Integrating SBAS-InSAR and Artificial Intelligence for Land Subsidence Monitoring in Ca Mau, Vietnam

Abstract

In recent years, land subsidence has become a significant concern in the Ca Mau region, Vietnam, causing major environmental and socio-economic challenges. InSAR technology has been effectively applied in monitoring surface deformation over large areas. However, traditional InSAR processing methods still face limitations due to noise, phase unwrapping errors, and atmospheric disturbances, which affect the reliability of subsidence results. This study employs an integrated approach combining SBAS-InSAR with Artificial Intelligence (AI) to enhance the accuracy and efficiency of land subsidence monitoring. AI algorithms are applied to improve phase unwrapping, reduce noise, and correct atmospheric effects, thereby improving the quality of deformation signals obtained from InSAR data. The method is tested using Sentinel-1 imagery acquired over the Ca Mau city area during the period from January 2022 to December 2023. The results indicate that integrating AI significantly enhances the accuracy of land subsidence detection, achieving strong agreement with leveling subsidence data (RBS = 0.80; RMSE = 3 mm).