A Study on the Prediction Methods of Slope Deformation Using SAR Data

Slope deformation is one of the most critical precursors to natural hazards such as landslides, embankment failures, and ground subsidence. Reliable early detection and prediction of slope deformations are essential to mitigate disaster risks and support resilient land management. This study reviews recent advances in slope deformation prediction methods using Synthetic Aperture Radar (SAR) and proposes a research method to enhance prediction accuracy and applicability. The review focuses on four main approaches: (i) InSAR time-series analysis for temporal deformation tracking, (ii) conversion of line-of-sight (LOS) displacements into three-dimensional ground motion components, (iii) nonlinear deformation forecasting using machine learning techniques, and (iv) data fusion between high-resolution satellite SAR and ground-based SAR (GB-InSAR) observations.

The analysis of SAR-based studies published over the past five years demonstrates that high-frequency and high-resolution SAR data, when combined with time-series analysis, can quantitatively capture progressive slope deformation and acceleration trends at millimeter-level precision. Integrated models that incorporate climatic, geological, and topographic factors achieved strong predictive performance, with coefficients of determination (R²) exceeding 0.9. Machine learning–based approaches, particularly those employing recurrent neural networks and ensemble algorithms, effectively represented nonlinear and seasonal deformation dynamics. However, prediction accuracy remains constrained by dense vegetation, limited satellite revisit intervals, and the directional sensitivity of LOS measurements, which can introduce uncertainty in estimating the true magnitude and direction of deformation.

This study investigated the strengths, limitations, and practical considerations of current SAR-based deformation prediction methods. The research findings confirm that multi-sensor integration, combining SAR data with meteorological, hydrological, and geotechnical information, can significantly improve the reliability and generalizability of slope deformation forecasts. Moreover, AI-based frameworks offers promising opportunities for interpretable and transferable models applicable to different slope environments. Based on an analysis of current research trends, this study provides a comprehensive overview of the latest SAR-based slope deformation prediction technologies and proposes a research method for developing a real-time monitoring and prediction system for slope deformation under the influences of climate change and anthropogenic factors 

 

ACKNOWLEDGEMENTS

Research for this paper was carried out under the KICT Research Program (project no. 20250285-001, Development of infrastructure disaster prevention technology based on satellites SAR.) funded by the Ministry of Science and ICT