The impact of high-density spatial sampling versus antenna orientation on 3D GPR fracture imaging

Three-dimensional Ground Penetrating Radar (3D GPR) surveys are necessary to reconstruct complex fracture geometries in the subsurface. Two of the most important factors controlling image quality are antenna orientation relative to fractures and density of acquisition grids. This study, conducted in the Madonna della Mazza quarry (Italy), compares two acquisition methods with the goal of optimizing the imaging of fractures and related 3D fracture networks. We acquired two very dense, orthogonal 3D GPR surveys with a 250 MHz antenna and 5 cm trace spacing on the same day, covering the same area of 20 x 20 m. By decimation of the original raw datasets, reduced survey densities of 10 cm and 20 cm spacing are simulated. The results show differences in the imaging quality of the two methods to depths of 75 cm, while for a depth of 130 cm and deeper, image quality is similar. At the same trace density/m(2), a single, unidirectional survey with a densely sampled grid is the preferred method rather than two surveys with orthogonal antenna orientations but larger profile spacing. The extra effort of conducting surveys with an acquisition grid of an eighth of wavelength of the antenna centre frequency guarantees that we can properly sample the high-frequency content of the GPR signal spectrum and results in optimum image quality regardless of fracture orientation. The simple survey design principle found in this study is universally applicable to any field condition, target geometry, and antenna frequency. Such high-density 3D GPR survey design enables the high-resolution characterization of 3D fracture networks on subsurface timeslices in near photographic quality.