Modeling and Uncertainty in Landslide Susceptibility Prediction Considering Coupling Mode of Landslide Types
-
摘要: 为全面考虑不同滑坡类型间的差异，提高滑坡易发性预测的准确性与工程应用价值，以重庆市万州区为例，应用支持向量机、C5.0决策树、逻辑回归和多层感知器模型分别对单一堆积层滑坡和岩崩进行易发性建模，再基于综合表现最优模型，分别采用直接耦合法、概率统计法和易发性比较法对不同滑坡类型进行耦合建模，并评估其不确定性.建模结果显示: C5.0决策树模型在单一类型滑坡易发性预测中表现最佳，AUC值均超过0.930；3种耦合方式的易发性预测结果与实际情况基本一致，其中易发性比较法具有较高的AUC值和频率比精度，预测性能最好，其次是概率统计法和直接耦合法.鉴于反映滑坡演化规律的全面性、实际场景的适应性以及防治决策的支持需求，考虑滑坡类型耦合方式的滑坡易发性优于仅考虑单一类型滑坡，但未来仍有必要更深入探索不同类型滑坡间的特异性与耦合方式的优化.Abstract: In order to comprehensively consider the differences between different types of landslides, improve the accuracy and engineering application value of landslide susceptibility prediction, taking Wanzhou District, Chongqing City as an example, support vector machine, C5.0 decision tree, logistic regression, and multilayer perceptron models were applied to model the susceptibility of a single stacked layer landslide and rockfall. Based on the optimal comprehensive performance model, direct coupling method, probability statistics method, and susceptibility comparison method were used to couple and model different types of landslides, and their uncertainties were evaluated. The results show that the C5.0 decision tree model performs best in susceptibility prediction for single-type landslides, with AUC values exceeding 0.930. The susceptibility prediction results of the three coupling methods are generally consistent with the actual situation, among which the comparative susceptibility method exhibits higher AUC values and accuracy in frequency ratio, demonstrating the best predictive performance. The probabilistic statistical method ranks second, followed by the direct coupling method. In terms of comprehensively reflecting the evolution patterns of landslides, adapting to actual scenarios, and supporting prevention and control decisions, considering the coupling of landslide types yields better landslide susceptibility predictions than considering only single-type landslides. However, it is still necessary to explore the specificity among different types of landslides and optimize coupling methods in greater depth in the future.
-
图 4 万州区地表覆被、水文因子和工程地质因子
Fig. 4. The surface cover, hydrological factors, and engineering geological factors in Wanzhou district
图 5 4种模型的堆积层滑坡易发性图
Fig. 5. Landslide susceptibility maps of the accumulation landslide for four models
图 7 堆积层滑坡和岩崩易发性预测的AUC精度
Fig. 7. AUC accuracy for predicting the susceptibility of landslides and rockfalls in sedimentary layers
图 8 考虑滑坡类型不同耦合方式的C5.0决策树模型滑坡易发性分布图
Fig. 8. Landslide susceptibility distribution of C5.0 decision tree model considering different coupling modes of landslide types
图 9 考虑滑坡类型不同耦合方式的C5.0决策树模型滑坡易发性指数分布图
Fig. 9. Landslide susceptibility index distribution of C5.0 decision tree model considering different coupling modes of landslide types
图 10 基于C5.0决策树模型的滑坡易发性AUC精度
Fig. 10. AUC accuracy of landslide susceptibility based on C5.0 decision tree model
