ABSTRACT
Imaging radar is a unique remote sensing system in that it uses its own source of target illumination, therefore providing imagery independent of solar illumination, and operates at a wavelength long enough to be able to penetrate clouds, making it insensitive to weather. The raw ground resolution of an imaging radar is far too coarse to be useful in identification of terrestrial targets, but mathematical recombination of all radar returns from a target while it is in the field of view of the sensor allows the computation of a synthetic aperture many kilometers long, and hence improves the resolution of the sensor to a few meters. Multichannel synthetic-aperture radar (SAR) is achieved through the sending and receiving of different polarizations of radar signal. After suitable noise filtering, polarimetric SAR responses can be decomposed to infer scattering types: surface, dihedral, and volume scatterers. From these decompositions, traditional classification techniques may be used to identify features on the ground, both discrete scatterers—strongly reflecting point objects like towers, poles, or other man-made structures—and distributed scatterers—fields, forests, and other natural environments. Examples are given, including identification of distributed scatterers in a region of Chinese Inner Mongolia, invasive weed growth in a prairie region in southern Alberta, Canada, and oil and gas infrastructure in central Alberta, Canada.
