Study On Phase Reconstruction Algorithm Of Spaceborne Synthetic Aperture Radar And Its Application

Ground deformation results from artificial factors(e.g.groundwater withdrawal,exploitation of underground mineral resources)or natural factors such as volcanic activity,crustal movement and glacier rafting.This phenomenon could cause severe damage to critical infrastructures(e.g.bridges,roads,oil-gas pipelines and dams),lead to serious environmental problems and economical loss(e.g.ground fissure,saltwater intrusion,and building collapse),poses great threats to normal production and living.Synthetic aperture radar(SAR)interferometric image stacking technique is a novel remote sensing technique that has been developed since the early 2000 s.This technique overcomes the drawbacks of the traditional DInSAR technique(e.g.spatial/temporal decorrelation,atmospheric disturbances,etc.),provides ground deformation measurements in millimeter accuracy and has shown great potential in various deformation monitoring applications.The basic idea of this technique is to form interferograms with different temporal/spatial baselines by utilizing multiple SAR images.This technique precisely separates deformation phase component on pixels with low phase dispersion.Due to the influence of scatterers decorrelation,only persistent scatterers(PS)can be highly coherent in differential interferograms.However,the density of PS is usually low under natural conditions,which produces great limitations on the application of the interferometric image stacking technique.In order to minimize decorrelation between stacked SAR images,the phase reconstruction technique was developed by Guarneri and Tebaldini in 2008.It reforms SAR observations by optimizing a maximum likelihood objective defined with respect to a coherence matrix and minimizes interferometric decorrelation in areas of moderate coherence.In 2011,Ferretti et al.used statistically homogeneous pixels(SHP)to identify the distributed scatterer,then generate coherence matrices.In this way,the loss of PS during the spatial average operation can be avoided.Since then,the phase reconstruction technique has been increasingly used in interferometric applications.Firstly,this thesis thoroughly studies the fundamental theory of SAR and phase reconstruction techniques and discussed the identification method of distributed scatterers is in detail.Secondly,the objective function of phase reconstruction is discussed at the mathematical level,its first order derivative is deduced in detail and on this foundation the frequently-used phase reconstruction algorithm is implemented.The objective function of phase reconstruction is defined on each elements with respect to the coherence matrix,and these elements actually represent the coherent statistical characteristics of the point-targets in corresponding interferograms.Therefore,the phase reconstruction process may be impacted by significantly decorrelated combinations,resulting in a decrease in phase reconstruction quality,leading to the loss of potential highly coherent targets.To address this problem,this thesis improves existing phase reconstruction method based on a simplified coherence network.This approach converts the maximum likelihood estimator to an optimization problem defined on a network.On the premise that the feasibility and the redundancy is guaranteed,the network is simplified by excluding the arcs corresponding to low coherence elements.In this way,the influence of severely biased interferometric phase estimates on phase reconstruction can be diminished.In addition,a targeted iterative strategy is designed based on the idea of phase linking to resolve the phase reconstruction problem with respect to the simplified network.Simulation results reveal that compared to traditional phase reconstruction algorithms,this method has an excellent computational efficiency and the computational accuracy is improved to a certain degree.To further validate this method and its application performance,the Sentinel-1 data acquired over Langfang region is processed in this thesis.Experimental results demonstrate that phase reconstruction method is capable of substantially increasing the number of highly coherent targets in SAR interferometric image stacking techniques and effectively improving the density of final measurements in areas of moderate coherence.