- Department of Civil and Environmental Engineering, Gangneung-Wonju National University, Gangneung, Korea, Republic of (rsgis.manik@gmail.com)
Rapid urban and industrial development in reclaimed coastal cities of the Korean Peninsula has imposed persistent anthropogenic loading and vibration on thick marine sediment, leading to frequent geotechnical hazards and exacerbating vulnerability to compound coastal hazards. Although preconstruction ground improvement techniques were implemented adequately, these thick sediment layers continue to undergo long-term consolidation and secondary compression, resulting in progressive subsidence that is often overlooked by MTInSAR-based ground subsidence monitoring approaches. Therefore, to characterize land subsidence dynamics and associated cascading hazards in reclaimed cities, we develop an ensemble kinematic and physical modeling framework using time-series SAR interferometry. For experimental purposes, we selected three major reclaimed coastal cities, i.e., Incheon, Mokpo, and Busan in South Korea. Initially, the multi-temporal Sentinel-1 SAR (2017-2023) data were processed using the Persistent Scatterer Interferometry (PSInSAR) technique to derive vertical displacement (VD) time series at persistent scatterer (PS) locations with temporal coherence >0.7. Thereafter, the VD time series of each PS point was smoothed using a Savitzky-Golay local polynomial regression to reduce noise while preserving long-term displacement signals. Consequently, subsidence kinematics were characterized by identifying the temporal regime using the Pruned Exact Linear Time (PELT) algorithm with an L2 loss function, and segment-wise first-order linear regression (R²>0.9) was applied to quantify phase-dependent displacement rates. The results exhibit that the VD rates in the reclaimed regions of Mokpo, Busan, and Incheon vary from -9.78 to 3.59 mm/yr, -41.19 to 1.85 mm/yr, and -9.89 to 1.79 mm/yr, with mean rates of -0.64, -4.15, and -0.94 mm/yr, respectively. We observed multiple VD phases characterized by a shift from quasi-linear settlement to episodic acceleration at each PS in all three cities, indicating that reclaimed sediments are still undergoing consolidation and stress redistribution within strata. Furthermore, the VD time series of each PS was modeled using hyperbolic settlement formulations to characterize the nonlinear consolidation behavior of reclaimed sediments. The strong agreement between hyperbolic model predictions and PSInSAR-derived VD indicates that land subsidence in reclaimed areas within these cities is predominantly consolidation-controlled. The modeled subsidence characteristics were further validated through analysis of in-situ borehole geotechnical data using the Casagrande plasticity chart. Moreover, the velocity decay ratios derived from the hyperbolic settlement model exhibit relatively high in Busan (0.733) and Incheon (0.603), indicating sustained, long-term settlement associated with secondary compression and transitional consolidation stages. On the other hand, the Mokpo reclaimed region exhibits a substantially lower decay ratio (0.341), indicating a rapid attenuation of subsidence velocity and near completion of primary consolidation, which is also consistent with its decade-old land reclamation history. Notably, it was observed that subsidence-related geohazards (i.e., sinkhole occurrences) have been reported more frequently in the reclaimed areas of Incheon and Busan than in Mokpo, providing independent evidence that further supports the modeled subsidence mechanism. The proposed ensemble framework exhibits that integrating kinematic segmentation with physical modeling of PSInSAR-derived VD time histories facilitates a transition from passive monitoring to predictive urban subsidence hazard assessment, which is crucial for long-term infrastructure planning in reclaimed coastal megacities under global climate change scenarios.
How to cite: Das Adhikari, M., Song, M.-S., Kim, S.-W., and Yum, S.-G.: Assessing Progressive Subsidence Hazards in Reclaimed Coastal Cities Using Ensemble Kinematic and Physical Modeling of PSInSAR Displacement Time Series, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4023, https://doi.org/10.5194/egusphere-egu26-4023, 2026.