Landslide Susceptibility Mapping Based on AI Technology
-
摘要: 基于AI技术的滑坡易发性制图具有高效、准确的优点，为推进其在滑坡灾害防治中的应用，在介绍和总结机器学习、深度学习以及集成学习模型的原理及特点的基础上，选择支持向量机、深度随机森林和随机森林等代表性模型在陕西省略阳县进行了应用分析；探讨了AI技术在滑坡易发性领域的应用及发展方向.研究结果表明：基于决策树的集成学习模型相比于逻辑回归、支持向量机等，表现出更高的效能，AUC值在0.90以上；LSM（landslide susceptibility mapping）中常用的类不平衡采样策略下，基于Boosting的集成模型具有优势，并且其受采样比的影响相对较小；对抗生成网络可以提高在深度学习模型在数据限制情况下的性能，本文中AUC值从0.77提升至0.82；滑坡理论模型与AI数据模型相结合，具有巨大的潜力；通过充分利用时序数据的AI模型可以提升模型的性能，并有助于揭示滑坡的链式灾害效应和时空演化特征；进行各种学习模型的系统性研究，对于AI技术在滑坡易发性制图中的应用具有重要的意义.Abstract: Utilizing AI technology for landslide susceptibility mapping offers the advantages of efficiency and accuracy.To promote its application in landslide disaster prevention and control, in this article it introduces and summarizes the principles and characteristics of machine learning, deep learning, and ensemble learning models. Representative models such as Support Vector Machines, Deep Random Forests, and Random Forests were applied for analysis in Lueyang County, Shaanxi Province. It discusses the application and development directions of AI technology in the field of landslide susceptibility. The results indicate that ensemble learning models based on decision trees, compared to logistic regression and support vector machines, demonstrate higher efficacy with AUC values above 0.90. Under the commonly used class imbalance sampling strategy in LSM (landslide susceptibility mapping), ensemble models based on Boosting show advantages and are relatively less affected by sampling ratios. Generative Adversarial Networks can enhance the performance of deep learning models under data constraints, where in this study, the AUC value increased from 0.77 to 0.82. Combining landslide theoretical models with AI data models has great potential; leveraging AI models that fully utilize time-series data can improve model performance and help reveal the chain disaster effects and spatio-temporal evolution characteristics of landslides; conducting systematic studies of various learning models is of significant importance for the application of AI technology in landslide susceptibility mapping.
-
图 1 机器学习的发文趋势(据Liu et al., 2023)
