InSAR-Derived Spatiotemporal Deformation Monitoring and Unsupervised Clustering-Based Evolution Pattern Analysis of the Shanghai Seawall

Objective: The Shanghai Seawall is located at the estuary of the lower Yangtze River with a humid climate and uneven precipitation distribution, so the operational safety of hydraulic conservancy projects is of utmost importance. Methods: This study focuses on high-resolution deformation monitoring and long-term evolutionary pattern analysis of the Shanghai seawall. Using a total of 120 high-resolution long-term TerraSAR-X acquisitions from 2013 to 2024, we systematically analyze the deformation evolution of the seawall by integrating Interferometric Synthetic Aperture Radar (InSAR) technique with unsupervised learning methods. To address the instability of clustering results caused by noise contamination i n InSAR deformation time series and the subtle differences among deformation patterns, a hybrid analytical framework is developed, consisting of Principal Component Analysis (PCA) for dimensionality reduction, Dynamic Time Warping (DTW)–based optimization, and K-means clustering. This framework improves traditional statistically distance-based clustering by incorporating temporal shape similarity constraints, thereby enhancing the separability and stability of deformation patterns in the temporal dimension. Distinct deformation evolution pattern clusters are identified and further interpreted using historical optical imagery and ancillary data. Results: The results show that significant subsidence areas along the seawall are mainly distributed in the reclaimed land area, agricultural land area, and estuarine sections, with a maximum subsidence rate of 13.76 mm/a. Among them, the seawalls near the Pudong Airport have been continuously subsiding since 2015. The seawalls located at the estuaries of the Dazhi River and the tributaries of the Huangpu River, and the outer seawalls near the confluence of Tongyuan East Road and Guanhai Boulevard, have subsided persistently throughout the observation period. In contract, the inner seawalls near the confluence of Tongyuan East Road and Guanhai Avenue showed a characteristic of rapid subsidence initially, followed by slow subsidence, and eventually stabilizing. Conclusions: The different subsidence pattern is mainly influenced by the soil consolidation and compression in the reclaimed land area, the over-exploitation of groundwater in the agricultural area, and the combined effects of sedimentation and the tidal forces in the estuary area. Based on the cluster-based analysis of subsidence, the "point-threshold" interpretation has been upgraded to a "pattern-interpretation". This enables the robust extraction of typical deformation evolution patterns, providing a basis and technical support for the segmented management, investigation and reinforcement of seawalls.