A novel MT-InSAR method for robust estimation of permafrost deformation by integrating u test and multitier network strategy: a case study of the Beiluhe region of the Qinghai–Tibet Plateau

• A novel MT-InSAR method addresses low-coherence challenges in permafrost regions. • U test that considers temporal backscatter variability improves SHP identification. • Multitier network effectively controls error propagation and enhances reliability. • The method improves the completeness and reliability of the deformation field. Distributed scatterer interferometry (DSI) is an effective technique for improving InSAR-derived deformation quality in low-coherence permafrost regions. Reliable identification of statistically homogeneous pixels (SHPs) and deformation inversion through a two-tier network are key steps of DSI. However, existing SHP selection methods based on parametric hypothesis testing generally assume temporal stationarity of surface backscattering, which is often violated in permafrost regions due to freeze‒thaw processes in the active layer. In addition, the conventional two-tier network strategy tends to produce unstable inversion networks in decorrelated permafrost areas, thereby degrading deformation reliability. To address these limitations, this study proposes a robust coherent scatterer InSAR (RCSI) method. First, considering temporal variations in surface backscattering, a U test is developed to identify SHPs with similar temporal backscattering behavior. Second, a multitier network strategy is proposed, in which persistent scatterers (PS) and distributed scatterers (DS) candidates are evenly partitioned into multiple groups according to posterior coherence, and deformation is progressively estimated using robust M−estimators. The effectiveness of RCSI is quantitatively validated using both simulated data and Sentinel-1 observations over the Beiluhe region on the Qinghai–Tibet Plateau. The results show that the U test can effectively identify SHPs under non-stationary backscattering conditions. Compared with the two-tier network strategy, the multitier network strategy achieves a 19.9% improvement in deformation accuracy, as validated against in situ leveling measurements, and yields a 42.6% increase in the number of measurement points (MPs). Our study demonstrates that RCSI can significantly enhance the quality of InSAR-derived deformation in permafrost regions.