Fig. 1. Publication trends on machine learning (Liu et al., 2023)
图 7 滑坡易发性制图因子图
a.坡向图；b.坡度图；c. 岩性图；d. 起伏度图；e. 曲率图；f. NDVI图；g. 断层距离图；h. 道路距离图；i. 水系距离图
Fig. 7. Landslide susceptibility mapping factor map
图 8 多种AI模型ROC曲线
a.采样比1∶1多种AI模型ROC曲线；b.采样比8∶2多种AI模型ROC曲线
Fig. 8. ROC curves of various AI models
图 9 深度随机森林模型ROC曲线
a.采样比1∶1深度随机森林模型ROC曲线；b.采样比8∶2深度随机森林模型ROC曲线
Fig. 9. ROC curve of deep random forest model
图 10 集成对抗神经网络的深度随机森林模型ROC曲线
a.采样比1∶1对抗神经网络与深度随机森林集成模型ROC曲线；b.采样比8∶2对抗神经网络与深度随机森林集成模型ROC曲线
Fig. 10. ROC curve of deep random forest model with integrated generative adversarial networks
表 1 机器学习基本原理及适用性分析
Table 1. Basic principles and applicability analysis of machine learning
模型 基本原理 在LSM中的应用与研究 优劣势分析 逻辑回归 线性模型广泛应用于二元分类问题，通过输入变量的线性函数(决策边界)将研究对象分为两个类别. 逻辑回归的目标是找到最合理的模型来表述滑坡与非滑坡的不同条件因素(如坡度、高程和坡向)之间的关系. 逻辑回归是线性模型，能够清晰解释每个特征的影响方向和大小，但是其不能很好地处理特征与滑坡之间的非线性关系. 决策树 决策树是一种基本的分类工具，通过不断将空间中的对象分组，逐步提高分类精度.它会根据不同特征逐层分类，直到达到所需的精确度. Saito et al. (2009)将决策树用于分析日本明石山大范围滑坡易发性，揭示滑坡发生与地形、地质条件的关系，指出决策树的优势是可解释性强.Alkhasawneh et al.(2014)用决策树模型确定造成马来西亚槟城岛大范围滑坡的24个因素的重要性，并通过集成学习方法对决策树的增强带来了更强的性能. 决策树和基于树的模型有较高的性能且有强的可解释性，但存在过拟合的现象. 支持向量机 支持向量机通过找到一个超平面来分类数据，目标是使数据点到超平面的间隔最大.关键作用的是距离超平面最近的“支持向量”点.核技巧是支持向量机的重要方法，它允许将低维无法线性分割的数据映射到更高维空间，使得数据可以被线性分割.这一过程依赖于核函数. 支持向量机的在LSM中的应用非常广泛(徐胜华等, 2020).有学者对支持向量机在LSM中的应用做了详细的综述并指出将该模型与其他机器学习或深度学习集成将带来更强大的性能(Huang and Zhao, 2018). SVM的显著优势是泛化错误率低，并且在样本数据不足或数据异常的情况下能保持高性能.但它比较依赖超参数的选择，且当特征维度高时表现不稳定. 
下载: 导出CSV
表 2 集成学习基本原理及适用性分析
Table 2. Basic principles and applicability analysis of ensemble learning
集成模型 基本原理 在LSM中的应用与研究 优劣势分析 Boosting Boosting是多个基学习器串行的集成方法.这种算法是先由初始训练数据训练一个基学习器，再根据基学习器的测试表现对样本数据分布进行调整. Sahin (2022)比较了梯度增强机(GBM)、分类增强器(CatBoost)、极端梯度增强(XGBoost)和轻度增强机(LightGBM)在LSM中应用的性能，研究结果表明Boosting在LSM中可以构建更加有效的模型. Boosting算法可以显著提升模型的预测准确性，能够很好地处理数据中复杂的非线性关系，适用于滑坡预测等复杂问题.但对于Boosting算法容易被数据集中的噪声和异常值影响，对类不平衡数据需要进行额外处理. Bagging (Breiman, 1996) Bagging是多个基学习器并行的集成方法，每个学习器的预测结果通过简单投票得出最终结果. 随机森林(Breiman, 2001)是Bagging的拓展，它在以决策树为基学习器的训练中加入了随机特征选择.Zhang et al. (2023)发现RF在重庆市奉节县的LSM应用性能优于XGBoost. 随机森林具有计算成本低、便于超参数整定以及在求解复杂非线性问题时具有鲁棒性等优点(Taalab et al., 2018).Bagging将多个模型的结果进行平均或投票，形成的最终模型较难解释.
