Fadhillah, M. F., Hakim, W. L., Park, S. J., & Lee, C. W. (2025). Integrating SAR and Optical Imagery Analysis for Liquefaction Phenomenon Identification of Post-Pohang Earthquake 2017, South Korea, Utilizing a Hybrid Deep-Learning Approach. IEEE Transactions on Geoscience and Remote Sensing. Click (a) Liquefaction susceptibility map generated using the hybrid DCNN-PSO model in the Pohang area. (b) Sentinel-1 DInSAR surface deformation analysis. (c) Performance metrics showing DCNN-PSO achieved the highest ROC-AUC of 0.86. Following the 2017 Pohang earthquake (Mw 5.4), this study introduces a comprehensive framework for identifying liquefaction by integrating multi-sensor remote sensing data and advanced machine-learning techniques. The research utilizes Sentinel-1 C-band SAR data to capture ground deformation through DInSAR and persistent/distributed scatterer analysis (ICOPS), alongside soil moisture evaluations using optical satellite imagery.The proposed methodology employs a hybrid deep-learning approach, combining Dense Convolutional Neural Networks (DCNN) with swarm-based metaheuristic optimization (PSO and GWO) to enhance detection reliability. The optimized DCNN-PSO model demonstrated superior performance with an AUC value of 0.86, accurately mapping the spatial distribution of liquefaction in quaternary alluvium areas. This integrated approach provides a robust tool for rapid post-disaster assessment and demonstrates the profound effectiveness of AI in evaluating complex geospatial phenomena. Continue reading...
