Development of an Imagewise Stacking Method to Mitigate Atmospheric Effect With an Application to Lianyan Railway

Much attention has been paid to the subsidence monitoring of large-scale man-made linear features, which are usually characterized by long-distance, large-scale, and nonuniform distribution. The multitemporal interferometric synthetic aperture radar (InSAR) technique can provide ground deformation maps for a large area with high spatial resolution and millimeter precision at low costs. The technique overcomes the limitations of ground-based survey instruments. However, the variable atmospheric turbulent component in flat areas has impacts on InSAR detection capability, especially over low-amplitude quasi-stable deformation areas. We propose an imagewise stacking method, which can improve the InSAR time-series analysis by reducing residual noise while retaining the ground deformation signal. We integrate persistent scatterer and distributed scatterer in our processing strategy and conduct a tentative test on the roadbed subsidence monitoring of the Lianyan high-speed railway using 47 medium-resolution Sentinel-1 images. Comparing the InSAR displacement time series and the results with a continuous BeiDou navigation satellite system network, the standard deviation of 3.3 mm and the mean absolute error of 2.2 mm are obtained. The proposed method reduced the above two error measures by 66.3% and 65.6%, respectively, compared with the spatiotemporal filtering. Moreover, the analysis of the spatiotemporal characteristics of the railway deformations also tells that the railway under study is stable.