Coherent Scatterers at Different Polarizations and Frequencies Rafael Zandona Schneider, Luca Marotti, Kostantinos P. Papathanassiou Microwaves and Radar Institute, German Aerospace Center, PO BOX 1116, 82230, Wessling, Germany Abstract SAR images of urban areas are characterized by strong geometrical distortions (layover, foreshortening, shadowing) and a high diversity of mainly man-made scattering mechanisms. In order to simplify the analysis of SAR images in urban regions the Coherent Scatterers (CSs) technique has been proposed for the detection of scatterers with point-like behaviour. These deterministic scatterers allow the extraction of information in a simpler and more direct way. In this work, the detection and characterization of CSs at different frequencies and different polarizations is addressed. For that, airborne data acquired at L-, C- and X-band by the E-SAR system of the German Aerospace Center (DLR) over the city of Munich in Germany are used. 1 Introduction The Coherent Scatterers (CSs) technique for the detection of scatterers with deterministic (i.e., point-like) scattering behaviour in SAR images was proposed in [1], [2]. The detection is performed by correlating SAR images from different looks. In section 2 an overview of the technique is provided. In [1] and [2] the characterization of CSs concerning their interferometric and polarimetric properties was evaluated. It was shown that they have in general low polarimetric en- tropy, high amplitude and high interferometric coherence, in agreement to the expected properties of point-like scat- terers. The potential of information extraction from CSs as the orientation about the Line of Sight and the dielectric properties of dihedral-like CSs was also addressed. In [3] a first comparison of CSs at L- and X-band was eval- uated for two urban testsites: Dresden and Oberpfaffen- hofen, both located in Germany. It was shown that when the images have similar resolution more CSs are detected at L- than at X-band. In this paper the detection and charac- terization of CSs and their properties at different frequen- cies (and different polarizations) is further evaluated. 2 Coherent Scatterers Detection The basic CSs approach [2] evaluates the spectral corre- lation between two parts of the full image spectrum (two sublooks) to identify deterministic scatterers. Care should be taken if the spectrum of the original image was weighted by a Hamming window in the processing time. In this case, the spectrum should be first divided by the corresponding Hamming function to obtain the original measured image spectrum. The two sublook images are then formed by di- viding the total spectrum into two parts (and applying a Hamming window in each one in order to suppress side- lobes), and by shifting them to the same central frequency. The coherence between the two sublook images is esti- mated and the pixels with high coherence values are in- terpreted as CSs. 3 The Munich Dataset In October 2005 SAR data was acquired at L-, C- and X- band over the city of Munich in Germany with the airborne system E-SAR of the German Aerospace Center (DLR). The data at L-band are full polarimetric, at C-band dual polarimetric (HH and HV) and at X-band the channels HH and VV were acquired in repeat-pass mode. The resolution in range direction of the SLC images at all frequencies is 1.5m and in azimuth is 0.75m at L-band and C-band, and 0.63m at X-band. A resampling in the az- imuth direction of the L- and X-band data had to be ap- plied in order to correct for the differences in the images sampling due to different airplane velocities during the ac- quisition time. Further, a coregistration procedure among all data was necessary. Fig. 1 shows the Pauli decomposition image at L-band of the Munich city. The azimuth direction is vertical and near range is on the right. On the lower left part of the im- age the big circular region is the Theresienwiese, where still the temporal structures for the Munich’s Oktoberfest were present (the images were acquired one week after the event). Above the Theresienwiese, the Munich central sta- tion and part of the rail tracks can be recognized. On the upper right part of the image, the forested region is the Englischer Garten. The Isar river crosses the lower right region of the image. Note the strong dihedral component (HH-VV) returned from buildings oriented parallel, or almost parallel, to the azimuth flight direction, and that the return from buildings oriented non-parallel to azimuth is dominated by the cross- polarization channel (HV).