1.School of Traffic and Transportation, Lanzhou Jiaotong University, Lanzhou Gansu 730070, China
2.Zhejiang Yiwu Port Co. , Ltd. , Yiwu Zhejiang 730070, China
3.Railway Industry Key Laboratory of Intelligent Control of Plateau Railway Transportation, Lanzhou Gansu 730070, China
Show less
文章历史+
Received
Published
2024-10-15
2025-07-07
Issue Date
2026-07-13
PDF (8016K)
摘要
为降低滑坡风险,绘制易受滑坡影响地区的地图至关重要。以兰渝铁路宕昌段为研究对象,采用SBAS-InSAR技术结合升降轨雷达数据,提取2021年11月至2022年10月的地表形变信息并分解为垂直向和东西向形变。利用支持向量机(SVM)模型对研究区域进行滑坡危险性分级,并结合垂直向形变速率进行铁路沿线滑坡动态危险性评价。结果表明:选取高程、降雨量等16个指标作为滑坡影响因子,经皮尔逊相关系数和信息增益比筛,选出13个对滑坡影响最大的指标建立SVM模型,将研究区域分为高、较高、中、较低、低危险性5类,经ROC曲线检验模型评价结果精度为0.901 9;兰渝铁路沿线1 km区域升轨地表形变速率为-33.11~32.33 mm · a-1,降轨地表形变速率为-50.67~48.98 mm · a-1,运用准三维形变分解式计算得出垂直向地表形变速率为-64.59~89.92 mm · a-1,东西向为-117.29~90.88 mm · a-1;与SVM模型评价结果相比,结合SBAS-InSAR变形速率滑坡动态危险性评价结果中低、较低、中危险区面积占比分别减少2.53%,3.68%和2.24%,较高、高危险区分别增加5.69%和2.76%,使危险性分区结果更可靠。
Abstract
To reduce landslide risk, it is crucial to map areas that are susceptible to landslides. Focusing on the Dangchang section of Lanzhou-Chongqing Railway, this study employs SBAS-InSAR technology with ascending/descending track radar data to extract surface deformation information from November 2021 to October 2022, decomposing it into vertical and east-west components. A Support Vector Machine (SVM) model classifies landslide hazard levels, integrated with vertical deformation rate to assess the dynamic landslide hazard along the railway. The results indicate that 16 indicators, including elevation and rainfall, are selected as factors affecting landslides. After screening through Pearson correlation coefficients and information gain ratios, 13 of the most significant indicators are chosen to establish the SVM model, categorizing the study area into five hazard levels: high, relatively high, medium, relatively low, and low. The model evaluation result, verified by the ROC curve, shows an accuracy of 0.9019. Ascending track deformation rate within 1 km railway corridor along the Lanzhou-Chongqing Railway ranges from -33.11 to 32.33 mm · a-1, while the rate for the descending track varies from -50.67 to 48.98 mm · a-1. A quasi-3D deformation decomposition yields the vertical surface deformation rate between -64.59 and 89.92 mm · a-1, and the east-west rate -117.29 to 90.88 mm · a-1. Compared to SVM results, dynamic evaluation shows that medium, relatively low, and low hazard zones decrease by 2.53%, 3.68%, and 2.24%, respectively; while high and relatively high hazard zones increase by 5.69% and 2.76%, respectively, enhancing hazard zoning reliability.