图 11 SVM、LR和MLP模型不同耦合方式的滑坡易发性图
Fig. 11. Landslide susceptibility maps of SVM, LR and MLP models in different coupling modes
图 12 SVM、LR和MLP模型不同耦合方式的AUC精度
Fig. 12. AUC accuracy of different coupling modes of SVM, LR and MLP models
表 1 部分基础环境因子的频率比
Table 1. Frequency ratios of some conditioning factors
环境因子 变量值 全区栅格数(个) 堆积层滑坡
栅格数(个)岩崩栅格数(个) 频率比 堆积层滑坡 岩崩 高程(m) 40~309 135 791 5 316 413 2.824 0.347 309~479 255 504 6 568 2 988 1.854 1.336 479~617 374 887 5 506 3 429 1.059 1.045 617~749 358 873 3 437 3 233 0.691 1.029 749~887 269 298 2 313 2 486 0.62 1.054 887~1 043 224 959 1 334 1 677 0.428 0.851 1 043~1 225 132 729 436 1 272 0.237 1.094 1 225~1 640 59 449 202 364 0.245 0.699 坡度(°) 0~6.0 176 522 146 404 0.060 0.261 6.0~12.9 287 215 738 736 0.185 0.294 12.9~18.7 355 417 5 110 1 223 1.037 0.393 18.7~24.0 297 566 6 588 1 634 1.597 0.627 24.0~29.5 328 133 7 149 2 930 1.572 1.020 29.5~35.6 208 177 3 576 3 296 1.239 1.808 35.6~43.1 117 490 1 512 3 215 0.928 3.125 43.1~70.6 40 970 293 2 424 0.516 6.757 平面曲率(m-1) 0~9.9 281 906 5 880 3 644 1.505 1.476 9.9~18.5 372 006 7 531 4 989 1.460 1.532 18.5~27.5 283 466 4 702 2 849 1.197 1.148 27.5~37.4 206 872 2 429 1 528 0.847 0.844 37.4~47.9 160 643 1 262 826 0.567 0.587 47.9~59.1 137 272 823 497 0.432 0.413 59.1~70.9 128 073 659 469 0.371 0.418 70.9~81.5 241 252 1 826 1 060 0.546 0.502 MNDWI -0.65~0.295 3 291 69 20 1.512 0.694 -0.295~0.216 239 660 3 112 2 523 0.937 1.202 -0.216~0.180 433 703 6 351 3 590 1.056 0.945 -0.180~0.141 432 576 6 518 3 529 1.087 0.932 -0.141~0.090 362 882 4 911 3 150 0.976 0.991 -0.090~0.039 205 522 2 147 1 791 0.754 0.995 -0.039~0.134 97 745 1 001 1 186 0.739 1.386 0.134~0.351 36 111 1 003 73 2.004 0.231 NDVI -0.280~0.100 32 786 316 35 0.695 0.122 0.100~0.237 33 204 297 167 0.645 0.574 0.237~0.322 105 611 1 112 286 0.760 0.309 0.322~0.380 182 036 2 463 582 0.976 0.365 0.380~0.448 367 504 5 767 1 837 1.132 0.571 0.448~0.500 372 734 6 163 2 971 1.193 0.910 0.500~0.565 476 685 6 551 5 484 0.991 1.314 0.565~0.751 240 930 2 443 4 500 0.731 2.133 岩性 泥岩 122 739 2 254 395 1.325 0.368 灰岩 1025 323 13 795 9 248 0.971 1.030 砂岩 262 991 6 267 1 443 1.719 0.627 火山岩 112 761 142 1 748 0.091 1.770 石英岩 145 000 1 688 1 092 0.840 0.860 页岩 142 676 966 1 936 0.488 1.550 水系距离(m) 0~100 194 492 1 314 2 300 0.487 1.351 100~300 308 301 7 247 5 038 1.696 1.866 300~600 405 155 8 134 2 924 1.448 0.824 600~1 000 403 438 4 153 2 905 0.743 0.822 1 000~4 540 500 104 4 264 2 695 0.615 0.615 
下载: 导出CSV
表 2 SVM、C5.0决策树和MLP模型的参数
Table 2. The parameters of SVM, C5.0 decision tree and MLP models
模型 参数名称 参数取值 堆积层滑坡 岩崩 SVM 停止标准 0.001 规则参数 6 10 回归精确度 0.1 核参数 0.09 C5.0 修剪严重性 80 试验次数 10 折叠次数 10 子分支最小记录数 15 MLP 学习率 0.01 动量 0.35 0.4 迭代时间 2 000
下载: 导出CSV
表 3 不同机器学习模型下各级堆积层滑坡和岩崩的易发性频率比值
Table 3. The frequency ratio of susceptibility to landslides and rockfalls in different levels of accumulation layers under different machine learning models
