Ground Penetrating Radar (GPR) datasets from 3D printed heterogeneous fracture aperture

While its effectiveness is well established, accurately characterizing heterogeneous aperture fields remains challenging. We address this through a controlled experiment analyzing GPR responses to a synthetic heterogeneous fracture.

We used spatially correlated random fields to model rough-wall surfaces separated by a mean aperture, creating 3D-printed moulds and casting gypsum blocks to mimic a true fracture. Then, we collected GPR reflection data on a 2D grid over the fracture and extracted both relative and absolute reflection coefficients. We demonstrate that GPR can infer fracture apertures even in the presence of heterogeneity.