Analysis of radar response from urban areas

Abstract

The output from regular mapping and monitoring of urban areas provides an important source of information for urban planners and decision makers. The use of remotely sensed data to provide this information has been successful in particular environments but has had only limited success in tropical zone countries where cloud and rain often restrict the useful acquisition of visible/infrared image data on a regular basis. In many cases, and particularly in east Asia, these are precisely the areas that most need the data. A number of researchers have examined the potential of using radar images to overcome these problems, because at the wavelengths used (X- to P-), radar is not affected by cloud or rain. Urban areas are a spatially complex mixture of both natural and built surfaces whose spectral and geometric properties are many and varied. Buildings for example, cause significant backscatter when irradiated by microwave radiation, which is dependent on wavelength, polarisation and incidence angle of the radar beam, and roughness, dielectric properties and size, shape and orientation of the buildings and their surface facets. To some extent all combinations of specular and diffuse backscatter are a function of the height and width of buildings, and thus give rise to the possibility of using backscatter as a measure of the bulk density of the built environment. Equations for backscattering mechanisms, often found in urban environments, are well known. These are for example, facets, point scatterers, dihedral and trihedral corner reflectors, cylinders and wedges. This paper examines the theoretical relationships between urban morphology and remote sensing response at radar wavelengths, provides some preliminary results on measures of urban classification using AirSAR quad polarised radar data from test sites over the city of Sydney, Australia, and proposes a solution to the problem of backscatter variation due to building orientation.