小样本机器学习方法在新疆滑坡灾害易发性评价中的应用

梁龙飞

梁龙飞. 2023. 小样本机器学习方法在新疆滑坡灾害易发性评价中的应用[J].工程地质学报, 31(4): 1394-1406. doi: 10.13544/j.cnki.jeg.2023-0226
引用本文: 梁龙飞. 2023. 小样本机器学习方法在新疆滑坡灾害易发性评价中的应用[J].工程地质学报, 31(4): 1394-1406. doi: 10.13544/j.cnki.jeg.2023-0226
Liang Longfei. 2023. Application of small-sample machine learning method in evaluation of landslide disaster susceptibility in Xinjiang[J]. Journal of Engineering Geology, 31(4):1394-1406. doi: 10.13544 /j.cnki.jeg.2023-0226
Citation: Liang Longfei. 2023. Application of small-sample machine learning method in evaluation of landslide disaster susceptibility in Xinjiang[J]. Journal of Engineering Geology, 31(4):1394-1406. doi: 10.13544 /j.cnki.jeg.2023-0226

小样本机器学习方法在新疆滑坡灾害易发性评价中的应用

doi: 10.13544/j.cnki.jeg.2023-0226
基金项目: 

新疆维吾尔自治区重点研发计划项目 2021B03004-3

详细信息
    作者简介:

    梁龙飞(1997-),男,硕士生,主要从事地质灾害与工程相互影响方面的研究. E-mail: TS22010127P31@cumt.edu.cn

  • 中图分类号: P642.22

APPLICATION OF SMALL-SAMPLE MACHINE LEARNING METHOD IN EVALUATION OF LANDSLIDE DISASTER SUSCEPTIBILITY IN XINJIANG

Funds: 

Key R&D Program of Xinjiang Uygur Autonomous Region 2021B03004-3

  • 摘要: 机器学习已经在滑坡易发性评价中大量应用且取得了较好的表现,但在进行大区域评价时,仍存在数据库样本需求量大,算力要求高;影响因素分级机械化,未考虑其与滑坡机理的相关性等情况。为减少数据库样本需求,本文提出了构建包含3种坡体状态的滑坡的数据库:已经发生过失稳的坡体、正处于失稳状态的坡体、失稳概率小的坡体,该数据库可以在数值上划分出临界值,便于模型更准确地识别滑坡,较大幅度地减小了数据量。针对影响因素分级机械化的问题,提出了基于频数分布图、累计曲线及其导数图的数理统计方式,更精细地描述因数与滑坡易发性的关系。以新疆滑坡灾害为例,验证了“包含3种坡体状态的数据库”与“基于数理统计的描述方法”的适用性,获得了新疆滑坡灾害易发性分区图。结果表明与传统数据库对比,在不明显改变精度的前提下,减少了90%以上的样本量;基于数理统计的描述方法可以绘制出更加细致的滑坡危险性分区图;活动性断裂和地形起伏度对新疆滑坡易发性起到重要的控制作用。
  • 图  1  新疆数字高程模型图

