A Physical Approach to POLSAR Time Series Change Analysis

In this letter, we present a new approach to the analysis of polarimetric radar time series, when interest centers on the changes in the full
$n \times n$
coherency matrix of the scene rather than a scalar backscatter coefficient. The key to the new idea is to form matrix differences over time, but always constrained so that the remainder is a physical coherency matrix, that is, formed by some combination of physical scattering mechanisms in the scene. We first detail the algorithm and compare it with other published approaches before illustrating its application to the L-band POLSAR data from the NASA-Jet Propulsion Laboratory (JPL) uninhabited aerial vehicle SAR (UAVSAR) system for study of agricultural changes around Saskatoon, Canada, over a two-year baseline.