Advances in meter-resolution synthetic aperture radar interferometry

Synthetic aperture radar (SAR) interferometry is the technique of using SAR as interfer-ometer, to measure the phase difference caused by topography or object displacement. Multipass SAR interferometry (InSAR) is so far the only method for assessing long-term millimetre-level deformation over large areas from space on an imaging basis. Very high resolution multipass InSAR techniques have made substantial development in the last dccadc in monitoring individual building using persistent scatterer interferometry (PS1) and SAR tomography (TomoSAR), in monitoring nonurban area using small baseline subset (SBAS) and SqueeSAR, and so on. To prepare for future SAR missions which will have higher resolution, larger coverage, and greater data volume, this thesis addresses the following four aspects of multipass InSAR techniques: computational efficiency, information fusion, contextual awareness, and statistical robustness. TomoSAR is the most competent InSAR method for urban area monitoring. But it is much more computationally intensive than any other multipass InSAR methods. To this end, an computationally efficient multi-dimensional TomoSAR algorithm for urban area was developed, by integrating tomographic SAR inversion and the well-known PSI. The results of PSI were uved for a pre-classification of single and double scatterers, and also used as prior in the tomographic inversion. Real data experiments show the proposed approach obtains results comparable to the one obtained by the most computationally expensive tomographic SAR algorithms (e.g. SI.1MMF.R), and save∗ computational time by a factor of fifty. Multi-aspect InSAR point clouds fusion is required for a complete monitoring of entire city, due to the SAR side-looking geometry. In this thesis, a robust algorithm, namely "L- shapc detection & matching", is proposed, especially for fusing two point clouds from cross- heading orbits, i.e. ascending and descending. The main idea of this algorithm is finding and matching the theoretically exact point correspondence which is the end positions of facades where the two point clouds close. Practical experiment shows the proposed method achieves sub-meter consistency with the state-of-the-art, and is much more computationally efficient. The development of a semantic urban infrastructure monitoring algorithm bp fusing InSAR and optical images was followed after the point cloud fusion. The attributes derived from optical images, e.g. colour, classification label, are transferred to the InSAR point cloud for a semantic level analysis of the deformation signal. The key lies on a strict 3-D geometric co-registration of SAR and optical images by reconstructing and matching the 3-D point clouds derived from the two types of images. Examples on bridges and railway monitoring arc demonstrated. Robust InSAR optimization is crucial for future SAR data, because of the unprecedented high resolution brings different observation statistics and much more dynamic interfero- metric phase. The proposed robust InSAR optimization (RIO) framework answers two open questions in multipass InSAR: (1) How to optimally treat images with a large phase error, e.g., due to unmodeled motion phase, uncompensated atmospheric phase, etc.? And (2) How to estimate the covariance matrix of a non-Gaussian complex InSAR multivariate, particularly those with nonstationary phase signals? For the former question, RIO employs a robust M-estimator to effectively down-weight these images, and for the latter question, a new method - The rank M-Estimator - is proposed. Simulated and real data experiments demonstrated that substantial improvement can be achieved in terms of the variance of estimates, comparing to the state-of-the-art estimators for both persistent and distributed scatterers. The proposed framework can be easily extended to other multipass InSAR techniques. The abovementioned algorithms were tested using TeiraSAR-X data of various test sites, especially for the efficient TomoSAR algorithm and the RIO framework.