Change detection of slow-moving landslide with multi-source SBAS-InSAR and Light-U2Net
Citation
Zhang, Y., Meng, X., Dijkstra, T., Jordan, C., Chen, G., Zeng, R., Novellino, A. (2025). Change detection of slow-moving landslide with multi-source SBAS-InSAR and Light-U2Net. International Journal of Applied Earth Observation and Geoinformation, 136: 104342. Link to paper
Abstract
The paper proposes an automatic framework for Boundary-Changed Slow-moving Landslide (BCSML) detection by integrating multi-source Small Baseline Subset InSAR (SBAS-InSAR), Convolutional Neural Network (CNN), and change detection methodologies. A novel and effective Light-U2Net was constructed with decreased complexity to identify Significant Deformation Zone (SDZ) and locate SML candidate. The proposed Light-U2Net model achieves high Precision (80.1%), Recall (80.2%), and F1-scores (80.1%) in Zayu County. The model's complexity has reduced by 42.4% without compromising identification accuracy compared to the original model. In the NLPF area, a total of 273 BCSMLs were detected, with 176 identified as expanding and 97 as shrinking, with BCSML identification accuracy reaching 90.47%.