Land subsidence poses an increasing hazard in urban areas such as the Bandung Metropolitan Area (BMA), Indonesia. This study monitors land subsidence from 2018 to 2024 and develops predictive susceptibility maps. Deformation was analyzed using Improved Combined Scatterers Interferometry with Optimized Point Scatterers (ICOPS) applied to Sentinel-1 SAR datasets. Subsidence susceptibility was modelled using convolutional neural networks (CNN) and CNN models optimized with the Gray Wolf Optimizer (CNN-GWO) and Imperialist Competitive Algorithm (CNN-ICA), trained with topographic, land cover, population, geological, and hydrological factors. Results show that major subsidence exceeding 10 cm/year concentrated in urbanizing zones underlain by compressible deposits and exposed to construction loads. Comparative validation between ICOPS results and GNSS was conducted, demonstrating the reliability of ICOPS monitoring. A land subsidence susceptibility map with CNN-GWO achieving superior accuracy (AUC = 0.919) compared to CNN-ICA (AUC = 0.904) and standalone CNN (AUC = 0.874). Lithology, precipitation, groundwater level change, and elevation emerged as the most influential subsidence drivers. This integrated satellite monitoring and deep learning framework provides a transferable approach for subsidence risk assessment in vulnerable urban regions.
Achmad et al. (Sat,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: