A study on spatio-temporal filtering in the spirit of SqueeSAR

Several years ago an advanced approach for processing InSAR time series was introduced under the name SqueeSAR. At its core is an estimation procedure for the phase history of distributed scatterers (DS), which is based on the covariance matrix of the complex pixel values over time. As in practice we often encounter the situation that data do not behave perfectly Gaussian, we started to investigate an augmentation of the under-lying stochastic model. In this paper we present new results from tests of a large number of parameter settings and give recommendations for their choice.