Hakim, W. L., Fadhillah, M. F., Won, J. S., Park, Y. C., & Lee, C. W. (2025). Advanced time-series InSAR analysis to estimate surface deformation and utilization of hybrid deep learning for susceptibility mapping in the Jakarta metropolitan region. GIScience & Remote Sensing, 62(1). Click (a) ICOPS-MintPy based mean vertical deformation rate map in Jakarta: A high-resolution land subsidence distribution map derived from Sentinel-1 data spanning 2015 to 2023. (b) Susceptibility map generated by the LSTM-GWO hybrid model: Predicted land subsidence risk zones identified through optimized deep learning algorithms. (c) ROC-AUC performance metrics: A graph demonstrating that the LSTM-GWO model achieved the highest predictive accuracy with an AUC of 0.976. This study presents a comprehensive framework to precisely measure surface deformation and predict subsidence susceptibility in the Jakarta Metropolitan Region (JMR), an area facing severe land subsidence and flood risks due to excessive groundwater extraction. To address nonlinear deformation behaviors, we proposed an advanced time-series InSAR methodology that integrates the Improved Combined Scatterer Interferometry with Optimized Point Scatterers (ICOPS) method with the MintPy algorithm, utilizing multi-temporal Sentinel-1 SAR data from 2015 to 2023. Validated against GPS/GNSS observation point data, the proposed ICOPS-MintPy approach demonstrated superior accuracy, achieving lower RMSE values (e.g., 0.93 cm/year at the CTGR station) compared to the conventional ICOPS method.Furthermore, the high-precision subsidence inventory map derived from InSAR measurement points was used to generate susceptibility maps by integrating deep learning models (CNN and LSTM) with metaheuristic optimization algorithms (GWO and ICA). The evaluation results revealed that the LSTM-GWO hybrid model outperformed all other models, achieving an outstanding AUC of 0.976. Spatial correlation analyses indicated that over 90% of the subsidence occurred in settlement areas situated on alluvial fan and alluvial landforms, providing critical insights for sustainable urban hazard mitigation policies. Continue reading...