下载: 导出CSV
-
Alkhasawneh, M. S., Ngah, U. K., Tay, L. T., et al., 2014. Modeling and Testing Landslide Hazard Using Decision Tree. Journal of Applied Mathematics, (1): 929768. https://doi.org/10.1155/2014/929768 Breiman, L, 1996. Bagging Predictors. Machine Learning, 24: 123-140. Breiman, L, 2001. Random Forests. Mach. Learn., 45(1): 5-32. https://doi.org/10.1023/a:1010933404324 Catani, F., Lagomarsino, D., Segoni, S., et al., 2013. Landslide Susceptibility Estimation by Random Forests Technique: Sensitivity and Scaling Issues. Natural Hazards and Earth System Sciences, 13(11): 2815-2831. https://doi.org/10.5194/nhess-13-2815-2013 Chang, Z. L., Catani, F., Huang, F. M., et al., 2023. Landslide Susceptibility Prediction Using Sope Unit-Based Machine Learning Models Considering the Heterogeneity of Conditioning Factors. Journal of Rock Mechanics and Geotechnical Engineering, 15(5): 1127-1143. https://doi.org/10.1016/j.jrmge.2022.07.009 Dikshit, A., Pradhan, B., Alamri, A. M., 2021. Pathways and Challenges of the Application of Artificial Intelligence to Geohazards Modelling. Gondwana Research, 100: 290-301. https://doi.org/10.1016/j.gr.2020.08.007 Dosovitskiy, A., Beyer, L., Kolesnikov, A., et al., 2020. An Image is Worth 16×16 Words: Transformers for Image Recognition at Scale. ArXiv e-Prints: arXiv: 2010.11929. https://doi.org/10.48550/arXiv.2010.11929 Dou, J., Xiang, Z. L., Xu, Q., et al., 2023. Application and Development Trend of Machine Learning in Landslide Intelligent Disaster Prevention and Mitigation. Earth Science, 48(5): 1657-1674(in Chinese with English abstract). Fang, Z. C., Wang, Y., Peng, L., et al., 2020. Integration of Convolutional Neural Network and Conventional Machine Learning Classifiers for Landslide Susceptibility Mapping. Computers & Geosciences, 139: 104470. https://doi.org/10.1016/j.cageo.2020.104470 Hochreiter, S., Schmidhuber, J., 1997. Long Short-Term Memory. Neural Computation, 9(8): 1735-1780. https://doi.org/10.1162/neco.1997.9.8.1735 Huang, Y., Zhao, L., 2018. Review on Landslide Susceptibility Mapping Using Support Vector Machines. Catena, 165: 520-529. https://doi.org/10.1016/j.catena.2018.03.003 Jia, J., Mao, Y. M., Meng, X. J., et al., 2023. Comparison of Landslide Susceptibility Evaluation by Deep Random Forest and Random Forest Model: A Case Study of Lueyang County, Hanzhong City. Northwestern Geology, 56(3): 239-249(in Chinese with English abstract). Kirschbaum, D., Stanley, T., 2018. Satellite-Based Assessment of Rainfall-Triggered Landslide Hazard for Situational Awareness. Earth's Future, 6(3): 505-523. https://doi.org/10.1002/2017ef000715 Krizhevsky, A., Sutskever, I., Hinton, G., 2012. ImageNet Classification with Deep Convolutional Neural Networks. Communications of the ACM, 60(6): 84-90. https://doi.org/10.1145/3065386 Li, L. P., Lan, H. X., Guo, C. B., et al., 2017. Geohazard Susceptibility Assessment along the Sichuan-Tibet Railway and Its Adjacent Area Using an Improved Frequency Ratio Method. Geoscience, 31(5): 911-929 (in Chinese with English abstract). Li, X. P., Chong, J. X., Lu, Y. B., et al., 2022. Application of Information Gain in the Selection of Factors for Regional Slope Stability Evaluation. Bulletin of