| Ground subsidence has characteristics of hysteretic reaction, slow developing, but this slow chronic geological disasters can bring serious security problems, such as building collapse, underground pipeline damage, thus threating people’s life and production safety. In China, ground subsidence mainly occurs in the Yangtze River delta, the North China Plain and the Fenwei Basin area. The North China Plain has the most serious ground subsidence. With development of economy and increment of population, the excessive exploitation of underground water leads to the formation of groundwater funnel, which seriously affects the sustainable development of the city.DInSAR has the capabilities of all time and all weather imaging and high resolution observing, which make it become an effective method for regional surface subsidence monitoring. In recent years, with the successful launch of the high resolution satellite, such as TerraSAR-X and COSMO-SkyMed, the multi-temporal radar interferometry (MT-InSAR) has widely used to monitor urban subsidence, and can enhance the density of the coherent point target (CT) and the accuracy of subsidence measurements. However, due to the spatial and temporal decoration and other factors, the non-urban has relatively less coherent target, which decreases deformation details.In order to obtain the surface deformation information with high accuracy and high spatial coverage, on the basis of MT-InSAR technology, this thesis introduces the non-local filtering method to improve the interferometric phase’s quality to increase the recognition of coherent target in non-urban area. The details are as follows:(1) In order to solve the problem of the loss of edge information in the processing of high resolution SAR image by the traditional filtering method, in this thesis, the non-local filtering method considering both the amplitude and phase information of the image is used to improve the quality of the interferometric phase. It extends the filtering coverage from local area to the overall image field, and makes full use of the redundant information in the image to better retain the edge details, which can be more effective to restore interferometric fringes in low coherence regions.(2) Using simulated data and real data to test the adaptability of non-local filtering for high resolution SAR images. The simulation data and TanDEM-X data are tested by using multilook, adaptive filtering and non-local interferometric filtering. Compared with the results of other two filtering methods, non-local filter performance better in the suppression of noise level and the maintaining of the details of the image.(3) Because in the non-urban area decorrelation is more serious, it is often impossible to detect the deformation information of such region. On the basis of conventional MT-InSAR technology, the non-local filtering method is introduced to improve the quality of each interferogram, so as to improve the spatial sampling rate and deformation estimation accuracy of coherent targets in the non-urban area.(4) Ground deformation monitoring experiment was carried out by using the 39 TerraSAR-X images collected from April 2009 to December 2010 in Tianjin area. Both the conventional MT-InSAR and the MT-InSAR based on the non-local filtering (NLMT-InSAR) were used for deformation extraction. Comparative analysis shows that, CT density detected by NLMT-InSAR was increased by 41%. NLMT-InSAR has higher spatial sampling rate and more comprehensive deformation information in non-urban areas. The results of MT-InSAR and NLMT-InSAR are verified by the leveling data, the root mean square error are 4.75 mm and 3.16 mm respectively, demonstrating that the NLMT-InSAR method has higher accuracy.Theoretical analysis and experimental studies have shown that in the case of non-urban region’s spatial and temporal decorrelation, the MT-InSAR method based on the non-local filtering can improve the accuracy of deformation estimates and spatial coverage of the CT in the region. |
PDF Full Text Request