Linking Crack Precursors to Landslide Runout Dynamics via Deep Learning Mapping and SPH Simulation
DOI:
https://doi.org/10.64862/ajeg.2025.2sp.122.276Keywords:
Deep learning, single- and multi-temporal UAV-based ground-crack mapping, Post-failure kinematics, Smoothed Particle Hydrodynamics (SPH), Spatiotemporal evolution, Kinematic run-out predictionAbstract
To investigate the landslide’s spatiotemporal evolution and failure mechanisms, a comprehensive field campaign based on multi-source data was conducted. Deep learning models (U-Net and ResU-Net) were applied to UAV orthophotos to automatically extract ground cracks, with ResU-Net achieving the highest accuracy (89.7%). Stability analysis using the limit equilibrium method and Monte Carlo simulations revealed a substantial reduction in the safety factor to 0.822 and a 72.82% increase in failure probability under extreme rainfall conditions. Smoothed Particle Hydrodynamics (SPH) simulations further predicted a 48 s run-out with a peak velocity of 28 m/s. This integrated framework highlights the potential of combining deep learning, probabilistic analysis, and particle-based modeling for quantitative landslide hazard assessment and early warning.
References
Abraham, M.T., Satyam, N., Reddy, S.K.P., Pradhan, B., 2021. Runout modeling and calibration of friction parameters of Kurichermala debris flow, India. Landslides 18(2), 737– 754. https://doi.org/10.1007/s10346-020-01540-1
Aminpour, M., Alaie, R., Khosravi, S., Kardani, N., Moridpour, S., Nazem, M., 2023. Slope stability machine learning predictions on spatially variable random fields with and without factor of safety calculations. Comput. Geotech. 153, 105094. https://doi.org/10.1016/j.compgeo.2022.105094
Daud H., Dou J., Khan N.G., et al., 2025. Tree- based machine learning and flow simulation for debris flow susceptibility, runout propagation, and dynamic in the higher Himalayas. Math. Geosci. 57, 1195-1233. https://doi.org/10.1007/s11004-025-10196-3
Jie D, Ali P. Y, Abdelaziz M, et al., 2020. Different sampling strategies for predicting landslide susceptibilities are deemed less consequential with deep learning, Science of The Total Environment, 720, 137320. https://doi.org/10.1016/j.scitotenv.2020.137320.
Lele Z, Jie D, Zilin X, et al. 2025. Spatiotemporal reconstruction and post-failure stability analysis of rainfall-induced landslides via multi-modal SAR-optical data fusion. Catena 259, 109399. https://doi.org/10.1016/j.catena.2025.109399.
Meng W, Wanqing S, et al., 2022. Phase-field modeling of cracking process in partially saturated porous media and application to rainfall-induced landslides, Engineering Geology, 310, 106884. https://doi.org/10.1016/j.enggeo.2022.106884.
Zilin X, Jie D, Huiming T et al., 2023. Vegetation-landslide nexus and topographic changes post the 2004 Mw 6.6 Chuetsu earthquake. Catena (Amst), Vol. 223, No. 106946. https://doi.org/10.1016/j.catena.2023.106946.
Downloads
Published
Data Availability Statement
Data will be made available on request.
Issue
Section
License
Copyright (c) 2025 Nepal Society of Engineering Geology (NSEG)
This work is licensed under a Creative Commons Attribution 4.0 International License.
