Abstract
Interferometric Synthetic Aperture Radar (InSAR) is used to generate Digital Elevation Map (DEM) from multiple high resolution SAR images. The phase values of the complex reflection images are used to yield scattering phase center (SPC) which is then used to extract elevation information for each pixel of an image. Interferometry assumes that the return from a pixel can be associated with a single SPC at a fixed elevation. This is not the case for forested terrain as elevation of vegetation and elevation of ground results in two SPCs leading to errors in DEM estimates of the terrain. Fully polarimetric SAR (PolSAR) measurements have the potential to separate different scattering mechanisms based on their polarimetric signatures. The multiple signal classification (MUSIC) algorithm is an array based direction of arrival (DOA) estimation technique that exploits eigen structure of an input data matrix. It can be implemented to provide asymptotically unbiased estimates of number of signals, DOAs and polarizations. In this thesis, MUSIC is used by incorporating polarimetry to separate the SPCs associated with ground and vegetation components in simulated polarimetric SAR interferometry (Pol-InSAR) measurements. The sensitivity of the MUSIC algorithm to noise and vegetation level is also investigated. The probability of detection of SPC associated with vegetation component decreases with decreasing SNR. Furthermore, the error in estimate of SPC for ground component increases with an increase in vegetation level due to higher vegetation bias. This MUSIC approach is verified by using L-band Pol-InSAR field measurements from Glen Affric region.