Unsupervised Clustering of InSAR Time-Series Deformation in Mandalay Region from 2022 to 2025 Using Dynamic Time Warping and Longest Common Subsequence

Urban land subsidence poses a significant threat in rapidly urbanizing regions, threatening infrastructure integrity and sustainable development. This study focuses on Mandalay, Myanmar, and presents a novel clustering framework—Dynamic Time Warping and Trend-based Longest Common Subsequence with Agglomerative Hierarchical Clustering (DTLCS-AHC)—to classify spatiotemporal deformation patterns from Small Baseline Subset (SBAS) Interferometric Synthetic Aperture Radar (InSAR) time series derived from Sentinel-1A imagery covering January 2022 to March 2025. The method identifies four characteristic deformation regimes: stable uplift, stable subsidence, primary subsidence, and secondary subsidence. Time–frequency analysis employing Empirical Mode Decomposition (EMD) and Discrete Fourier Transform (DFT) reveals seasonal oscillations in stable areas. Notably, a transition from subsidence to uplift was detected in specific areas approximately seven months prior to the Mw 7.7 earthquake, but causal relationships require further validation. This study further establishes correlations between subsidence and both urban expansion and rainfall patterns. A physically informed conceptual model is developed through multi-source data integration, and cross-city validation in Yangon confirms the robustness and generalizability of the approach. This research provides a scalable technical framework for deformation monitoring and risk assessment in tropical, data-scarce urban environments.