Evaluation of landslide hazards susceptibility based on machine learning: Taking the Three Gorges Reservoir Area as an Example
-
摘要: 三峡库区滑坡灾害分布广、数量多、规模大、危害严重，因此开展滑坡灾害易发性评价对该地的地灾防治与处理具有重要参考意义。本文提取了地层岩性、地质构造、坡度、坡向、曲率、斜坡形态、植被指数、水系等17 个因子，选用逻辑回归模型、支持向量机模型、集成学习的梯度提升迭代决策树模型和深度学习中的长短期记忆神经网络与卷积神经网络耦合模型四个机器学习模型进行滑坡灾害易发性评价，选取最优评价模型，完成三峡库区的易发性分区评价，总结研究区易发性空间区划特性。对比四种模型的AUC（Area Under Curve）精度可以得出结论：GBDT模型（Gradient Boosting Decision Tree Model）的AUC精度相对较高，优于其他三个模型，更适合三峡库区的滑坡易发性研究。GBDT的易发性评价结果显示：研究区内极高易发性区域和高易发性区域主要集中于渝东、鄂西一带以及长江沿岸和支流沿岸。研究结果是对整个库区的易发性进行评价，可为后续库区的防灾减灾提供参考。Abstract: Landslide hazards in Three Gorges Reservoir Area are widespread, numerous, large-scale, and seriously disastrous, so carrying out a landslide hazard susceptibility evaluation is of great reference significance for the prevention and treatment of geodetic hazards in the area. In this paper, we extracted 17 factors such as stratigraphic lithology, geological structure, slope, slope direction, curvature, slope morphology, vegetation index, water system, etc., and selected four machine learning models, including logistic regression model, sup port vector machine model, gradient boosting iterative decision tree model with integrated learning, and the coupled model of long and short-term memory neural network and convolutional neural network with deep learning are used for the evaluation of landslide hazard susceptibility, and the optimal evaluation model is selected to complete the evaluation of the Three Gorges reservoir area. The optimal evaluation model is selected to complete the susceptibility zoning evaluation in the Three Gorges reservoir area and to summarize the spatial zoning characteristics of susceptibility in the study area. Comparing the AUC (Area Under Curve) accuracies of the four models, it can be concluded that the GBDT model (Gradient Boosting Decision Tree Model) has a relatively high AUC accuracy, which is better than the other three models, and it is more suitable for the landslide susceptibility study of the Three Gorges reservoir area. The susceptibility evaluation results of GBDT show that the very high susceptibility areas and high susceptibility areas in the study area are mainly concentrated in the east Henan and west Hubei areas, as well as along the Yangtze River and its tributaries. The results of the study evaluate the susceptibility of the whole reservoir area, which could provide a reference for the subsequent hazard prevention and mitigation in the reservoir area.
-
-
[1] 白世彪,王建,闾国年,周平根,侯圣山,徐素宁.2007.基于GIS和双变量分析模型的三峡库区滑坡灾害易发性制图[J].山地学报,25(1):85-92.
[2] 蔡文学,罗永豪,张冠湘,钟慧玲.2017.基于GBDT与Logistic回归融合的个人信贷风险评估模型及实证分析[J].管理现代化,37(2):1-4.
[3] 常宏.2022.三峡库区潜在顺层岩质滑坡识别的宏观判据[J].华南地质,38(2):265-272.
[4] 邓念东,石辉,文强,李宇新,曹晓凡.2021.信息量支持下的随机森林模型的崩塌易发性评价[J].科学技术与工程,21(6):2210-2217.
[5] 何静.2019.基于机器学习的滑坡灾害空间预测及风险评估[D].电子科技大学硕士学位论文.
[6] 胡旭东.2019.基于集成学习的地质灾害易发性评价研究[D].中国地质大学硕士学位论文.
[7] 扈秀宇.2020.基于GIS 与堆栈融合模型的长春市地质灾害易发性和危险性评价研究[D].吉林大学硕士学位论文.
[8] 李远远,梅红波,任晓杰,胡旭东,李梦迪.2018.基于确定性系数和支持向量机的地质灾害易发性评价[J].地球信息科学学报,20(12):1699-1709.
