Subsidence characterization of karst sinkholes using satellite remote sensing: A Missouri case study

Greene County in Missouri has experienced a significant increase in sinkhole occurrences over recent decades due to its karst geology. This study focuses on investigating ground subsidence related to karst sinkholes using satellite-based remote sensing techniques and aims to develop a sinkhole susceptibility map utilizing Geographic Information System (GIS) methodologies. Interferometric Synthetic Aperture Radar (InSAR) data from Sentinel-1 satellites, covering the period from 2018 to 2020, were employed to detect and analyze ground deformation patterns. The InSAR analysis revealed an annual subsidence rate of up to 30 mm along the satellite's line-of-sight, indicating active ground movements in the region. To predict areas susceptible to future sinkhole development, a sinkhole inventory dataset was compiled from the Missouri Department of Natural Resources (MoDNR), and an Artificial Neural Network (ANN) machine learning model was applied. Topographic conditioning factors were derived from high-resolution Light Detection and Ranging (LiDAR) data to enhance the predictive modeling. The results demonstrated a strong correlation between areas of significant deformation detected by InSAR and regions identified as highly susceptible to sinkholes in the susceptibility map. Furthermore, newly identified sinkholes coincided with zones of high subsidence, validating the predictive capacity of the ANN model. This study underscores the effectiveness of integrating satellite remote sensing with machine learning techniques to detect subtle ground deformation and to map zones at risk of future sinkhole formation. The proposed approach offers valuable insights for sustainable urban development, land-use planning, and hazard mitigation strategies in karst regions like Greene County.