    Figure  1.  Digital elevation model map of Xinjiang

    图  2  滑坡与距断裂距离关系图

    Figure  2.  Relationship between landslide and distance to fault

    图  11  影响因素分级图

    Figure  11.  Hierarchical diagram of influencing factors

    图  3  滑坡与距河流距离关系图

    Figure  3.  Relationship between landslide and distance to river

    图  4  滑坡与水流强度指数关系图

    Figure  4.  The relationship between landslide and SPI

    图  5  滑坡与地形湿度指数关系图

    Figure  5.  Relationship between landslide and TWI

    图  6  滑坡与高程关系图

    Figure  6.  Relationship between landslide and DEM

    图  7  滑坡与起伏度关系图

    Figure  7.  Relationship between landslide and TR

    图  8  滑坡与坡度关系图

    Figure  8.  Relationship between landslide and slope angle

    图  9  滑坡与坡向关系图

    Figure  9.  Relationship between landslide and slope aspect

    图  10  滑坡与剖面曲率关系图

    Figure  10.  Relationship between landslide and PC

    图  12  模型指标对比图

    Figure  12.  Comparison chart of model indicators

    图  13  特征值重要性图

    Figure  13.  Eigenvalue importance map

    图  14  滑坡易发性分区图

    Figure  14.  Landslide susceptibility map

    表  1  基础数据来源表

    Table  1.   Basic data source table

    影响因素 数据来源
    数字高程 SRTM 30m
    坡度 SRTM 30m
    坡向 SRTM 30m
    剖面曲率 SRTM 30m
    水流强度指数 SRTM 30m
    起伏度 SRTM 30m
    距河流距离 新疆水文地质图
    地形湿度指数 SRTM 30m
    距断裂距离 中国及毗邻海区活动断裂分布图
    下载: 导出CSV

    表  2  支持向量机的超参数

    Table  2.   Hyperparameters of SVM

    种类 参数
    C 1
    kernel linear
    gamma 5
    下载: 导出CSV

    表  3  随机森林的超参数

    Table  3.   Hyperparameters of RF

    种类 参数
    n estimators 100
    max depth 5
    criterion Gini
    下载: 导出CSV

    表  4  多层感知机的超参数

    Table  4.   Hyperparameters of MLP

    种类 参数
    hidden layer sizes (10,10)
    activation Relu
    solver Adam
    max iter 1000
    下载: 导出CSV

