Coherence linearity and SKP-structured matrices in multi-baseline PolInSAR

In this paper we examine two different approaches to the estimation of ground and volume interferometric coherences from single and multi-baseline PolInSAR data. One approach, largely retained in literature, is to force the hypothesis that, in each interferogram, the variation of the interferometric coherence with polarization gives rise to a line in the complex plane. The other approach is to process all interferograms jointly by assuming that the data covariance matrix is structured as a Sum of Kronecker Products (SKP). Those two approaches are compared from the algebraic and statistical point of view. Results from simulations show that joint processing all avail able interferograms improves coherence estimation, even though a careful implementation is required to avoid underestimation of coherences close to one in magnitude.