Four-component Scattering Power Decomposition Using Rotated Coherency Matrix

Since the successful launch of ALOS in 2006, we have seen significant advances in the utilization of fully polarimetric SAR (PolSAR) data. We are now in the phase of PolSAR applications for sensing and monitoring the Earth’s environment. Specific application areas include natural and man-made disaster monitoring, forest monitoring, crop assessment, oceanography, etc. In this report, a method of utilization of fully polarimetric data by means of target decomposition is given. Target decomposition still remains a main topic of PolSAR research area. Comprehensive reviews have been presented by Lee [1] and Cloude and Pottier [2]. There are two kinds in the incoherent decomposition. One is the most widely used H/ /A method developed by Cloude and Pottier based on the eigenvalue analysis [3-6]. Touzi recently proposed a new incoherent decomposition by applying the coherent Kennaugh decomposition in Pauli basis [7-8]. The other is a physical scattering model-based method, first developed by Freeman and Durden [9]. Several methods have been proposed to this model-based decomposition [10-14]. These methods rely on the ensemble averaging of the coherency or covariance matrix. In the incoherent analysis for PolSAR data, there are 9 real-valued independent parameters as the second order statistics of polarimetric information [1-2]. The physical scattering model tries to account for these 9 parameters. The three-component decomposition accounts for 5 terms [9]. The four-component scheme accounts for 6 parameters by adding a fourth component, the helix scattering [10]. An [11] and Yamaguchi [12] reduced the number of independent parameter from 9 to 8, by rotation of the measured coherency matrix, leaving unaccounted number as 2. The remaining 2 correspond to the real and imaginary part of T13 element in the coherency matrix, and they are small minor terms. These 2 parameters are not accounted in any physical model-based decomposition up to now [1] for complete matching. In this report, we present some results of the fourcomponent scattering power decomposition with rotation of coherency matrix applied to ALOS-PALSAR polarimetric data sets. We have seen a great improvement in polarimetric image by this new decomposition method. 2. FOUR-COMPONENT SCATTERING POWER DECOMPOSITION