Calibration and classification of SIR-C SAR data in moderate slope areas

The goal of this paper is to present the capabilities of SIR-C synthetic aperture radar (SAR) magnitude data (dual frequency, dual polarization) for land cover classification purposes in areas with moderate slope variations. The work consists of two parts. In a first step SIR-C SAR data acquired during the SRL-2 mission over a hilly to mountainous area in Central Switzerland is geometrically and radiometrically calibrated, by taking into account high resolution digital elevation model (DEM) data. In a second step the backscattering coefficient is classified using an unsupervised and a supervised method. The first one is a scale-space clustering algorithm, while the second one is a quadratic classifier. Since topographic variations correspond also to vegetation changes, additional a priori topographic information, such as elevation and slope data, must be included during the classification step. The different classification results are then compared with the digital ground truth. Results clearly shows that, in spite of strong relief-induced distortions, an accurate classification can be achieved, if the SAR data is correctly calibrated and a priori information is used