在滑坡监测研究中,吴绿川等[7]融合InSAR技术与光学遥感技术对贵州多地形变开展监测,通过设定形变速率阈值识别高危区域,并结合归一化植被指数(Normalized Difference Vegetation Index,NDVI)及滑坡形态参数进行灾害风险综合评估。廖明生等[8]运用D-InSAR技术精准定位三峡库区滑坡的位置,分析区域形变特征。王桂杰等[9]系统阐述D-InSAR技术原理,提出永久散射体(Persistent Scatterer,PS)方法和短基线集(Small Baseline Subset,SBAS)方法,有效克服传统InSAR易受时空相干性干扰的技术瓶颈。雷玲等[10]利用PS-InSAR技术监测伯克利山滑坡,测得最大形变速率达每年30 mm · a-1。许广河等[11]通过PS-InSAR分析西吉县滑坡,结合历史资料和实地验证,揭示了滑坡活动与降水及老滑坡复活的关联性。张雄伟[12]成功利用D-InSAR与PS-InSAR技术监测公路地质灾害形变,并分析其形变特征。褚洪义等[13]采用基于Sentinel-1A的PS-InSAR技术发现喀喇昆仑公路历史滑坡体仍存在持续变形。Soltanieh等[14]通过PS-InSAR监测汤普森河谷滑坡,发现其最大位移速率为每年29~52 mm · a-1。上述研究表明InSAR技术在滑坡识别中具有显著优势,但受限于单一轨道数据源,其监测可靠性仍需通过多源数据协同验证。
滑坡危险性评价是评估特定区域内地质灾害发生的空间与时间概率。许冲等[15]基于汶川滑坡数据库,用层次分析法确定地层岩性、断裂等因子的权重,结合地理信息系统(Geographic Information System,GIS)平台绘制了汶川地震滑坡危险性图。孟晓捷等[16]采用类似方法评价天水麦积区的滑坡风险。唐凤娇等[18]联合频率比法和信息量法评估金沙江溪洛渡库区滑坡危险性。刘筱怡[18]采用信息量法和逻辑回归模型,选取多项指标评价大渡河流域古滑坡,发现逻辑回归模型更优,并结合InSAR技术对滑坡危险性进行了动态评价。熊小辉等[19]比较了多种模型与逻辑回归模型的耦合效果,发现证据权-逻辑回归模型在四川普格县评价滑坡危险性中精度最高。姚林强[20]采用确定性系数和逻辑回归模型评价兰州地质灾害风险,并结合SBAS-InSAR数据提高滑坡危险性结果的精准度。唐维厘[21]应用信息熵模型筛选因子,采用逻辑回归模型评价黄土区域滑坡危险性。杨创奇等[22]建立了耦合熵指数-逻辑回归树模型并应用于吴起县滑坡评价。然而,这些模型的受试者工作特征曲线(Receiver Operating Characteristic,ROC)下方面积(Area Under Curve,AUC)多在0.7~0.9间,其评估滑坡危险性精度有待提升。
本文采用SBAS-InSAR技术获取研究区域高精度时序形变信息,于升降轨数据形变分解计算滑坡体垂直与水平向形变分量,揭示滑坡三维运动特征。提取潜在滑坡影响因子,并通过信息增益比(Information Gain Ratio,IGR)量化各因子与形变数据的关联强度,筛选主导驱动因子。基于筛选后的因子集,构建支持向量机(Support Vector Machine,SVM)模型,实现滑坡形变-环境因子关系的非线性建模。在此基础上,分别输出基于SVM的滑坡危险性模型评价结果与基于SBAS-InSAR的滑坡动态危险性分析结果,并通过不同危险性评价结果对比验证多方法融合在滑坡动态危险性评估中的互补性与可靠性,为滑坡时空演化预测与风险分级防控提供一体化技术支撑。
针对滑坡动态危险性研究,首先通过SBAS-InSAR技术提取滑坡区域高精度时序形变信息,利用多景SAR影像构建干涉对网络并抑制时空失相关噪声,获取滑坡体长期形变时间序列与累计形变量。随后基于升降轨数据形变分解,结合升轨与降轨卫星的视线方向(Line of Sight,LOS)向形变观测值及轨道几何参数,通过加权最小二乘反演解算滑坡体垂直与水平向形变分量,揭示滑坡三维运动特征。开展滑坡影响因子相关性分析,从地形地貌(坡度和坡向)、地质构造(岩性和断层)、水文气象(降雨和地下水位)等维度筛选潜在影响因子,并采用信息增益比量化各因子与形变数据的关联强度,确定主导驱动因子。基于筛选后的因子集,构建支持向量机(SVM)模型,利用非线性核函数映射高维特征空间,通过模型参数优化确定最佳惩罚系数C与核参数Gamma,实现滑坡形变-环境因子关系的非线性建模。在此基础上,分别输出基于支持向量机的滑坡危险性模型评价结果(ROC曲线等)与基于SBAS-InSAR技术的滑坡动态危险性分析结果(根据形变速率阈值划分危险等级)。最终通过不同危险性评价结果对比,验证多方法融合在滑坡动态危险性评估中的互补性与可靠性。
最后,进行形变速率反演。第1次反演估算形变速率和残余地形,优化干涉图结果。第2次反演计算时间序列上的位移,从而得到精确的形变速率如图10所示。由图10可知:兰渝铁路沿线1 km区域升轨地表形变速率主要集中在-33.11~32.33 mm · a-1之间,d区有零星区域形变速率超过了-33.11 mm · a-1;其降轨地表形变速率主要分布在-50.67~48.98 mm · a-1区间,b区、c区部分区域地表形变速率超过了48.98 mm · a-1,d区部分区域地表形变速率超过了-50.67 mm · a-1。
图11中,垂直形变取正表示沉降、取负表示抬升,东西形变取正表示向东、取负表示向西。由图11可知:垂直向地表形变速率主要在-64.59~89.92 mm · a-1之间,c区和d区部分区域表现为地表沉降,b区和c区部分区域表现为地表抬升,其余部分地表形变稳定,东西向地表形变速率主要在-117.29~90.88 mm · a-1之间,a区、b区和c区部分区域地表水平向东运动,d区零星区域表现为水平向西运动。
然后,将3.1节获得的地表垂向形变速率进行等级划分,共分为5个等级:低危险区(形变速率绝对值小于5 mm · a-1)、较低危险区(形变速率绝对值在5~10 mm · a-1范围)、中危险区(形变速率绝对值在10~15 mm · a-1范围)、较高危险区(形变速率绝对值在15~20 mm · a-1范围)、高危险区(形变速率绝对值大于20 mm · a-1)。
