Rainfall-Induced landslide susceptibility prediction considering spatial heterogeneity

Abstract

Existing landslide susceptibility prediction methods often fail to fully account for the spatial heterogeneity of environmental factors such as topography, soil, and vegetation, nor do they accurately reflect the impact of extreme rainfall on landslide susceptibility. To overcome these limitations, this study proposes two innovative methods. First, to address the issue of spatial heterogeneity, a Deep Embedding Clustering (DEC) approach is introduced. DEC utilizes an autoencoder to map environmental factors to a lower-dimensional space, capturing nonlinear relationships between variables and performing clustering in this space. Unlike traditional methods, DEC does not rely on simple distance measures; instead, it jointly optimizes clustering centers and feature representations, enabling more precise regional delineation, which significantly enhances prediction accuracy and adaptability to varying environments. Second, to address the static nature of rainfall thresholds, a mixed distribution modeling strategy is proposed for both non-extreme and extreme rainfall. In this strategy, non-extreme rainfall is modeled using the Gamma distribution to describe cumulative effects, while extreme rainfall is modeled using the Generalized Pareto Distribution (GPD) to model extreme values, with thresholds dynamically determined using the Pickands theorem. Additionally, a Bayesian online parameter updating mechanism is implemented to dynamically adjust distribution parameters, recalibrating the model when real-time rainfall data deviates from historical distributions, significantly reducing response time and improving the model's adaptability to changing rainfall patterns. By combining Deep Embedding Clustering (DEC) and the mixed distribution rainfall threshold model, this study achieves more precise spatial zoning and dynamic rainfall responses, greatly improving prediction accuracy and timeliness. Compared to traditional models relying on uniform thresholds, the experimental results show that landslide density and event numbers have increased from 0.038 events/km² and 44 events to 0.044 events/km² and 59 events, respectively, validating the importance of incorporating spatial heterogeneity and distinct rainfall event types in landslide susceptibility prediction.

Keywords

Landslide susceptibility / Deep embedded clustering (DEC) / Spatial heterogeneity / Mixed-distribution rainfall threshold / Multi-task learning adaptive neural tree model

CRediT authorship contribution statement

Xingfu Zhang: Writing – original draft, Visualization, Validation, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Abi Erdi: Writing – review & editing, Supervision.

Declaration of competing interest

The authors declare the following financial interests (e.g., any funding for the research project)/personal relationships (e.g., the author is an employee of a profitable company) which may be considered as potential competing interests: We declare that we have no financial and personal relationships with other people or organizations that can inappropriately influence our work. Meanwhile, there is no professional or other personal interest of any nature or kind in any product, service and/or company that could be construed as influencing the position presented in, or the review of, the manuscript entitled.

References

[1] Banfi, F., Bevacqua, E., Rivoire, P., Oliveira, S.C., Pinto, J.G., Ramos, A.M., De Michele, C., 2023. Temporal clustering of precipitation for detection of potential landslides. Natural Hazards and Earth System Sci. Discussions 2023, 1-21.

[2] Baum, R.L., Savage, W.Z., Godt, J.W., 2008. TRIGRS: a Fortran Program for Transient Rainfall Infiltration and Grid—based Regional slope—stability Analysis, Version 2.0. US Geological Survey, Reston, VA, USA.

[3] Berti, M., Martina, M., Franceschini, S., Pignone, S., Simoni, A., Pizziolo, M., 2012. Probabilistic rainfall thresholds for landslide occurrence using a Bayesian approach. J. Geophys. Res.: Earth Surf. 117 (F4).

[4] Bhandari, B.P., Dhakal, S., Tsou, C.—Y., 2024. Assessing the prediction accuracy of frequency ratio, weight of evidence, shannon entropy, and information value methods for landslide susceptibility in the Siwalik hills of Nepal. Sustainability 16 (5), 2092.

[5] Change, I.P.O.C., 2007. Climate change 2007: the physical science basis. Agenda 6 (7), 333.

[6] Chen, X., Chen, W., 2021. GIS—Based landslide susceptibility assessment using optimized hybrid machine learning methods. Catena 196, 104833.

[7] Chen, X., Shi, Y., Wen, Y., Shi, R., Mi, X., 2023. Landslide susceptibility evaluation based on certainty factor and support vector machines. Sci. Technol. Eng. 23 (2), 518-527.