Engineering Geology and the Environment, 81(11): 470. https://doi.org/10.1007/s10064-022-02970-y Liu, F., Deng, Y. H., Mu, H. D., et al., 2023. A Study of the Stability Evaluation Method of Rainfall-Induced Shallow Loess Landslides Based on the Maxent-Sinmap Slope Model. Hydrogeology & Engineering Geology, 50(5): 146-158(in Chinese with English abstract). Liu, S. L., Wang, L. Q., Zhang, W. G., et al., 2023. A Comprehensive Review of Machine Learning-Based Methods in Landslide Susceptibility Mapping. Geological Journal, 58(6): 2283-2301. https://doi.org/10.1002/gj.4666 Lyu, L., Chen, T., Li, J., 2023. An Interpretation Study on the ML Models for Landslide Susceptibility Mapping. IGARSS 2023—2023 IEEE International Geoscience and Remote Sensing Symposium. July 16-21, 2023, Pasadena, CA, USA. IEEE, 7241-7244. https://doi.org/10.1109/igarss52108.2023.10282171 Miao, F. S., Wu, Y. P., Xie, Y. H., et al., 2018. Prediction of Landslide Displacement with Step-Like Behavior Based on Multialgorithm Optimization and a Support Vector Regression Model. Landslides, 15(3): 475-488. https://doi.org/10.1007/s10346-017-0883-y Miao, F. S., Zhao, F. C., Wu, Y. P., et al., 2023. Landslide Susceptibility Mapping in Three Gorges Reservoir Area Based on GIS and Boosting Decision Tree Model. Stochastic Environmental Research and Risk Assessment, 37(6): 2283-2303. https://doi.org/10.1007/s00477-023-02394-4 Ngo, P. T. T., Panahi, M., Khosravi, K., et al., 2021. Evaluation of Deep Learning Algorithms for National Scale Landslide Susceptibility Mapping of Iran. Geoscience Frontiers, 12(2): 505-519. https://doi.org/10.1016/j.gsf.2020.06.013 Pham, B. T., Prakash, I., Singh, S. K., et al., 2019. Landslide Susceptibility Modeling Using Reduced Error Pruning Trees and Different Ensemble Techniques: Hybrid Machine Learning Approaches. Catena, 175: 203-218. https://doi.org/10.1016/j.catena.2018.12.018 Pradhan, B., Dikshit, A., Lee, S., et al., 2023. An Explainable AI (XAI) Model for Landslide Susceptibility Modeling. Applied Soft Computing, 142. https://doi.org/10.1016/j.asoc.2023.110324 Reichenbach, P., Rossi, M., Malamud, B. D., et al., 2018. A Review of Statistically-Based Landslide Susceptibility Models. Earth-Science Reviews, 180: 60-91. https://doi.org/10.1016/j.earscirev.2018.03.001 Sahin, E. K., 2022. Comparative Analysis of Gradient Boosting Algorithms for Landslide Susceptibility Mapping. Geocarto International, 37(9): 2441-2465. https://doi.org/10.1080/10106049.2020.1831623 Saito, H., Nakayama, D., Matsuyama, H., 2009. Comparison of Landslide Susceptibility Based on a Decision-Tree Model and Actual Landslide Occurrence: The Akaishi Mountains, Japan. Geomorphology, 109(3-4): 108-121. https://doi.org/10.1016/j.geomorph.2009.02.026. Stott, P., 2016. How Climate Change Affects Extreme Weather Events. Science, 352(6293): 1517-1518. https://doi.org/10.1126/science.aaf7271 Taalab, K., Cheng, T., Zhang, Y., 2018. Mapping Landslide Susceptibility and Types Using Random Forest. Big Earth Data, 2(2): 159-178. https://doi.org/10.1080/20964471.2018.1472392 Tian, N. M., Lan, H. X., Wu, Y. M., et al., 2020. Performance Comparison of BP Artificial Neural Network and CART Decision Tree Model in Landslide