Monitoring surface deformation of the Qinghai–Tibet permafrost region using an improved MT-InSAR method based on spatial–temporal constraints

• Main Authors: Jiachen Li, Qijie Wang, Ya Zhang, Kai Sun, Guanyou Gao
• Format: Article in Journal/Newspaper
• Language: English
• Published: Taylor & Francis Group 2024
• Subjects: Permafrost deformation; multi-temporal InSAR (MT-InSAR); segmented permafrost deformation model; spatial similarity constraint; Qinghai–Tibet Plateau; Mathematical geography. Cartography; GA1-1776; permafrost
• Online Access: https://doi.org/10.1080/17538947.2024.2406383; https://doaj.org/article/dd16592830f342a1a7fe14bd946add06

Monitoring ongoing permafrost degradation on the Qinghai–Tibet Plateau is challenging because of its underground characteristics. Interferometric synthetic aperture radar (InSAR) technology, which can detect large-scale and high-precision surface deformation, has become an effective tool for indirect permafrost monitoring. Traditional InSAR methods usually solve permafrost deformation independently and on a point-by-point basis with high coherence, while existing permafrost deformation models ignore the interannual variation in deformation. This paper proposes an improved multi-temporal InSAR (MT-InSAR) method based on spatial–temporal constraints for monitoring permafrost deformation. The method establishes a segmented permafrost deformation model considering the interannual variation in deformation over time. The spatial similarity constraint is subsequently introduced to establish the relationship between deformations at neighboring points, improving the results of permafrost monitoring. Both simulated and real experiments confirmed the ability of the proposed method to capture interannual variations in permafrost deformation with higher accuracy and point density. A real experiment with Sentinel-1 data revealed significant interannual variations in surface deformation within the Wudaoliang-Tuotuohe permafrost region from 2018–2021, characterized by decreased linear deformation and slight seasonal deformation changes. Compared with the traditional methods and in-situ data, the deformation accuracy and point density can be improved by approximately 15.6% and 19.4%, respectively.