Three-dimensional Displacement Estimation and Accuracy Validation on Highway Slopes Based on Multi-view Ground-based SAR

Objectives: Highway slope instability poses serious threats to transportation safety and infrastructure reliability. Accurate three-dimensional (3D) deformation monitoring is essential for slope stability assessment and disaster early warning. A practical method for estimating slope-parallel horizontal displacement, transverse horizontal displacement perpendicular to the roadway, and vertical displacement is proposed in this study, using multi-view ground-based synthetic aperture radar (GBSAR). Methods: A single GBSAR system was sequentially installed at three stations respectively with different azimuth viewing angles to acquire interferometric measurements in a time-sharing mode. A projection model was established to transform multi-view line-of-sight (LOS) displacements into three independent deformation components. To address the ill-posed nature of the inversion problem, Singular Value Decomposition (SVD) combined with Tikhonov-regularized least squares method was employed to estimate the 3D displacement components, which enhanced the numerical stability and suppressed the noise amplification.A in-situ experiment on a highway slope in Changsha, China was conducted, with 4 artificial corner reflectors (CRs) installed on the slope to validate the geometry projection function and the displacement accuracy. Results: The effectiveness of the proposed method has been confirmed. The root mean square errors (RMSEs) relative to CR measurements were estimated as 0.4mm for slopeparallel displacement, 0.4mm for vertical displacement, and 1.8mm for transverse displacement, respectively. Compared with conventional least squares estimation methods, the SVD combined with Tikhonov-regularized least squares method significantly reduced the error propagation caused by rank deficiency and improved reliability under complex slope conditions. Conclusions: The proposed multiview GBSAR 3D displacement monitoring method provides a cost-effective and accurate solution for 3D slope deformation monitoring using only a single radar system. It can offer high flexibility, low hardware costs compared to multi-device systems, and ensure robust performance in complex environments. This approach has strong potential for geohazard early warning and infrastructure safety management.