Increase of information by polarimetric remote sensing

Abstract

Polarimetric systems promise to be a valid method of providing more information for radar remote sensing. The gain by using polarimetry with respect to a single polarisation system, is reduced due to correlations between the single channels and due to noise. This reduction of performance is subject of this paper. Unlike in communication theory, the information content of radar data is not any more only a function of the signal to noise ratio and signal to depolarisation ratio, but it depends also on the correlations between the single channels, hence the capacity depends on the observed targets. This information content allows a definition of the maximum number of distinguishable classes in the given data. The usefulness of this approach to divide between data and information is shown with an examples. Its validity is demonstrated with double band full polarimetric SIR-C data.