Susceptibility Prediction. Journal of Geo-Information Science, 22(12): 2304-2316(in Chinese with English abstract). van Westen, C. J., van Asch, T. W. J., Soeters, R., 2006. Landslide Hazard and Risk Zonation: Why is It still so Difficult?Bulletin of Engineering Geology and the Environment, 65(2): 167-184. https://doi.org/10.1007/s10064-005-0023-0 Wang, H. J., Wang, L., Zhang, L. M., 2023. Transfer Learning Improves Landslide Susceptibility Assessment. Gondwana Research, 123: 238-254. https://doi.org/10.1016/j.gr.2022.07.008 Wang, H. J., Zhang L., Luo H., et al., 2021. AI-Powered Landslide Susceptibility Assessment in Hong Kong. Engineering Geology, 288: 106103. https://doi.org/10.1016/j.enggeo.2021.106103 Wang, Y., Fang, Z. C., Hong, H. Y., 2019. Comparison of Convolutional Neural Networks for Landslide Susceptibility Mapping in Yanshan County, China. Science of the Total Environment, 666: 975-993. https://doi.org/10.1016/j.scitotenv.2019.02.263 Wang, Y. K., Tang, H. M., Huang, J. S., et al., 2022. A Comparative Study of Different Machine Learning Methods for Reservoir Landslide Displacement Prediction. Engineering Geology, 298: 106544. https://doi.org/10.1016/j.enggeo.2022.106544 Ward, P. J., Blauhut, V., Bloemendaal, N., et al., 2020. Review Article: Natural Hazard Risk Assessments at the Global Scale. Natural Hazards and Earth System Sciences, 20(4): 1069-1096. https://doi.org/10.5194/nhess-20-1069-2020 Xu, S. H., Liu, J. P., Wang, X. H., et al., 2020. Landslide Susceptibility Assessment Method Incorporating Index of Entropy Based on Support Vector Machine: A Case Study of Shaanxi Province. Geomatics and Information Science of Wuhan University, 45(8): 1214-1222(in Chinese with English abstract). Yang, C., Liu, L. L., Huang, F. M., et al., 2023. Machine Learning-Based Landslide Susceptibility Assessment with Optimized Ratio of Landslide to Non-Landslide Samples. Gondwana Research, 123: 198-216. https://doi.org/10.1016/j.gr.2022.05.012 Zhang, W. G., He, Y. W., Wang, L. Q., et al., 2023. Landslide Susceptibility Mapping Using Random Forest and Extreme Gradient Boosting: A Case Study of Fengjie, Chongqing. Geological Journal, 58(6): 2372-2387. https://doi.org/10.1002/gj.4683 Zhu, J. J., Yang, M. Q., Ren, Z. J., 2023. Machine Learning in Environmental Research: Common Pitfalls and Best Practices. Environmental Science & Technology, 57(46): 17671-17689. https://doi.org/10.1021/acs.est.3c00026 Zuccaro, G., De Gregorio, D., Leone, M. F., 2018. Theoretical Model for Cascading Effects Analyses. International Journal of Disaster Risk Reduction, 30: 199-215. https://doi.org/10.1016/j.ijdrr.2018.04.019 窦杰, 向子林, 许强, 等, 2023. 机器学习在滑坡智能防灾减灾中的应用与发展趋势. 地球科学, 48(5): 1657-1674. doi: 10.3799/dqkx.2022.419 贾俊, 毛伊敏, 孟晓捷, 等, 2023. 深度随机森林和随机森林算法的滑坡易发性评价对比: 以汉中市略阳县为例. 西北地质, 56(3): 239-249. 李郎平, 兰恒星, 郭长宝, 等, 2017. 基于改进频率比法的川藏铁路沿线及邻区地质灾害易发性分区评价. 现代地质, 31(5): 911-929. 刘凡, 邓亚虹, 慕焕东, 等, 2023. 基于最大熵-无限边坡模型的降雨诱发浅层黄土滑坡稳定性评价方法研究. 水文地质工程地质, 50(5): 146-158. 田乃满, 兰恒星, 伍宇明, 等, 2020. 人工神经网络和决策树模型在滑坡易发性分析中的性能对比. 地球信息科学学报, 22(12): 2304-2316. 徐胜华, 刘纪平, 王想红, 等, 2020. 熵指数融入支持向量机的滑坡灾害易发性评价方法: 以陕西省为例. 武汉大学学报(信息科学版), 45(8): 1214-1222. -
点击查看大图
图(10) / 表(2)
计量
- 文章访问数: 8
- HTML全文浏览量: 3
- PDF下载量: 2
- 被引次数: 0