(2)利用SBAS-InSAR技术对研究区域进行形变信息提取,兰渝铁路沿线1 km区域升轨地表形变速率主要集中在-33.11~32.33 mm · a-1之间,降轨地表形变速率主要分布在-50.67~48.98 mm · a-1区间。经过准三维形变分解式计算得出兰渝铁路沿线垂直向地表形变速率主要在-64.59~89.92 mm · a-1之间,东西向地表形变速率主要在-117.29~90.88 mm · a-1之间。
ZHUJianjun, LIZhiwei, HUJun. Research Progress and Methods of InSAR for Deformation Monitoring [J]. Acta Geodaetica et Cartographica Sinica, 2017, 46 (10): 1717-1733. in Chinese
DONGJianjun, MEIYuan, YANBin, et al. D-InSAR Monitoring of Slope Safety and Stability in High-Altitude Dump Site [J]. Journal of Safety and Environment, 2022, 22 (5): 2514-2523. in Chinese
[7]
朱强.基于InSAR技术的滑坡隐患识别及形变监测研究[D].北京:中国地质大学,2021.
[8]
ZHUQiang. Research on Landslide Hazard Identification and Deformation Monitoring Based on InSAR Technology [D]. Beijing: China University of Geosciences, 2021. in Chinese
WANGChuan, TUKuan, CHENHua, et al. Application of Stacking InSAR and IPTA Technology to Detect Landslide Hazards in Longde County, Ningxia [J]. Journal of Natural Disasters, 2022, 31 (5): 222-234. in Chinese
PENGYi, WUYunlong, YANGYu, et al. Research on Small Baseline InSAR Technology—Taking Maoxian County in Sichuan Province as an Example [J]. Urban Geotechnical Investigation & Surveying, 2022 (6): 102-107. in Chinese
WULüchuan, WANGJianhui, FUYan. Early Identifying and Monitoring Landslides in Guizhou Province with InSAR and Optical Remote Sensing [J]. Bulletin of Surveying and Mapping, 2021 (7): 98-102. in Chinese
LIAOMingsheng, TANGJing, WANGTeng, et al. Application of High-Resolution SAR Data in Landslide Monitoring in the Three Gorges Reservoir Area [J]. Scientia Sinica: Terrae, 2012, 42 (2): 217-229. in Chinese
WANGGuijie, XIEMowen, QIUCheng, et al. Application of D-INSAR Technology in Large-Scale Landslide Monitoring [J]. Rock and Soil Mechanics, 2010, 31 (4): 1337-1344. in Chinese
LEILing, ZHOUYinqing, LIJingwen, et al. Application of PS-InSAR to Monitoring Berkeley Landslide [J]. Journal of Beijing University of Aeronautics and Astronautics, 2012, 38 (9): 1224-1226. in Chinese
XUGuanghe, HEXiaofeng, WANGGuorui, et al. InSAR Identification and Cause Analysis of Landslide Disaster in Ningxia (Xiji County) Section of the Yellow River Basin [J]. Bulletin of Surveying and Mapping, 2022 (6): 104-107. in Chinese
ZHANGXiongwei. Research on Monitoring and Early Warning of Geological Disasters in High-Altitude and Rainy Mountainous Areas Based on InSAR Technology [D]. Xi'an: Chang'an University, 2021. in Chinese
CHUHongyi, HUANGZhaohuan, YAOLinqiang, et al. Research on the Characteristics of Surface Deformation along the Karakorum Highway Based on PS-InSAR Technology [J]. Journal of Lanzhou University (Natural Sciences), 2021, 57 (5): 608-614. in Chinese
[27]
SOLTANIEHA., MACCIOTTAR. Updated Understanding of the Ripley Landslide Kinematics Using Satellite InSAR [J]. Geosciences, 2022, 12 (8): 298.
