Influence of DEM Resolution on Geomorphometric Representation, Explainable Artificial Intelligence and Temporal Transferability of Machine Learning Models for Landslide Susceptibility Mapping in the Nepal Himalaya
DOI:
https://doi.org/10.64862/Keywords:
Landslide susceptibility mapping, DEM resolution, Random forest, XGBoost, SHAP, Temporal transferability, Nepal Himalaya, Geomorphometric analysisAbstract
Landslide susceptibility mapping (LSM) is essential for risk mitigation in the Nepal Himalaya, yet the effects of digital elevation model (DEM) resolution on geomorphometric representation, model discrimination, explainability, and temporal transferability remain insufficiently resolved. This study evaluates 5, 12.5, 30, and 90 m DEMs using Random Forest (RF) and Extreme Gradient boosting (XGBoost) with spatially grouped 10-fold cross-validation. The modelling dataset comprised 54,952 samples from 486 landslide polygons (1955–2010), with temporal validation using 135 landslides (2011–2025). Results show resolution sensitivity in topographic ruggedness index (TRI), stream power index (SPI), slope, relief, and elevation, whereas plan and profile curvature were robust. Mean spatial cross-validated AUROC remained stable (0.779–0.792), with no significant resolution effect (Friedman χ² = 2.76, p = 0.430), but threshold-dependent metrics degraded at 90 m. SHapley Additive exPlanations (SHAP) showed stable feature rankings dominated by rainfall, elevation, lithology, and distance to roads, with rainfall–lithology as the strongest interaction. Multi-criteria scoring placed 12.5–30 m within the preferred range. Temporal validation preserved susceptibility ranking (AUROC 0.809–0.846) but revealed under-confidence, with Youden-optimal thresholds of 0.13–0.17. The results indicate that DEM coarsening changes terrain representation more strongly than environmental relationships learned by ensemble models, while probability calibration requires temporal updating.
References
Chang, K.-T., Merghadi, A., Yunus, A. P., Pham, B. T., and Dou, J. (2019), “Evaluating scale effects of topographic variables in landslide susceptibility models using GIS-based machine learning techniques,” Scientific Reports, 9, 12296. https://doi.org/10.1038/s41598-019-48773-2
Friedman, M. (1937). The use of ranks to avoid the assumption of normality implicit in the analysis of variance. Journal of the American Statistical Association, 32(200), 675–701. https://doi.org/10.1080/01621459.1937.10503522
Gaidzik, K., and Ramírez-Herrera, M. T. (2021). The importance of input data on landslide susceptibility mapping. Scientific Reports, 11, 19334. https://doi.org/10.1038/s41598-021-98830-y
Lundberg, S. M., and Lee, S.-I. (2017). A unified approach to interpreting model predictions. Advances in Neural Information Processing Systems, 30, 4765–4774. https://doi.org/10.48550/arXiv.1705.07874
Merghadi, A., Yunus, A. P., Dou, J., Whiteley, J., ThaiPham, B., Bui, D. T., Avtar, R., and Abderrahmane, B. (2020). Machine learning methods for landslide susceptibility studies: A comparative overview of algorithm performance. Earth-Science Reviews, 207, 103225. 10.1016/j.earscirev.2020.103225
Nocentini, N., Rosi, A., Piciullo, L., Liu, Z., Segoni, S., and Fanti, R. (2024). Regional-scale spatiotemporal landslide probability assessment through machine learning and potential applications for operational warning systems. Landslides, 21, 2369–2387. 10.1007/s10346-024-02287-9
Roberts, D. R., Bahn, V., Ciuti, S., Boyce, M. S., Elith, J., Guillera-Arroita, G., Hauenstein, S., Lahoz-Monfort, J. J., Schroeder, B., Thuiller, W., Warton, D. I., Wintle, B. A., and Dormann, C. F. (2017). Cross-validation strategies for data with temporal, spatial, hierarchical, or phylogenetic structure. Ecography, 40(8), 913–929. 10.1111/ecog.02881
Usta, Z., Akıncı, H., and Akın, A. T. (2024). Comparison of tree-based ensemble learning algorithms for landslide susceptibility mapping in Murgul (Artvin), Turkey. Earth Science Informatics, 17, 1459–1481. 10.1007/s12145-024-01259-w
Wubalem, A. (2022). The impact of DEM resolution on landslide susceptibility modeling. Arabian Journal of Geosciences, 15, 967. https://doi.org/10.1007/s12517-022-10241-z
Yan, G., Tang, G., Li, S., Lu, D., Xiong, L., and Liang, S. (2023). Uncertainty in regional scale assessment of landslide susceptibility using various resolutions. Natural Hazards, 117, 399–423. https://doi.org/10.1007/s11069-023-05865-7
Youden, W. J. (1950). Index for rating diagnostic tests. Cancer, 3(1), 32–35. https://doi.org/10.1002/1097-0142(1950)3:1
Zhao, X., Chen, W., Tsangaratos, P., and Ilia, I. (2024). Evaluating landslide susceptibility: The impact of resolution and hybrid integration approaches. Geomatics, Natural Hazards and Risk, 15(1), 2409198.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Nepal Society of Engineering Geology (NSEG)
This work is licensed under a Creative Commons Attribution 4.0 International License.
