Sub-band interferometry on polarimetric SAR dataset

Complex SAR image spectral analysis has been shown to provide interesting information about the scattering media, highlighting differences in behavior between homogeneous areas with a fully-developed speckle and point scatterers potentially presenting a more coherent radar spectral response. In the line of previously published works by other groups dealing with coherent scatterers, we have studied the statistics of their occurrence as a function of several parameters like frequency, resolution or polarization. The first step of our procedure consists in selecting two or more sub-bands in the Doppler spectrum (in the azimuth direction) or in the distance direction. The sub-bands are subsequently frequency-shifted in order to provide a nonzero correlation. The resulting inter sub-band interferograms are explored to identify the points presenting a significant coherence and the relevance of the associated phase is investigated. The number of detected coherent scatterers, and the associated complex correlation coefficient is analyzed, with respect to the different polarizations, the width of the sub-bands and the distance between the sub-bands. Different polarimetric high resolution SAR datasets acquired with the ONERA airborne system RAMSES have been used. The large panel of RAMSES acquired datasets allows the exploration of the influence of frequency (P, L and X bands) and landscape. The theoretical approach will be detailed, and experimental results will be discussed.