Measuring Land Surface Deformation over Soft Clay Area Based on an FIPR SAR Interferometry Algorithm—A Case Study of Beijing Capital International Airport (China)

Abstract

Long-term settlement monitoring of infrastructure built in soft clay areas is of great importance. When using InSAR technology for soft clay settlement monitoring, deformation modeling is a key process. In most InSAR deformation modeling, each component of the total deformation is expressed directly with a fixed functional model in phase functions and assumed to occupy an equal weight. This causes equal weight assumption uncertainty and ignores the actual certain contribution of each phase component related to certain deformational factors. Moreover, the commonly used mathematical empirical models in traditional InSAR are not suitable for describing the nonlinear characteristics of the temporal settlement evolution for soft clay. To address these limitations, we propose an SAR interferometry algorithm, namely, FIPR (FastICA Poisson-curve reciprocal accumulation method), which separates the original InSAR signal based on FastICA to extract each deformation component, and then the models can each extract deformation components and estimate the unknown parameters based on a reciprocal accumulation method. Each independent component and the obtained deformation parameters are used to generate the final deformation time series. Both simulated and real data experiments were designed. The simulated experimental results indicated that the sICA (spatial independent component analysis) separated results were much closer to the original signals than those of the tICA (temporal independent component analysis), with their RMSE lower than 2 mm, and the sICA is thus more highly recommended. Beijing Capital International Airport in China was selected as the study area in the real data experiment. Using 24 high-resolution TerraSAR-X radar satellite images from January 2012 to February 2015, the time-series deformation was obtained, with the maximum cumulative subsidence of 126 mm. The modeling accuracy for the proposed model was estimated as ±2.6 mm, with an improvement of 36.6% compared to the EWA-LM (linear model with equal weight accumulation) algorithm and 16.1% compared to the EWA-PC (Poisson curve with equal weight accumulation) algorithm. The RMSE with external leveling measurements was estimated as ±1.0 mm, with 69.7% improvement compared to EWA-LM and 50% to EWA-PC. This indicated that FIPR can reduce the uncertainty of artificial assumptions in deformation modeling and improve the accuracy of deformation analysis for highways in soft clay areas, providing a reference for road maintenance and management.