CALIBRATION OF STOKES AND SCATTERING MATRIX FORMAT POLARIMETRIC SAR DATA

Several different approaches to calibration of polarimetric radar image data have appeared recently in the literature. In particular, several papers by JPL authors have offered different perspectives on the polarimetric radar calibration problem. In this paper, we compare two of these approaches and demonstrate their functional equivalence. We show how one of the algorithms can be adapted to calibrate both scattering matrix and Stokes matrix format polarimetric radar data. Klein [1] has presented an approach for calibrating polarimetric SAR data in scattering matrix format, which relies on backscatter reciprocity and the lack of correlation between like- and cross-pol backscatter found in many natural targets. van Zyl [21 has presented an approach to the calibration of Stokes matrix (compressed) format polarimetric radar data, which is based on the assumption of reciprocity between the radar receive (R) and transmit (T) matrix representations. Both these approaches require one or more trihedral corner reflectors within the imaged scene. Freeman et al. [3] have presented an approach based on using three polarimetric active radar calibrators (PARC's) within the imaged scene. In general, techniques such as that described in [31, using three known (man-made) targets located at a single point will not be successful in calibrating all types of polarimetric radar data, because of the possibility of large variations in cross-talk amplitude and phase across the image swath. In this paper, we show that the assumptions about the backscatter and the polarimetric radar system for the two approaches described in Klein's and van Zyl's papers are equivalent. We demonstrate that, to first order in the radar system cross talk (i.e., neglecting terms of second order and above) an exact solution to the Stokes matrix format data calibration problem exists. We call this the first-order solution, for convenience. Next, we show that van Zyl's approach can give this first-order solution for appropriately symmetrized polarimetric radar data. Then we show how, if the data is properly symmetrized, van Zyl's approach can be used to calibrate both scattering matrix and Stokes matrix format data. These conclusions should be generally applicable to polarimetric imaging radar system data. The method is illustrated on NASA/JPL DC-8 SAR data and the results of calibrating in either scattering matrix or Stokes matrix format are shown to be in excellent agreement.