    表  5  数据库对比表

    Table  5.   Database comparison table

    实验 张阳(2021) 孙德亮(2019) 周晓亭(2022) 本文
    研究区(×104km2) 0.2291 0.4087 0.2248 166
    滑坡 1251 1520 221 1204
    比值 0.546 0.372 0.090 0.001
    边坡状态 滑坡/非滑坡 滑坡/非滑坡 滑坡/非滑坡 滑坡/滑坡隐患点/非滑坡
    模型精度 89.94%~91.23% 83.00%~94.00% 50.55%~79.67% 57.00%~71.00%
    下载: 导出CSV
  • Arabameri A,Pradhan B,Rezaei K,et al. 2019. GIS-based landslide susceptibility mapping using numerical risk factor bivariate model and its ensemble with linear multivariate regression and boosted regression tree algorithms[J]. Journal of Mountain Science,16 (3): 595-618. doi: 10.1007/s11629-018-5168-y
    Arabameri A, Saha S, Roy J, et al. 2020. Landslide susceptibility evaluation and management using different machine learning methods in the Gallicash River Watershed, Iran[J]. Remote Sensing, 12(3): 475. doi: 10.3390/rs12030475
    Breiman L. 2001. Random forests[J]. Machine Learning, 45 : 5-32. doi: 10.1023/A:1010933404324
    Chen K, Zhu Y. 2007. A summary of machine learning and related algorithms[J]. Statistics & Information Forum, 22 (5): 105-112.
    Cheng H X, Lin Y J, Wang Y. 2023. Extraction and feature of precipitation clustering areas in Xinjiang based on GPM[J]. Journal of Arid Land Resources and Environment, 37 (3): 98-105.
    Das I, Stein A, Kerle N, et al. 2012. Landslide susceptibility mapping along road corridors in the Indian Himalayas using Bayesian logistic regression models[J]. Geomorphology, 179 (60): 116-125.
    Deng Q, Feng X, Zhang P, et al. 2000. Active tectonics of Tianshan Mountains[M]. Beijing: Seismological Press: 1-399.
    Deng Q D, Zhang P Z, Ran Y K, et al. 2003. Active tectonics and earthquake activities in China[J]. Earth Science Frontiers, 10 (S1): 66-73.
    Feng M. 2020. Risk assessment of landslide geological disasters in Daning County based on machine learning model[D]. Xi'an: Chang'an University.
    Guo Y L. 2022. Spatial-temporal evolution characteristics and influencing mechanism of lake evaporation in Xinjiang[D]. Lanzhou: Lanzhou University.
    Hu W Z. 1994. Arid environment, landslides and debris flow in Xinjiang and its prevention and controlling[J]. Geological Hazards and Environmental Protection, 5 (3): 1-7.
    He J. 2019. Spatial prediction and risk assessment of landslides based on machine learning[D]. Chengdu: University of Electronic Science and Technology of China.
    Kullback S, Rosenblatt H M. 1957. On the analysis of multiple regression in k categories[J]. Biometrika, 44(1-2): 67-83. doi: 10.1093/biomet/44.1-2.67
    Li R S, Ji W H, He S P, et al. 2011. The two tectonic domain division discussion between the ancient Asian and Tethys in western China[J]. Xinjiang Geology, 29 (3): 247-250.
    Liu L Y, Gao H Y. 2023. Landslide susceptibility assessment based on coupling of WOE model and Logistic regression model[J]. Journal of Engineering Geology, 31 (1): 165-175.
    Liu Y H, Fang R K, Su Y C, et al. 2021. Machine learning based model for warning of regional landslide disasters[J]. Journal of Engineering Geology, 29 (1): 116-124.
    Liu S Y, Yao X J, Guo W Q, et al. 2015. The contemporary glaciers in China based on the Second Chinese Glacier Inventory[J]. Acta Geographica Sinica, 70 (1): 3-16.
    Qiu H J. 2012. Study on the regional landslide characteristic analysis and hazard assessment: A case study of Ningqiang County[D]. Xi'an: College of Urban and Environment Northwest University.
    Roy J, Saha S, Arabameri A, et al. 2019. A novel ensemble approach for landslide susceptibility mapping(LSM)in Darjeeling and Kalimpong districts, West Bengal, India[J]. Remote Sensing, 11(23): 2866. doi: 10.3390/rs11232866
