The {Robust} {InSAR} {Optimization} {Framework} with {Application} to {Monitoring} {Cities} on {Volcanoes}

This paper introduces the Robust InSAR Optimization (RIO) framework to the multi-pass InSAR techniques, such as PSI, SqueeSAR and TomoSAR whose current optimal estimators were derived based on the assumption of Gaussian distributed stationary data, with seldom attention towards their robustness. The RIO framework effectively tackles two common problems in the multi-pass InSAR techniques: 1. treatment of images with bad quality, especially those with large uncompensated atmospheric phase, and 2. the covariance matrix estimation of non-stationary distributed scatterer (DS). The former problem is dealt with using a robust M-estimator which effectively down-weight the images that heavily violate the model, and the latter is addresses with a new method: the Rank M-Estimator (RME) by which the covariance is estimated using the rank of the DS. RME requires no flattening/estimation of the interferometric phase, thanks to the property of mean invariance of rank. The robustness of RME is achieved by using an M-estimator, i.e. amplitude-based weighing function in covariance estimation. The RIO framework can be easily extended to most of the multi-pass InSAR techniques.     «

This paper introduces the Robust InSAR Optimization (RIO) framework to the multi-pass InSAR techniques, such as PSI, SqueeSAR and TomoSAR whose current optimal estimators were derived based on the assumption of Gaussian distributed stationary data, with seldom attention towards their robustness. The RIO framework effectively tackles two common problems in the multi-pass InSAR techniques: 1. treatment of images with bad quality, especially those with large uncompensated atmospheric phase, and 2....     »