Automated Detection of Ground Deformation on Alabama Highways Using Remote Sensing

Details

• Creators:
  Khosravi, Ali ; Ghorbani, Zahra ; Zulfeqar, Najibullah ; Montgomery, Jack
• Corporate Creators:
  Auburn University. Highway Research Center
• Corporate Contributors:
  Alabama. Department of Transportation
• Subject/TRT Terms:
  Automation Decision Support Systems Deformation Detection And Identification Highways Highway Slopes Infrastructure InSAR Landslide Monitoring Landslides LiCSBAS Machine Learning Monitoring NSBAS PS-InSAR PyGMTSAR, Ground Deformation Radar Remote Sensing Slopes

  More +
• Publication/ Report Number:
  931-093
• Resource Type:
  Tech Report
• Geographical Coverage:
  United States ; Alabama
• Corporate Publisher:
  Auburn University. Highway Research Center
• Abstract:
  Ground deformation poses a significant risk to Alabama’s transportation infrastructure, particularly along highway corridors affected by landslides, embankment settlement, and hydrologically driven slope movement. This study develops and demonstrates an integrated satellite-based monitoring framework that combines multi-scale Interferometric Synthetic Aperture Radar (InSAR) processing, ground-based validation, environmental data integration, and machine learning–enhanced detection to support operational infrastructure monitoring across Alabama. Sentinel-1 C-band SAR data (2016–2025) were processed using complementary Small Baseline Subset (SBAS/NSBAS) and Persistent Scatterer (PS-InSAR) methodologies implemented through LiCSBAS and PyGMTSAR. The NSBAS workflow provided statewide deformation screening and regional situational awareness, while PS-InSAR enabled high-resolution monitoring of localized infrastructure assets and unstable slopes. Cross-comparison of methods demonstrated strong consistency in long-term deformation trends while highlighting the higher sensitivity of PS-InSAR for detecting localized acceleration and early-stage instability in vegetation-dominated environments. A comprehensive validation framework was implemented using independent geodetic and field datasets, including continuous GNSS observations, ALDOT inclinometer measurements, corner reflector deployments, and environmental datasets such as precipitation and soil moisture. Results show that satellite-derived time series successfully captured both catastrophic and slow-moving deformation processes, including pre-failure acceleration at the US-231 landslide site and seasonal hydrologically driven creep at SR-219. GNSS comparisons confirmed that NSBAS products reliably represent regional deformation behavior, while PS-InSAR demonstrated strong agreement with localized field instrumentation. Corner reflector installations established the foundation for future absolute validation and precision monitoring as additional satellite acquisitions become available. Building on validated deformation products, a web-based ALDOT GeoMonitor Portal was developed to integrate InSAR, GNSS, and environmental data within a unified decision-support environment. In addition, a deep learning–enhanced Early Detection and Alert System (EDAS) was developed using a decomposition-based modeling framework combining ARIMA and CNN–Attention–BiGRU architectures to detect anomalous displacement behavior and provide short-term predictive capability. Application at the Littleville corridor demonstrated the system’s ability to identify deformation anomalies and quantify hydrologically driven acceleration patterns relevant to operational monitoring.
• Format:
  PDF
• Collection(s):
  US Transportation Collection
• Main Document Checksum:
  urn:sha-512:06cdedf572c3f2ac2d72d0d2ec9e083bc833a5c37a7d7c0bf7af972bdbf4b8c248caaea64eb6ca9ada00592f734d6670cbd7bfcd9c60323ca967919a6ded5b91
• Download URL:
  https://rosap.ntl.bts.gov/view/dot/91201/dot_91201_DS1.pdf
• File Type:
  [PDF - 9.53 MB ]