In urban areas with high population density and significant economic assets, both above‐ground and underground developments exert considerable influence on the safety of existing multistory and high‐rise buildings constructed in earlier periods, and a number of these aging buildings are facing increasingly severe safety risks with the extension of their service life. Traditional safety monitoring methods are challenged in large‐scale monitoring due to lower monitoring efficiency, making the study of large‐scale and wide‐area safety screening methods of great significance. Time series interferometric synthetic aperture radar (InSAR) technology, by extracting permanent scatterer (PS) points, enables long‐term monitoring of extensive urban buildings and infrastructure. However, the general deformation parameters of PS points obtained through InSAR calculations often fail to meet the requirements for building risk analysis. Therefore, a structural differential settlement index (SDSI) calculation and monitoring method for urban buildings based on InSAR was proposed in this study. The improved Huber loss function–based approach in the SDSI model significantly enhanced fitting stability and eliminated the influence of outliers on the least squares method results. A total of 41 COSMO‐SkyMed images from 2020 to 2022 were acquired in this study. The deformation inversion calculation method was used to calculate the SDSI, which was compared with ground measurement data from the same period for validation. The results show that the SDSI is highly correlated with the ground measurement tilt data, with an average correlation coefficient of over 0.70. Meanwhile, a case study was conducted on several buildings located in an area of Shenzhen affected by underground shield tunneling construction. The classification accuracy of valid samples within the experimental area reached 84%, which can provide an effective idea and method for InSAR urban building monitoring and groundwork for follow‐up safety assessments of building structures.
Li et al. (Thu,) studied this question.