Novel approaches of borehole-GPR data processing and visualization – application for unexploded ordnance detection

Schmidt, Volkmar ; Bredeck, André

Abstract

Borehole ground-penetrating radar (BGPR) measurements allow for the detection of objects and structures in the subsurface and are often applied to the detection of unexploded ordnance (UXO). If omnidirectional borehole antennas in reflection mode are used for the measurement, the localization of UXO is only possible if the data from a multitude of boreholes are analysed. Data analysis is usually still done by manual picking of reflections. We propose novel approaches to process and visualize data from BGPR measurements in a more advanced and appealing manner. Therein, the reflected energy recorded in the radargrams is projected back to all potential reflection points in the three-dimensional space around the boreholes. If the projection direction is considered, we obtain a vectorized energy projection image. Superposition of projected energy yields an easy-to-grasp indicator of possible locations of UXO and of regions of interest that ought to be investigated in more detail. These approaches have been applied to synthetic data and to data measured on a test site with buried UXO. The results show that energy projection is a useful tool for BGPR data visualization, although the result is dependent on data pre-processing. The proposed methods provide novel representations of BGPR data based on an objective algorithm which will at least complement the conventional methods.