Quantitative Long-Term Change Monitoring by Maximizing Polarimetric Sensitivity

Owing to its high reproducibility, synthetic aperture radar (SAR) is a reliable change detection tool. Although several maximization techniques for maximizing the time variance of polarimetric SAR (PolSAR) data have been proposed, quantitative long-term change monitoring has never been achieved owing to the infeasibility of selecting a pair of polarization states for the transmitting and receiving antennas and their variation during monitoring. This study introduces polarimetric sensitivity, an index characterized by the time variation of received power. Polarimetric sensitivity can be maximized by selecting a physically realizable pair of polarization states. Then, the selected pair, i.e., an axis of comparison, is maintained constant throughout the observation period so that long-term continuous change can be quantitatively measured. The technique was verified first using a numerically simulated scattering power from a simple oriented dipole and then by analyzing real time-series PolSAR images obtained by the second Advanced Land Observation Satellite (ALOS-2). Observing the seasonal changes in a broad-leaf forest in Tomakomai Experimental Forest (TOEF) in Hokkaido, Japan, from winter to summer and the secular changes in the earthquake disaster area in Kumamoto, Japan, confirmed that the sensitivity of each pixel in the proposed scheme was maximized, and the received power reconstructed by the optimized polarization states exhibited continuous variation in time. Moreover, the optimization technique revealed that co-polarization was considered for most pixels in the forested and disaster regions to achieve optimal sensitivity.