Experimental validation of an entropy-based classification scheme using a wide-band polarimetric radar

The extension of an entropy-based classification scheme based on polarimetric SAR for land applications and inversion of biophysical parameters is currently being investigated. In this paper, the potential and limitations of this technique are outlined by using a number of wide-band polarimetric data sets acquired in the European Microwave Signature Laboratory under fully controlled conditions. The targets used in this experimental validations include various tree types and maize plants of different densities and heights. These data sets have been analyzed by displaying the loci of the measured data in the entropy-alpha plane. The application of this scheme leads to interesting results and the subsequent interpretation gives an indication about the dominant scattering mechanism associated to a given target (as a function of the working frequency and the incidence angle). For example, the estimated scattering mechanism of trees with disk-shape (e.g. ficus) or needle-shape (e.g. fir tree) leaves is clearly different and shows an opposite trend in the entropy-alpha space as the frequency is increased. Other interesting results include the different responses from a single tree and a group of trees. A group of fir trees, to take an example, presents a minor spread in the loci, a higher entropy and a lower dependence on the incidence angle compared to an individual tree.