Automatic Reference Points Selection for Large-Scale InSAR Data Processing via Analytic Hierarchy Process

Abstract

With the increasing availability and volume of SAR data, interferometric synthetic aperture radar (InSAR) processing is gradually evolving toward automation and large-scale applications, where reference point (RP) selection plays a critical role in both processing efficiency and deformation accuracy. To address the limitations of conventional RP selection methods, such as deformation shifts, error accumulation, and low automation, this study proposes an automatic multi-RPs selection method based on the analytic hierarchy process (AHP). The method integrates multisource constraints for candidate screening, and then constructs a multicriteria weighting framework using AHP to determine the optimal RP within each connected component. Taking Zhuhai City and its major island regions (Guangdong, China) as the study area, deformation results derived from different selection methods are compared based on the SBAS-InSAR technique. The results indicate that: first, although different methods yield consistent spatial deformation patterns, the proposed method achieves superior reference consistency with a mean deformation rate of −0.5 mm/yr (near zero-deformation); second, in isolated island regions, the proposed method accurately retrieves true surface deformation, whereas other methods exhibit evident phase unwrapping errors;third, comparative experiments show that the proposed method achieves higher reliability and robustness, with deformation rate distributions showing smaller mean (0.5 mm/yr) and standard deviation (1.4 mm/yr) compared to other methods; finally, locally, the proposed method yields more stable RPs (mean: 1.6 mm/yr) than conventional methods (−5 to +10 mm/yr). Overall, the proposed method demonstrates higher reliability and generalization across scales, mitigating deformation distortion in spatially discontinuous areas, while enhancing automation and efficiency. This study provides a practical solution for large-scale InSAR processing in complex environments.