Target decomposition theorems applied to multiple scattering of polarised waves

Recent SAR observations of polarised backscatter from the Greenland Ice sheet have shown anomalous behaviour (Rignot, 1993), with several features in common with earlier earth based radar observations of the icy Galilean satellites (Ostro, 1990). In particular, the observed circular and linear polarisation ratios are much higher than those observed for any other type of natural scattering mechanism and this, combined with a larger than expected reflectivity, has set a challenge for the modelling community. A common feature of all feasible models is the assumption of wave propagation in low loss but highly anisotropic media. This implies that the observed radar signal is the result of many scattering events, each of which is polarisation dependent. That any polarisation structure survives such averaging is indeed surprising but is entirely consistent with recent analytical and numerical studies of enhanced backscatter or weak localisation in random media (Mishchenko, 1993). The present author proposes a probabilistic Target Decomposition (TD) theorem based on eigenvector analyses of the observed coherency matrix. This the author suggests provides a unified framework in which to investigate the polarisation properties of such anomalous behaviour. After outlining the main features of the polarimetric observations the key features of the target decomposition are briefly summarized approach and applied to a simplified model of multiple scattering from low loss particles. In this way a probabilistic model is proposed for the study of depolarisation effects in random media problems.