滑坡类型 分类 滑坡易发性预测各级别的频率比 SVM C5.0 LR MLP 堆积层滑坡 极低 0.100 0.006 0.089 0.094 低 0.320 0.065 0.315 0.333 中等 0.689 0.426 0.695 0.720 高 1.405 2.925 1.484 1.347 极高 3.851 13.674 4.078 3.579 频率比精度 82.5% 97.1% 83.5% 81.1% 平均值 0.363 0.230 0.362 0.389 标准差 0.24 0.247 0.244 0.263 岩崩 极低 0.027 0.008 0.031 0.030 低 0.253 0.080 0.276 0.250 中等 0.599 0.434 0.656 0.680 高 1.474 2.575 1.557 1.623 极高 8.719 15.054 8.726 8.352 频率比精度 92.1% 97.1% 91.4% 91.2% 平均值 0.230 0.198 0.249 0.243 标准差 0.253 0.255 0.249 0.266
下载: 导出CSV
表 4 C5.0决策树模型预测滑坡易发性的滑坡频率比
Table 4. Landslide frequency ratio for predicting landslide susceptibility using C5.0 decision tree
C5.0决策树 分类 全区占比(%) 滑坡岩崩栅格数(个) 发生占比(%) 频率比 频率比精度(%) 直接耦合法 极低 39.1 1 084 2.6 0.068 84.4 低 20.6 2 784 6.8 0.330 中等 15.9 5 192 12.7 0.797 高 12.6 9 259 22.6 1.791 极高 11.7 22 655 55.3 4.709 概率统计法 极低 30.5 91 0.2 0.007 97.0 低 22.5 376 0.9 0.041 中等 19.3 1 580 3.9 0.200 高 16.3 7826 19.1 1.169 极高 11.3 31 101 75.9 6.689 易发性比较法 极低 32.5 79 0.2 0.006 96.5 低 24.4 493 1.2 0.049 中等 19.2 2 100 5.1 0.267 高 14.3 9 254 22.6 1.576 极高 9.6 29 048 70.9 7.385
下载: 导出CSV
表 5 4种模型在不同耦合方式下滑坡易发性指数统计特征
Table 5. Statistical characteristics of landslide susceptibility indexes under different coupling modes
模型 直接耦合法 概率统计法 易发性比较法 平均值 标准差 平均值 标准差 平均值 标准差 SVM 0.368 0.242 0.431 0.257 0.426 0.259 C5.0 0.313 0.280 0.348 0.288 0.325 0.291 LR 0.366 0.242 0.361 0.247 0.360 0.247 MLP 0.382 0.252 0.453 0.281 0.449 0.286
下载: 导出CSV
-
Cao, J., Zhang, Z., Wang, C. Z., et al., 2019. Susceptibility Assessment of Landslides Triggered by Earthquakes in the Western Sichuan Plateau. CATENA, 175: 63-76. https://doi.org/10.1016/j.catena.2018.12.013 Chen, W., Pourghasemi, H. R., Naghibi, S. A., 2018. A Comparative Study of Landslide Susceptibility Maps Produced Using Support Vector Machine with Different Kernel Functions and Entropy Data Mining Models in China. Bulletin of Engineering Geology and the Environment, 77(2): 647-664. https://doi.org/10.1007/s10064-017-1010-y Deng, M. D., Ju, N. P., Wu, T. W., et al., 2024. Evaluation of Susceptibility under Different Landslide Sample Points and Polygonal Expression Modes. Earth Science, 49(5): 1565-1583(in Chinese with English abstract). 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). Feby, B., Achu, A. L., Jimnisha, K., et al., 2020. Landslide Susceptibility Modelling Using Integrated Evidential Belief Function Based Logistic Regression Method: A Study from Southern Western Ghats, India. Remote Sensing Applications: Society and Environment, 20: 100411. https://doi.org/10.1016/j.rsase.2020.100411 Fu, Z. Y., Li, D. Q., Wang, S., et al., 2023. Landslide Susceptibility Assessment Based on Multitemporal Landslide Inventories and TrAdaBoost Transfer Learning. Earth Science, 48(5): 1935-1947(in Chinese with English abstract). Gao, X. Y., Wang, J. Z., Mao, X., et al., 2023. The Susceptibility Assessment of Landslide Based on Bi-GRU Network. Science of Surveying and Mapping, 48(4): 