2D dike deformation monitoring using an optimized InSAR spatial unwrapping network

Interferometric synthetic aperture radar (InSAR) scatterers on dikes are typically distributed in narrow linear patterns, which makes reliable phase unwrapping challenging for multi-temporal InSAR (MT-InSAR). This limitation restricts accurate retrieval of two-dimensional (2D) deformation, particularly the decomposition of slope and normal components. To address this, a 2D dike deformation monitoring based on an optimized spatial unwrapping network (SUN) is proposed. A weighted shortest-path algorithm incorporating temporal coherence is applied to remove unreliable arcs in the initial SUN. A sliding-window strategy is further introduced to densify the network with high-quality connections. The optimized SUN is then directly integrated into the three-dimensional phase unwrapping framework. Additionally, a 2D deformation decomposition model tailored to dike geomorphology is developed. The method is applied to the Houtribdijk in the Netherlands using ascending and descending Sentinel-1A datasets from 2018 to 2022. Results show that the optimized SUN significantly reduces unwrapping errors, decreasing the standard deviation of residuals by up to 49%. Most dike segments remained stable during the monitoring period, while the reinforced segment exhibited notable subsidence, with maximum deformation rates of −5.07 cm/year in the slope direction and −4.35 cm/year in the normal direction. Cross-validation with ground measurement and EGMS products confirms the reliability of the results. Combined analysis reveals that sand compaction associated with the reinforcement project is the primary driver of surface deformation, while water level variations contribute to displacement and seasonal deformation in certain segments. The proposed approach enhances phase unwrapping reliability and enables reliable 2D deformation characterization for dike monitoring applications.