[9] 连志鹏,厉一宁,刘磊,王宁涛.2022.基于ArcGIS 的湖北省远安县地质灾害风险定量评价[J]. 华南地质,38(4):680-688.
[10] 刘艳辉,方然可,苏永超,肖锐铧.2021.基于机器学习的区域滑坡灾害预警模型研究[J]. 工程地质学报,29(1):116-124.
[11] 陆继翔,张琪培,杨志宏,涂孟夫,陆进军,彭晖.2019.基于CNN-LSTM 混合神经网络模型的短期负荷预测方法[J].电力系统自动化,43(8):131-137.
[12] 邱丹丹.2017.基于多源数据融合的滑坡风险分析研究[D].中国地质大学博士学位论文.
[13] 宋花玲.2006.ROC 曲线的评价研究及应用[D].第二军医大学硕士学位论文.
[14] 孙德亮.2019.基于机器学习的滑坡易发性区划与降雨诱发滑坡预报预警研究[D].华东师范大学博士学位论文.
[15] 唐川,马国超.2015.基于地貌单元的小区域地质灾害易发性分区方法研究[J].地理科学,35(1):91-98.
[16] 田春阳.2020.辽宁省西丰县地质灾害易发性评价[D].辽宁师范大学硕士学位论文.
[17] 王芳.2017.万州区滑坡灾害风险评价与管理研究[D].中国地质大学博士学位论文.
[18] 王慧,文海家,胡东萍,谢朋.2015.山地环境地质灾害易发性县级区划研究——以重庆城口县为例[A].//2015 年全国工程地质学术年会论文集,588-593.
[19] 王念秦,郭有金,刘铁铭,朱清华.2019.基于支持向量机模型的滑坡危险性评价[J].科学技术与工程,19(35):70-78.
[20] 夏辉,殷坤龙,梁鑫,马飞.2018.基于SVM-ANN模型的滑坡易发性评价——以三峡库区巫山县为例[J].中国地质灾害与防治学报,29(5):13-19.
[21] 夏辉.2018.三峡库区重庆库段区域滑坡风险评价[D].中国地质大学硕士学位论文.
[22] 许嘉慧,张虹,文海家,孙德亮.2021.基于逻辑回归的巫山县滑坡易发性区划研究[J].重庆师范大学学报(自然科学版),38(2):48-56.
[23] 闫举生,谭建民.2018.基于ANN和LR的远安县滑坡易发性评价[J].山西建筑,44(20):48-50.
[24] 杨光.2020.基于SBAS-InSAR 技术优化的茂县区域滑坡敏感性评价[D].吉林大学硕士学位论文.
[25] 殷宗敏,赵宝强,叶润青.2022.时序InSAR技术在三峡库首区潜在滑坡识别中的应用研究[J]. 华南地质,38(2):273-280.
[26] 张俊,殷坤龙,王佳佳,刘磊,黄发明.2016.三峡库区万州区滑坡灾害易发性评价研究[J].岩石力学与工程学报,35(2):284-296.
[27] 张利芹,李浩,顾超,潘会彬,付鹏伟.2020.基于信息量法的重庆云阳县(三峡库区)地质灾害易发性评价[J].矿产勘查, 11(12):2809-2815.
[28] 赵祈溶,曹顺红,文武飞,周丽芸.2021.基于证据权法的湖南省石门县皂市水库滑坡易发性评价[J].华南地质,37(2):216-225.
[29] 周超.2018.集成时间序列InSAR 技术的滑坡早期识别与预测研究[D].中国地质大学博士学位论文.
[30] Atkinson P M，Massari R. 1998. Generalised linear modelling of susceptibility to landsliding in the Central Apennines, Italy [J]. Computers & Geosciences, 24(4):373-385.
[31] Budetta P, Santo A, Vivenzio F. 2006. Landslide hazard mapping along the coastline of the Cilento region (Italy) by means of a GIS-based parameter rating approach[J]. Geomorphology, 94(3-4):340-352.
[32] Lee S, Ryu J H, Lee M J, Won J S. 2003.Use of an artificial neural network for analysis of the susceptibility to landslides at Boun, Korea[J]. Environmental Geology, 44(7): 820-833.
-
计量
- 文章访问数: 1034
- PDF下载数: 81
- 施引文献: 0
下载: