3D SAR coherent change detection for monitoring the ground under a forest canopy

Abstract

A method of applying repeat-pass synthetic aperture radar coherent change detection to forest-covered terrain is proposed and analysed. Backscatter from the canopy can be suppressed whilst the complex reflectivity of the ground is preserved by vertically beamforming multiple across-track image channels acquired on each pass. A multichannel coherence metric is formulated assuming a dual-layer (surface and volume) model of the scene. The optimal volume suppression is derived and then quantified using the interferometric random-volume-over-ground model for all pairs of channels. It is shown that good volume suppression is achievable with just three channels, and the resulting coherence estimation is reasonably accurate if the effective ground and volume scattering responses are of similar strength. Significant degradation occurs if knowledge of the ground height is inaccurate. Two aspects of radar design are discussed in detail: the selection of a wavelength that permits both adequate foliage penetration and high sensitivity to subtle changes in the ground clutter, and the sensitivity above noise required to measure the ground backscatter, which is typically weak at foliage-penetrating wavelengths.