[8] Dilley, M., 2005. Natural Disaster Hotspots: a Global Risk Analysis, vol. 5. World Bank Publications.

[9] Guzzetti, F., Reichenbach, P., Cardinali, M., Galli, M., Ardizzone, F., 2005. Probabilistic landslide hazard assessment at the basin scale. Geomorphology 72 (1—4), 272-299.

[10] Leonarduzzi, E., Molnar, P., McArdell, B.W., 2017. Predictive performance of rainfall thresholds for shallow landslides in Switzerland from gridded daily data. Water Resour. Res. 53 (8), 6612-6625.

[11] Li, K., Ni, T., Xue, J., Jiang, Y., 2023. Deep soft clustering: simultaneous deep embedding and soft—partition clustering. J. Ambient Intell. Hum. Comput. 1-13.

[12] Li, Y., Xu, X., Guo, S., 2024. Coupled effects of heavy rainfall and falling water—level exhibit the most significant influences on reservoir bank failures. Earth Surf. Process. Landf. 49 (10), 3177-3188.

[13] Lu, F., Zhang, G., Wang, T., Ye, Y., Zhao, Q., 2025. Geographically weighted random forest based on spatial factor optimization for the assessment of landslide susceptibility. Remote Sens. 17 (9), 1608.

[14] Ma, T., Li, C., Lu, Z., Wang, B., 2014. An effective antecedent precipitation model derived from the power—law relationship between landslide occurrence and rainfall level. Geomorphology 216, 187-192.

[15] Merghadi, A., Yunus, A.P., Dou, J., Whiteley, J., ThaiPham, B., Bui, D.T., Avtar, R., Abderrahmane, B., 2020. Machine learning methods for landslide susceptibility studies: a comparative overview of algorithm performance. Earth Sci. Rev. 207, 103225.

[16] Millán—Arancibia, C., Lavado—Casimiro, W., 2023. Rainfall thresholds estimation for shallow landslides in Peru from gridded daily data. Nat. Hazards Earth Syst. Sci. 23 (3), 1191-1206.

[17] Mu, K., Xie, W., Liu, Q., others, 2022. Research on landslide susceptibility evaluation based on logistic regression and LR coupling model. J. Catastrophol. 37 (3), 212-218.

[18] Naidu, S., Sajinkumar, K., Oommen, T., Anuja, V., Samuel, R.A., Muraleedharan, C., 2018. Early warning system for shallow landslides using rainfall threshold and slope stability analysis. Geosci. Front. 9 (6), 1871-1882.

[19] Petley, D., 2012. Global patterns of loss of life from landslides. Geology 40 (10), 927-930.

[20] Reichenbach, P., Rossi, M., Malamud, B.D., Mihir, M., Guzzetti, F., 2018. A review of statistically—based landslide susceptibility models. Earth Sci. Rev. 180, 60-91.

[21] Shahabi, H., Shirzadi, A., Ronoud, S., Asadi, S., Pham, B.T., Mansouripour, F., Geertsema, M., Clague, J.J., Bui, D.T., 2021. Flash flood susceptibility mapping using a novel deep learning model based on deep belief network, back propagation and genetic algorithm. Geosci. Front. 12 (3), 101100.

[22] Sidle, R.C., Bogaard, T.A., 2016. Dynamic Earth system and ecological controls of rainfall—initiated landslides. Earth Sci. Rev. 159, 275-291.

[23] Wang, Y., Zou, J., Wang, K., Liu, C., Yuan, X., 2023. Semi—supervised deep embedded clustering with pairwise constraints and subset allocation. Neural Netw. 164, 310-322.

[24] Xie, J., Girshick, R., Farhadi, A., 2016. Unsupervised deep embedding for clustering analysis. International Conference on Machine Learning, pp. 478-487.

[25] Zhang, S., Tan, S., Wang, H., Shi, Y., Ding, D., Sun, Y., Gao, H., 2024. To explore the optimal solution of different mapping units and classifiers and their application in the susceptibility evaluation of slope geological disasters. Ecol. Indic. 163, 112073.

[26] Zheng, Y., Jia, C., Yu, J., Li, X., 2023. Deep embedded clustering with distribution consistency preservation for attributed networks. Pattern Recogn. 139, 109469.