Inversion Technique for PolSAR Parameters Enabling Quantitative Comparison in Space and Time

The space-based polarimetric synthetic aperture radar (PolSAR) image has been expected as a quantitative long-term monitoring tool owing to its distinctive reproducibility and sensitivity. This study devised an inversion technique of physical surface parameters from PolSAR data by considering an axis of comparison so as to realize a quantitative comparison in space and time. First, it was shown that the quantitative change monitoring cannot be achieved via popular PolSAR decomposition techniques owing to the unstable axes of comparison. Then, a novel inversion algorithm was devised by brute force with a solid axis of comparison in combination with an evaluation utilizing a feature of the Hermitian matrix. The technique was verified using numerical simulation for various surfaces such as vegetation, urban, and sea surfaces, demonstrating unique inversion with reasonable accuracy. The technique was also validated using real PolSAR images obtained by Advanced Land Observation Satellite (ALOS-2). Various objects located in different places were selected from a scene in San Francisco, whereas a deciduous forest was used in Hokkaido, Japan, to examine the time-variation of the parameters. The results were promising based on a scattering theory of PolSAR. Moreover, some distinguished statistical features were recognized in the forested area and should be studied in the future.