    Sevgen E, Kocaman S, Nefeslioglu H A, et al. 2019. A novel performance assessment approach using photogrammetric techniques for landslide susceptibility mapping with logistic regression, ANN and random forest[J]. Sensors, 19(18): 3940. doi: 10.3390/s19183940
    Shang M, Xiong D B, Zhang H Q, et al. 2022. Landslide displacement prediction model based on time series and mixed kernel function SA-SVR[J]. Journal of Engineering Geology, 30 (2): 575-588.
    Shen Y P, Su H C, Wang G Y, et al. 2013. The responses of glaciers and snow cover to climate change in Xinjiang(l): Hydrological effect[J]. Journal of Glaciology and Geocryology, 35 (3): 513-527.
    Shi W M, Zheng Y R, Tang B M, et al. 2003. Discussion on stability analysis method for landslides[J]. Rock and Soil Mechanics, 24 (4): 545-548, 552.
    Shu J, Xu Z B, Meng D Y. 2022. Small sample learning in big data era[EB/OL]. 2022-06-10. https://arxiv.org/abs/1808.04572.
    Su J S, Zhang B F, Xu X. 2006. Advances in machine learning based text categorization[J]. Journal of Software, 17 (9): 1848-1859. doi: 10.1360/jos171848
    Sun D L. 2019. Mapping landslide susceptibility based on machine learning and forecast warning of landslide induced by rainfall[D]. Shanghai: East China Normal University.
    Tan L X. 2008. Research on landslide hazard risk assessment based on support vector machine supported by GIS[D]. Changsha: Central South University.
    Tu H M, Liu Z D. 1990. Demonstrating on optimum statistic unit of relief amplitude in China[J]. Journal of Hubei University(Natural Science), 12 (3): 266-271.
    Vapnik V N. 1999. The nature of statistical learning theory[J]. New York: Springer-Verlag: 409.
    Wang S B, Zhuang J Q, Zheng J, et al. 2022. Landslide susceptibility evaluation based on Deep Learning along Kangding-Litang section of CZ Railway[J]. Journal of Engineering Geology, 30 (3): 908-919.
    Wang J J, Yin K L, Xiao L L. 2014. Landslide susceptibility assessment based on GIS and weighted information value: A case study of Wanzhou district, Three Gorges Reservoir[J]. Chinese Journal of Rock Mechanics and Engineering, 33 (4): 797-808.
    Zhou X T. 2022. Recognition and dynamic susceptibility assessment of landslides based on multi-source data[D]. Shanghai: East China University of Science and Technology.
    Xie P, Gu Y L, Zhang Y H, et al. 2017. Precipitation and drought characteristics in Xinjiang during 1961-2015[J]. Arid Land Geography, 40 (2): 332-339.
    Xu C, Dai F C, Yai X, et al. 2009. GIS-based landslide susceptibility assessment using analytical hierarchy process in Wenchuan earthquake region[J]. Chinese Journal of Rock Mechanics and Engineering, 28 (S2): 3978-3985.
    Xu C, Dai F C, Xu X W. 2010. Wenchuan earthquake-induced landslides: an overview[J]. Geological Review, 56 (6): 860-874.
    Xu Q. 2012. Theoretical studies on prediction of landslides using slope deformation process data[J]. Journal of Engineering Geology, 20 (2): 145-151.
    Zhan Y. 2021. Mapping landslide hazard risk using machine learning—Algorithm in Guixi, Jiangxi, Chin[D]. Nanchang: East China University of Technology.
    Yao X J, Liu S Y, Guo W Q, et al. 2012. Glacier changes in Altay Mountains in China from 1960 to 2009—Based on the second glacier inventory of China[J]. Journal of Natural Resources, 27 (10): 1734-1745.
    Zhang M S, Li T L. 2011. Triggering factors and forming mechanism of loess landslide[J]. Journal of Engineering Geology, 19 (4): 530-540.