XUChong, DAIFuchu, YAOXin, et al. Evaluation of Landslide Susceptibility in Wenchuan Earthquake Area Based on Analytic Hierarchy Process Supported by GIS [J].Chinese Journal of Rock Mechanics and Engineering, 2009, 28 (): 3978-3985. in Chinese
MENGXiaojie, ZHANGXinshe, ZENGQingming, et al. The Susceptibility Evaluation of Loess Landslide Based on Weighted Information Value Method—Taking 1∶50,000 Map of Maiji District of Tianshui City as an Example [J]. Northwestern Geology, 2022, 55 (2): 249-259. in Chinese
TANGFengjiao, QIShengwen, GUOSongfeng, et al. Study on Spatiotemporal Distribution and Susceptibility of Reservoir-Induced Landslides in Xiluodu Reservoir Area of Jinsha River [J]. Journal of Engineering Geology, 2022, 30 (3): 609-620. in Chinese
[34]
刘筱怡.基于多元遥感技术的古滑坡识别与危险性评价研究[D].北京:中国地质科学院,2020.
[35]
LIUXiaoyi. Identification and Hazard Assessment of Ancient Landslides Based on Multi-Source Remote Sensing Technology [D]. Beijing: Chinese Academy of Geological Sciences, 2020. in Chinese
XIONGXiaohui, WANGChanglin, BAIYongjian, et al. Comparison of Landslide Susceptibility Assessment Based on Multiple Hybrid Models at County Level: a Case Study for Puge County, Sichuan Province [J]. Chinese Journal of Geological Hazard and Control, 2022, 33 (4): 114-124. in Chinese
YAOLinqiang. Analysis of Surface Deformation Characteristics and Susceptibility Evaluation of Geological Disasters in Lanzhou Area Based on SBAS-InSAR Technology [D]. Lanzhou: Lanzhou University, 2021. in Chinese
[40]
唐维厘.基于SBAS-InSAR的黄土滑坡监测和易发性评价[D].北京:中国地质大学,2021.
[41]
TANGWeili. Loess Landslide Monitoring and Susceptibility Evaluation Based on SBAS-InSAR [D]. Beijing: China University of Geosciences, 2021. in Chinese
YANGChuangqi, TAOPan, YANGZheng. Landslide Susceptibility Zoning Based on Logistic Regression Tree Coupled Entropy Index Model: Case Study of Landslide in Wuqi County, Yan'an City, Shaanxi Province [J]. People's Yangtze River, 2022, 53 (5): 128-134. in Chinese
[44]
BERARDINOP, FORNAROG, LANARIR, et al. A New Algorithm for Surface Deformation Monitoring Based on Small Baseline Differential SAR Interferograms [J]. IEEE Transactions on Geoscience and Remote Sensing, 2002, 40 (11): 2375-2383.
[45]
施友丽.基于时序InSAR库区滑坡监测识别与易发性评价研究[D].昆明:昆明理工大学,2022.
[46]
SHIYouli. Research on Landslide Monitoring, Identification and Susceptibility Evaluation in Reservoir Area Based on Time Series InSAR [D]. Kunming: Kunming University of Science and Technology, 2022. in Chinese
[47]
LIB Q, JIANGW L, LIY S, et al. Monitoring and Analysis of Woda Landslide (China) Using InSAR and Sentinel-1 Data [J]. Advances in Space Research, 2023, 72 (5): 1789-1802.
ZHUKuanxing. Early Identification and Surface Deformation of Potential Landslides from Diexi to Feihong in the Upper Reaches of Minjiang River [D]. Jilin: Jilin University, 2022. in Chinese
YUHaihua. Research on Surface Deformation Monitoring Method of Jiangding Cliff Slope Based on Time-Series InSAR Technology [D]. Chengdu: Southwest Jiaotong University, 2022. in Chinese
ZHAOBeibei, HUANGHaifeng, DENGYonghuang, et al. InSAR Monitoring and Analysis of Fanjiaping Landslide in the Three Gorges Reservoir area Based on Sentinel-1A [J]. Renmin Yangtze River, 2022, 53 (10): 103-107. in Chinese
[54]
HUANGW B, DINGM T,LI, Z H,et al. An Efficient User-Friendly Integration Tool for Landslide Susceptibility Mapping Based on Support Vector Machines: SVM-LSM Toolbox [J]. Remote Sensing, 2022, 14 (14): 3408.
LIUJie, WUZhen. GIS-Based Landslide Risk Assessment of Zhouqu-Wudu Section of Bailongjiang Basin [J]. Journal of Earthquake Engineering, 2020, 42 (6): 1723-1734. in Chinese