Subsurface radar imaging from space

Abstract

Ground Penetrating Radar (GPR) and Synthetic Aperture Radar (SAR) are two widely used techniques for acquiring radar images. GPR, as its name suggests, produces radar images of the below ground environment. SAR is a remote sensing technique which allows moving radar systems to produce radar images with dramatically improved resolutions over conventional radar systems. Despite their benefits, both GPR and SAR suffer from certain limitations. In the case of GPR, the radar system has to be in close proximity with the subsurface volume being surveyed, which limits the process to relatively small areas that are easily accessible. SAR allows large areas to be surveyed rapidly from large distances, but cannot distinguish buried objects from surface objects. This thesis focuses on a radar technique that offers the opportunity to overcome these limitations and allow subsurface radar imaging of large areas using radar data gathered by remote sensing systems. This novel technique is known as Virtual Bandwidth SAR (VB-SAR). VB-SAR utilises changes in soil moisture over a series of SAR images to differentiate buried objects from objects on the surface. In addition to this differentiation, VB-SAR also allows extremely high (centimetre scale) subsurface range resolutions to be obtained from SAR images with range resolutions measured in metres. This research has experimentally demonstrated the basic feasibility of performing remote subsurface radar imaging with the VB-SAR scheme. Within the laboratory environment a buried target has been successfully imaged using VB-SAR and the fundamentals of VB-SAR have been verified. Dramatic increases in subsurface range resolutions have been demonstrated, as has the ability of the VB-SAR scheme to work correctly over a range of radar frequencies, observation angles and polarisations. This laboratory work has been enabled by use of the Tomographic Profiling (TP) imaging scheme. TP is a synthetic aperture based imaging algorithm, but unlike conventional SAR TP produces images with a constant look angle over the entire imaging scene. This enabled the performance of the VB-SAR imaging scheme to be easily evaluated over a range of look angles using a single radar dataset and simplified the experimental setup. In addition to the experimental work, simulation exercises have been conducted and image processors have been implemented. Simulation, using a simulator created as part of this work, has allowed testing of the VB-SAR scheme in a range of scenarios (sidelooking SAR, different soils, multiple buried targets). The image processor work has implemented a high performance TP processor and a practical VB-SAR imager.