Monitoring land subsidence using Sentinel-1A, persistent scatterer InSAR, and machine learning techniques

Abstract

Land subsidence is a significant environmental issue in coastal regions, leading to infrastructure damage and increased vulnerability to flooding. This study focuses on monitoring land subsidence in the coastal region of Maceió using Sentinel-1A Synthetic Aperture Radar (SAR) data in conjunction with Persistent Scatterer Interferometry (PSI) and machine learning (ML) techniques. Sentinel-1A data, including both ascending and descending track datasets, were utilized to capture ground displacement over time. The analysis incorporated a range of environmental variables, including slope, elevation, aspect, curvature, land use/land cover (LULC), geology, Hillshade, and flow direction, which were used as training sets for the ML models. The Persistent Scatterer in SAR (PSI) technique was employed to extract precise displacement data from the SAR imagery, revealing land subsidence trends over the study period. ML models were applied to analyse the correlation between the subsidence patterns and environmental factors, enhancing the accuracy and reliability of the results. The ML techniques utilized included regression models and classification algorithms to predict and interpret subsidence rates and spatial distribution. The results show that land subsidence in Maceió ranged from 52 mm/year to −60 mm/year, with significant spatial variability. The findings highlight areas of rapid subsidence that may be linked to geological and anthropogenic factors, such as urbanization and groundwater extraction. Incorporating machine learning methods into this analysis improved the spatial resolution of the subsidence estimates, offering a more comprehensive understanding of the underlying causes and trends. This study demonstrates the potential of integrating remote sensing, InSAR, and ML techniques for monitoring and understanding land subsidence in coastal regions. The results can inform mitigation strategies and urban planning in areas susceptible to subsidence and related hazards.