Three Dimensional Migration and Forward Modelling of Ground Penetrating Radar Data.

摘要

Carnegie Mellon University is automating the use of Ground Penetrating Radar (GPR) for the cleanup of hazardous waste sites. The current focus is on the development of an automated subsurface mapping system to locate buried objects and geological structures so that sources and migratory pathways of contaminants can be be identified and catalogued. The subsurface maps are produced using the non-invasive sensing capabilities of GPR, GPR operates similar to conventional radar, but the data acquired is more difficult to process due to the nonhomogeneous nature of the subsurface medium. Though sometimes used in waste site characterization, GPR deployment, data acquisition, and interpretation are human driven processes. The potential of GPR to generate accurate three dimensional subsurface maps has not been fully realized previously. The Site Investigation Robot uses robots to position a GPR transducer to exploit the accurate, repeatable positioning available from automated equipment. By combining the use of a position cognizant, all terrain mobile robot and a linear scanning mechanism, GPR records are acquired in a two dimensional grid on the ground surface. This method of collection simplifies processing and the positional location of features of interest in the data. To achieve accurate positional accuracy of the located subsurface objects, correct modeling of the subsurface medium and the antenna is crucial. The subsurface medium can be modelled as a composition of multiple subvolumes, each one being characterized by its own electrical parameters.