    Zhao K L, Jin X L, Wang Y Z. 2021. Survey on few-shot learning[J]. Journal of Software, 32 (2): 349-369.
    陈凯, 朱钰. 2007. 机器学习及其相关算法综述[J]. 统计与信息论坛, 22 (5): 105-112. https://www.cnki.com.cn/Article/CJFDTOTAL-TJLT200705022.htm
    程红霞, 林粤江, 王勇. 2023. 基于GPM的新疆降水聚集区域提取及特征分析[J]. 干旱区资源与环境, 37 (3): 98-105. https://www.cnki.com.cn/Article/CJFDTOTAL-GHZH202303014.htm
    邓起东, 冯先岳, 张培震, 等. 2000. 天山活动构造[M]. 北京: 地震出版社: 1-399.
    邓起东, 张培震, 冉勇康, 等. 2003. 中国活动构造与地震活动[J]. 地学前缘, 10 (S1): 66-73. https://www.cnki.com.cn/Article/CJFDTOTAL-DXQY2003S1011.htm
    冯满. 2020. 基于机器学习模型的大宁县滑坡灾害危险性评价[D]. 西安: 长安大学.
    郭燕玲. 2022. 新疆湖泊蒸发的时空演变特征和影响机制研究[D]. 兰州: 兰州大学.
    何静. 2019. 基于机器学习的滑坡灾害空间预测及风险评估[D]. 成都: 电子科技大学.
    胡卫忠. 1994. 新疆的干旱环境与滑坡、泥石流及其防治对策[J]. 地质灾害与环境保护, 5 (3): 1-7. https://www.cnki.com.cn/Article/CJFDTOTAL-DZHB199403000.htm
    李荣社, 计文化, 何世平, 等. 2011. 中国西部古亚洲与特提斯两大构造域划分问题讨论[J]. 新疆地质, 29 (3): 247-250. https://www.cnki.com.cn/Article/CJFDTOTAL-XJDI201103002.htm
    刘璐瑶, 高惠瑛. 2023. 基于证据权与Logistic回归模型耦合的滑坡易发性评价[J]. 工程地质学报, 31 (1): 165-175. doi: 10.13544/j.cnki.jeg.2020-482
    刘时银, 姚晓军, 郭万钦, 等. 2015. 基于第二次冰川编目的中国冰川现状[J]. 地理学报, 70 (1): 3-16. https://www.cnki.com.cn/Article/CJFDTOTAL-DLXB201501002.htm
    刘艳辉, 方然可, 苏永超, 等. 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. doi: 10.13544/j.cnki.jeg.2020-533
    邱海军. 2012. 区域滑坡崩塌地质灾害特征分析及其易发性和危险性评价研究[D]. 西安: 西北大学.
    尚敏, 熊德兵, 张惠强, 等. 2021. 基于时间序列与混合核函数SA-SVR的滑坡位移预测模型研究[J]. 工程地质学报, 30 (2): 575-588. doi: 10.13544/j.cnki.jeg.2021-0584
    沈永平, 苏宏超, 王国亚, 等. 2013. 新疆冰川、积雪对气候变化的响应(Ⅰ): 水文效应[J]. 冰川冻土, 35 (3): 513-527. https://www.cnki.com.cn/Article/CJFDTOTAL-BCDT201306001.htm
    时卫民, 郑颖人, 唐伯明. 2003. 滑坡稳定性评价方法的探讨[J]. 岩土力学, 24 (4): 545-548, 552. https://www.cnki.com.cn/Article/CJFDTOTAL-YTLX200304014.htm
    苏金树, 张博锋, 徐昕. 2006. 基于机器学习的文本分类技术研究进展[J]. 软件学报, 17 (9): 1848-1859. https://www.cnki.com.cn/Article/CJFDTOTAL-RJXB200609005.htm
    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]. 上海: 华东师范大学.
    谭立霞. 2008. GIS支持下基于支持向量机的滑坡灾害危险性评价研究[D]. 长沙: 中南大学.
    涂汉明, 刘振东. 1990. 中国地势起伏度最佳统计单元的求证[J]. 湖北大学学报(自然科学版), 12 (3): 266-271. https://www.cnki.com.cn/Article/CJFDTOTAL-HDZK199003016.htm
    王佳佳, 殷坤龙, 肖莉丽. 2014. 基于GIS和信息量的滑坡灾害易发性评价——以三峡库区万州区为例[J]. 岩石力学与工程学报, 33 (4): 797-808. https://www.cnki.com.cn/Article/CJFDTOTAL-YSLX201404018.htm
    王世宝, 庄建琦, 郑佳, 等. 2021. 基于深度学习的CZ铁路康定-理塘段滑坡易发性评价[J]. 工程地质学报, 30 (3): 908-919. doi: 10.13544/j.cnki.jeg.2021-0115
    谢培, 顾艳玲, 张玉虎, 等. 2017.1961~2015年新疆降水及干旱特征分析[J]. 干旱区地理, 40 (2): 332-339. https://www.cnki.com.cn/Article/CJFDTOTAL-GHDL201702012.htm
    许冲, 戴福初, 徐锡伟. 2010. 汶川地震滑坡灾害研究综述[J]. 地质论评, 56 (6): 860-874. https://www.cnki.com.cn/Article/CJFDTOTAL-DZLP201006014.htm
    许冲, 戴福初, 姚鑫, 等. 2009. GIS支持下基于层次分析法的汶川地震区滑坡易发性评价[J]. 岩石力学与工程学报, 28 (S2): 3978-3985. https://www.cnki.com.cn/Article/CJFDTOTAL-YSLX2009S2104.htm
    许强. 2012. 滑坡的变形破坏行为与内在机理[J]. 工程地质学报, 20 (2): 145-151. http://www.gcdz.org/article/id/11113
    姚晓军, 刘时银, 郭万钦, 等. 2012. 近50a来中国阿尔泰山冰川变化——基于中国第二次冰川编目成果[J]. 自然资源学报, 27 (10): 1734-1745. https://www.cnki.com.cn/Article/CJFDTOTAL-ZRZX201210010.htm
    张茂省, 李同录. 2011. 黄土滑坡诱发因素及其形成机理研究[J]. 工程地质学报, 19 (4): 530-540. http://www.gcdz.org/article/id/10051
    张阳. 2021. 机器学习方法在滑坡灾害易发性区划中的应用[D]. 南昌; 东华理工大学.
    赵凯琳, 靳小龙, 王元卓. 2021. 小样本学习研究综述[J]. 软件学报, 32 (2): 349-369. https://www.cnki.com.cn/Article/CJFDTOTAL-RJXB202102005.htm
    周晓亭. 2022. 基于多源数据的滑坡识别及其易发性动态评价[D]. 上海: 东华理工大学.
  • 加载中
图(14) / 表(5)
计量
  • 文章访问数:  129
  • HTML全文浏览量:  22
  • PDF下载量:  52
  • 被引次数: 0
出版历程
  • 收稿日期:  2023-05-30
  • 修回日期:  2023-07-18
  • 刊出日期:  2023-08-25

目录

    /

    返回文章
    返回