221-230(in Chinese with English abstract). Golkarian, A., Naghibi, S. A., Kalantar, B., et al., 2018. Groundwater Potential Mapping Using C5.0, Random Forest, and Multivariate Adaptive Regression Spline Models in GIS. Environmental Monitoring and Assessment, 190: 149. https://doi.org/10.1007/s10661-018-6507-8 Guo, Z. Z., Shi, Y., Huang, F. M., et al., 2021. Landslide Susceptibility Zonation Method Based on C5.0 Decision Tree and K-Means Cluster Algorithms to Improve the Efficiency of Risk Management. Geoscience Frontiers, 12(6): 101249. https://doi.org/10.1016/j.gsf.2021.101249 Guo, Z. Z., Yin, K. L., Huang, F. M., et al., 2019. Evaluation of Landslide Susceptibility Based on Landslide Classification and Weighted Frequency Ratio Model. Chinese Journal of Rock Mechanics and Engineering, 38(2): 287-300(in Chinese with English abstract). He, R. J., Zhang, W. G., Dou, J., et al., 2024. Application of Artificial Intelligence in Three Aspects of Landslide Risk Assessment: A Comprehensive Review. Rock Mechanics Bulletin, 3(4): 100144. https://doi.org/10.1016/j.rockmb.2024.100144 Hong, H. Y., Tsangaratos, P., Ilia, I., et al., 2020. Introducing a Novel Multi-Layer Perceptron Network Based on Stochastic Gradient Descent Optimized by a Meta-Heuristic Algorithm for Landslide Susceptibility Mapping. Science of the Total Environment, 742: 140549. https://doi.org/10.1016/j.scitotenv.2020.140549 Huang, F. M., Teng, Z. K., Guo, Z. Z., et al., 2023. Uncertainties of Landslide Susceptibility Prediction: Influences of Different Spatial Resolutions, Machine Learning Models and Proportions of Training and Testing Dataset. Rock Mechanics Bulletin, 2(1): 100028. https://doi.org/10.1016/j.rockmb.2023.100028 Huang, F. M., Xiong, H. W., Yao, C., et al., 2023. Uncertainties of Landslide Susceptibility Prediction Considering Different Landslide Types. Journal of Rock Mechanics and Geotechnical Engineering, 15(11): 2954-2972. https://doi.org/10.1016/j.jrmge.2023.03.001 Huang, F. M., Ye, Z., Jiang, S. H., et al., 2021. Uncertainty Study of Landslide Susceptibility Prediction Considering the Different Attribute Interval Numbers of Environmental Factors and Different Data-Based Models. CATENA, 202: 105250. https://doi.org/10.1016/j.catena.2021.105250 Huang, X., Huang, J., He, Z. C., et al., 2024. Study on the Method of Estimating the Volume of Fragmental Rockfall Based on Image Recognition. Hydrogeology & Engineering Geology, 51(3): 140-148(in Chinese with English abstract). Hungr, O., Leroueil, S., Picarelli, L., 2014. The Varnes Classification of Landslide Types, an Update. Landslides, 11(2): 167-194. https://doi.org/10.1007/s10346-013-0436-y Kaya Topaçli, Z., Ozcan, A. K., Gokceoglu, C., 2024. Performance Comparison of Landslide Susceptibility Maps Derived from Logistic Regression and Random Forest Models in the Bolaman Basin, Türkiye. Natural Hazards Review, 25(1): 04023054. https://doi.org/10.1061/nhrefo.nheng-1771 Li, W. B., Fan, X. M., Huang, F. M., et al., 2021. Uncertainties of Landslide Susceptibility Modeling under Different Environmental Factor Connections and Prediction Models. Earth Science, 46(10): 3777-3795(in Chinese with English abstract). Liu, H. Z., Xu, H., Bao, H. J., et al., 2022. Optimization Experiment of Early Identification of Landslides Susceptibility Areas in Medium and Small Mountainous Catchment Based on Ensemble Learning. Advanced Engineering Sciences, 54(6): 12-20(in Chinese with English abstract). Liu, J. P., Liang, E. J., Xu, S. H., et al., 2022. Multi-Kernel Support Vector Machine Considering Sample Optimization Selection for Analysis and Evaluation of Landslide Disaster Susceptibility. Acta Geodaetica et Cartographica Sinica, 51(10): 2034-2045(in Chinese with English abstract). Liu, Z. Q., Gilbert, G., Cepeda, J. M., et al., 2021. Modelling of Shallow Landslides with Machine Learning Algorithms. Geoscience Frontiers, 12(1): 385-393. https://doi.org/10.1016/j.gsf.2020.04.014 Ma, X. J., Zhang, Y. F., Hou, L. J., et al., 2021. Study on Classification and Formation Mechanism of Landslide in Red Bed Area. Railway Engineering, 61(2): 75-78(in Chinese with English abstract). Miao, F. S., Wu, Y. P., Li, L. W., et al., 2021. Prediction of Joint Roughness Coefficient of Rock Mass Based on Boosting-Decision Tree C5.0. Journal of Zhejiang University (Engineering Science), 55(3): 483-490(in Chinese with English abstract). Nava, L., Carraro, E., Reyes-Carmona, C., et al., 2023. Landslide Displacement Forecasting Using Deep Learning and Monitoring Data across Selected Sites. Landslides, 20(10): 2111-2129. https://doi.org/10.1007/s10346-023-02104-9 Nsengiyumva, J. B., Luo, G. P., Amanambu, A. C., et al., 2019. Comparing Probabilistic and Statistical Methods in Landslide Susceptibility Modeling in Rwanda/Centre-Eastern Africa. Science of The Total Environment, 659: 1457-1472. https://doi.org/10.1016/j.scitotenv.2018.12.248 Song, Z. F., Zhang, Y., She, T., et al., 2023. Determination of Regional Landslide Rainfall Warning Threshold Based on Susceptibility Zoning: A Case Study in Longling County of Yunnan Province. The Chinese Journal of Geological Hazard and Control, 34(4): 22-29(in Chinese with English abstract). Sun, C. M., Ma, R. Y., Shang, H. X., et al., 2020. Landslide Susceptibility Assessment in Xining Based on Landslide Classification. Hydrogeology & Engineering Geology, 47(3): 173-181(in Chinese with English abstract). Tang, H. M., Li, C. D., Gong, W. P., et al., 2022. Fundamental Attribute and Research Approach of Landslide Evolution. Earth Science, 47(12): 4596-4608(in Chinese with English abstract). Tien Bui, D., Tuan, T. A., Klempe, H., et al., 2016. Spatial Prediction Models for Shallow Landslide Hazards: A Comparative Assessment of the Efficacy of Support Vector Machines, Artificial Neural Networks, Kernel Logistic Regression, and Logistic Model Tree. Landslides, 13(2): 361-378. https://doi.org/10.1007/s10346-015-0557-6 Varnes, D. J., 1958. Landslide Types and Processes. Landslides and Engineering Practice, 24: 20-47. Xu, C., Dai, F. C., Xu, X. W., et al., 2012. GIS-Based Support Vector Machine Modeling of Earthquake-Triggered Landslide Susceptibility in the Jianjiang River Watershed, China. Geomorphology, 145: 70-80. https://doi.org/10.1016/j.geomorph.2011.12.040 Zeng, T. R., Wang, L. F., Zhang, Y., et al., 2024. Landslide Susceptibility Modeling and Interpretability Based on CatBoost-SHAP Model. The Chinese Journal of Geological Hazard and Control, 35(1): 37-50(in Chinese with English abstract). Zhang, L. F., Wang, J. Y., Zhang, M. S., et al., 2022. Evaluation of Regional Landslide Susceptibility Assessment Based on BP Neural Network. Northwestern Geology, 55(2): 260-270(in Chinese with English abstract). Zhang, X. C., Jiang, Y. L., Wang, Y. J., et al., 2024. Evaluation of Landslide Susceptibility Based on Multi-Objective Optimization Method. Journal of Soil and Water Conservation, 38(1): 104-112, 121(in Chinese with English abstract). Zhang, Y. B., Xu, P. Y., Liu, J., et al., 2023. Comparison of LR, 5-CV SVM, GA SVM, and PSO SVM for Landslide Susceptibility Assessment in Tibetan Plateau Area, China. Journal of Mountain Science, 20(4): 979-995. https://doi.org/10.1007/s11629-022-7685-y Zhou, C., Yin, K. L., Cao, Y., et al., 2018. Landslide Susceptibility Modeling Applying Machine Learning Methods: A Case Study from Longju in the Three Gorges Reservoir Area, China. Computers & Geosciences, 112: 23-37. https://doi.org/10.1016/j.cageo.2017.11.019 邓明东, 巨能攀, 吴天伟, 等, 2024. 不同滑坡样本点和多边形表达模式下的易发性评价. 地球科学, 49(5): 1565-1583. 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 付智勇, 李典庆, 王顺, 等, 2023. 基于多时空滑坡编录和TrAdaBoost迁移学习的滑坡易发性评价. 地球科学, 48(5): 1935-1947. 高轩宇, 王继周, 毛曦, 等, 2023. 双向门控循环单元网络的滑坡易发性评价. 测绘科学, 48(4): 221-230. 郭子正, 殷坤龙, 黄发明, 等, 2019. 基于滑坡分类和加权频率比模型的滑坡易发性评价. 岩石力学与工程学报, 38(2): 287-300. 黄祥, 黄健, 贺子城, 等, 2024. 基于图像识别估算碎裂岩崩体积方法研究. 水文地质工程地质, 51(3): 140-148. 李文彬, 范宣梅, 黄发明, 等, 2021. 不同环境因子联接和预测模型的滑坡易发性建模不确定性. 地球科学, 46(10): 3777-3795. 刘海知, 徐辉, 包红军, 等, 2022. 基于集成学习的山区中小流域滑坡易发区早期识别优化试验. 工程科学与技术, 54(6): 12-20. 刘纪平, 梁恩婕, 徐胜华, 等, 2022. 顾及样本优化选择的多核支持向量机滑坡灾害易发性分析评价. 测绘学报, 51(10): 2034-2045. 马贤杰, 张玉芳, 侯李杰, 等, 2021. 红层地区滑坡的分类及形成机制. 铁道建筑, 61(2): 75-78. 苗发盛, 吴益平, 李麟玮, 等, 2021. 基于Boosting-决策树C5.0的岩体结构面粗糙度预测. 浙江大学学报(工学版), 55(3): 483-490. 宋昭富, 张勇, 佘涛, 等, 2023. 基于易发性分区的区域滑坡降雨预警阈值确定: 以云南龙陵县为例. 中国地质灾害与防治学报, 34(4): 22-29. 孙长明, 马润勇, 尚合欣, 等, 2020. 基于滑坡分类的西宁市滑坡易发性评价. 水文地质工程地质, 47(3): 173-181. 唐辉明, 李长冬, 龚文平, 等, 2022. 滑坡演化的基本属性与研究途径. 地球科学, 47(12): 4596-4608. 曾韬睿, 王林峰, 张俞, 等, 2024. 基于CatBoost-SHAP模型的滑坡易发性建模及可解释性. 中国地质灾害与防治学报, 35(1): 37-50. 张林梵, 王佳运, 张茂省, 等, 2022. 基于BP神经网络的区域滑坡易发性评价. 西北地质, 55(2): 260-270. 张兴存, 蒋玉琳, 王玉杰, 等, 2024. 基于多目标优化方法的滑坡易发性评价. 水土保持学报, 38(1): 104-112, 121. -
点击查看大图
图(12) / 表(5)
计量
- 文章访问数: 198
- HTML全文浏览量: 7
- PDF下载量: 7
- 被引次数: 0
