of the wave phases corresponding to the first arriving wave. On the other hand, Eikonal-equation-based modeling presents an approach being highly effective for directly computing first-arrival traveltimes.