A new approach for interpretation of SAR-data using polarimetric techniques

Radar polarimetry is a well established technique for classification of land use features. Several investigations have reported the use of polarimetric data to map Earth terrain types and land covers and have advanced a good understanding of polarimetric scattering mechanisms. In this paper 3 different approaches are outlined. The first one is a neural network classification using entropy H and α-angle from the Cloude decomposition theorem (CDT) as input parameters. The second approach is an extended H-α-classification using not only the H and α parameters but also the first eigenvalue. Finally an approach is discussed using the 6 eigenvalues from C-band and L-band. The main advantage of these methods is strong correlation between the results and the physical reality which makes this methods suitable for un-supervised and automatic classification. The potential of this methods is demonstrated using spaceborne data acquired during the 2nd SIR-C/X-SAR mission of the test site Oberpfaffenhofen, Germany (data take 30.00)