Determining land subsidence in western Taiwan by using a new approach of weighting PS-InSAR with redundant observations

Since the emerging of the technique of persistent scatterer interferometric synthetic aperture radar (PS-InSAR), diverse algorithms and methods are developed for finding stable scatterers. Also, different models and approaches are proposed for reducing phase errors caused by diverse influencing factors such as atmospheric sphere, topographic characteristics and noise. Then, the terrain surface deformation velocity at each persistent scatterer (PS) is determined. Although PS-InSAR is already widely applied for monitoring low-speed terrain surface deformation velocity component, most researches study mainly on the estimation of linear deformation velocity. Therefore, this paper presents a new approach for improving the data processing of PS-InSAR. Moreover, some ground check data are used to evaluate the quality of PS-InSAR on monitoring terrain surface deformation. In order to increase the robustness of PS-InSAR, the technique of redundant observation in the field of surveying and geomatics is utilized on the one hand. For each computation of PS-InSAR, different sets of SAR images in the same time period are adopted so that the ability of blunder detection is increased. On the other hand, the decorrelation effect caused by diverse factors is taken into account so that different weights are given to all sets of SAR images. Weighting averages are then computed to determine the most probable deformation velocity vectors on all scatterers. Higher outer reliability is expected. Finally, the results are compared with the ground check data determined by high precision levelling. Their quality is thus estimated.