Research On Hierarchical And Block Estimation Method For PS-InSAR Technique Towards Deformation Monitoring In Large-scale Areas

As a new geodetic technique,interferometric Synthetic Aperture Radar(In SAR)technique has become one of the standard techniques for large-scale deformation monitoring because of its all-weather,all-weather,large-scale and high-precision characteristics.As a representative,persistent scatterer interferometric synthetic aperture radar(PS-In SAR)technique can achieve a refined ground deformation measurement results by taking advantage of the full-space-resolution of SAR images,and has algorithm advantages unmatched by other In SAR technologies.So PS-In SAR technique has been used for many ground deformation monitoring caused by natural or human activities.But limited by the algorithm design,classical PS-In SAR method is only suit for small areas.With the rapid development of SAR satellite technologies and the increase of application requirements,PS-In SAR thchnique has encountered key bottlenecks in accuracy and efficiency,which can be summarized in the following two aspects:(1)With the improvement of SAR image quantity,resolution and ground coverage,ground observation targets increase exponentially,In SAR technology has entered the era of big data.Classical PS-In SAR algorithm is difficult to adapt to the new development status and cause the problem of decreasing precision in deformation monitoring in large-scale areas.(2)With the deeper understanding of surface displacement,deformation monitoring mission has been expected to be performed in cities,provinces and even countries.How to effectively reduce the computation time and memory requirements has been one of the major obstables for PS-In SAR in large-scale areas.The development of sersor and requirement also provides a new chance for the improvement of PS-In SAR algorithm.In order to solve the problems of accuracy and efficiency of PS-In SAR technique in large-scale deformation monitoring,this thesis take PS-In SAR algorithm flow as the breakthrough point and focuses on networking construction,deformation estimation and deformation time series optimization.These researches establish the whole technique process of the hierarchical and block estimation method for PS-In SAR technique,and provide new routines for large-scale deformation monitoring of PS-In SAR thchnique.The main contributions and innovarions of this thesis are as follows:(1)A hierarchical and block network constructing method for PS-In SAR technique towards large-scale deformation monitoring is proposed.This method can suppress error propagation effectively and meanwhile increase the computational efficiency significantly.In large-scale areas,PS points with different phase quality have obvious mixed distribution.Classical method with a unified network and only one control point may cause error propagation and lead to the decrease of accuracy.Furthermore,massive monitoring targets need overmuch time and memory requirements,which can lead to low efficiency or even cannot be performed.Based on the theories of traverse survey,this thesis proposed a hierarchical and block network constructing method for PS-In SAR technique.Firstly,the whole study area is partitioned into several regular blocks by considering the effect of spatial atmospheric delays;Then,a two-steps is used to select the control point for each block.Every block is constructed in secondary PS networks independently,and all control poings is used to connect the high-order network.Points in the high-order network are used as reference point for the secondary networks.By the hierarchical network and step-by-step constraints,the result accuracy can be ensured.Furthermore,the block estimation can be performed independently and then we can reduce computing time and memory consumption drastically by parallel process.This method has been used in southern California with more than 13,000 km~2 area and 20000000 PS targets.Experimental results show that the deformation rate accuracy is better than 2mm/year and deformation time series accuracy is 3.7mm.Compared with classical IPTA method,data processing time consumption reduced by about 40 times.(2)We proposed a novel PS-In SAR deformation estimation method for large-scale area based on smooth constraints of are deformation phase in time domain.The new method solves the problem of incomplete deformation time series restoration in classical method which use the stractecy of modeling deformation and separated of residual deformation by spatial-temporal filtering.The new method donot relay on the deformation model and can estimate deformation directly,which improve the flexibility and efficiency of PS-In SAR deformation estimation.In classical method,deformation model is used to fitting the real ground deformation,and model parameters is calculated in the method.Then the unfitted deformation is separated from residual deformation by spatial-temporal filtering.In large-scale area,there are various types of deformation and fixed model fitting can cause inapplicability problem.Moreover,unfitted deformation is not always a low-frequency signal in space or time domain,accumulated deformation calculateing may loss durong spatial-temporal filtering process.In this thesis,we analysis the time domain variation law of each component in are phase and interferometric phase integral combination method and single parameter solution space searching method is used to separate and unwrap phase caused by topographic error.Then a direct detection based on smooth constraints of are deformation phase in time domain and correction is used to unwrapping different deformation phase.Finally,deformation time series can be extracted directly by least square network adjustment method.The synthetic data sets test and real data sets experiment show that,deformation time series restoration accuracy is better than 5mm by using the new method and the deformation rate estimation accuracy is improves by 34.38%.Moreover,the new method has less time and computer memory consumption.Compared with classical PS-In SAR method,time consumption reduced by about 40 times and about 20 times in simulater data experiment and real data experiment,respectively.(3)We peoposed a PS-In SAR deformation time series post-processing method considering stationarity of time series.The new method solves the problems that it is difficult to determine the optimal filtering window and the results is more sensitive to noise level for the traditional filter method and achieves the high-precision and mass deformation time series extraction for large-scale area.There are various types of deformation in large-scale areas,the optimal window of traditional filter method changes between different time series and between different period in the same time series and it is very easy to cause the residual noise or deformation loss.In addition,the results are changed under different noise level.So the traditional fiter method could not be used in deformation time series mass post-processing.In this paper,we analysis the common change characteristics of ground deformation and build a model library for large-scale area which contain common ground deformation model.Then twice fitting process are performed by considering goodness of fit and time series stationarity to realize the transformation of sequence from non-stationary to stationary.Finally,a large-window filtering is used to extract residual deformation signal and the whole deformation time series can be obtained precisely.Experimental results show that the new method can obtain more accuracy results in both regular and irregular sequences than classical filter method,and the new method has more superiority in filter window selection and noise handling ability.