Advanced satellite radar interferometry for small-scale surface deformation detection
|dc.contributor.supervisor||Prof. Peter Lilly|
|dc.contributor.supervisor||Assoc. Prof. Mike Stewart|
|dc.contributor.supervisor||Dr. Zbigniew Perski|
Synthetic aperture radar interferometry (InSAR) is a technique that enables generation of Digital Elevation Models (DEMs) and detection of surface motion at the centimetre level using radar signals transmitted from a satellite or an aeroplane. Deformation observations can be performed due to the fact that surface motion, caused by natural and human activities, generates a local phase shift in the resultant interferogram. The magnitude of surface deformation can be estimated directly as a fraction of the wavelength of the transmitted signal. Moreover, differential InSAR (DInSAR) eliminates the phase signal caused by relief to yield a differential interferogram in which the signature of surface deformation can be seen. Although InSAR applications are well established, the improvement of the interferometry technique and the quality of its products is highly desirable to further enhance its capabilities. The application of InSAR encounters problems due to noise in the interferometric phase measurement, caused by a number of decorrelation factors. In addition, the interferogram contains biases owing to satellite orbit errors and atmospheric heterogeneity These factors dramatically reduce the stlectiveness of radar interferometry in many applications, and, in particular, compromise detection and analysis of small-scale spatial deformations. The research presented in this thesis aim to apply radar interferometry processing to detect small-scale surface deformations, improve the quality of the interferometry products, determine the minimum and maximum detectable deformation gradient and enhance the analysis of the interferometric phase image. The quality of DEM and displacement maps can be improved by various methods at different processing levels. One of the methods is filtering of the interferometric phase.However, while filtering reduces noise in the interferogram, it does not necessarily enhance or recover the signal. Furthermore, the impact of the filter can significantly change the structure of the interferogram. A new adaptive radar interferogram filter has been developed and is presented herein. The filter is based on a modification to the Goldstein radar interferogram filter making the filter parameter dependent on coherence so that incoherent areas are filtered more than coherent areas. This modification minimises the loss of signal while still reducing the level of noise. A methodology leading to the creation of a functional model for determining minimum and maximum detectable deformation gradient, in terms of the coherence value, has been developed. The sets of representative deformation models have been simulated and the associated phase from these models has been introduced to real SAR data acquired by ERS-1/2 satellites. A number of cases of surface motion with varying magnitudes and spatial extent have been simulated. In each case, the resultant surface deformation has been compared with the 'true' surface deformation as defined by the deformation model. Based on those observations, the functional model has been developed. Finally, the extended analysis of the interferometric phase image using a wavelet approach is presented. The ability of a continuous wavelet transform to reveal the content of the wrapped phase interferogram, such as (i) discontinuities, (ii) extent of the deformation signal, and (iii) the magnitude of the deformation signal is examined. The results presented represent a preliminary study revealing the wavelet method as a promising technique for interferometric phase image analysis.
|dc.title||Advanced satellite radar interferometry for small-scale surface deformation detection|
|curtin.department||Department of Spatial Sciences|