首页|基于SBAS-InSAR技术和BP神经网络的高位远程滑坡危险性分析研究

基于SBAS-InSAR技术和BP神经网络的高位远程滑坡危险性分析研究

扫码查看
原文链接
  • NETL
  • NSTL
  • 万方数据
高位远程滑坡具有突发性强、致灾范围广等特点,严重威胁人民生命和财产安全.针对滑坡危险性评价过程中评价单元划分缺乏对滑坡形状考虑的问题,利用SBAS-InSAR和BP神经网络模型对云南头寨滑坡进行危险性评价研究.首先,利用2018 年4 月—2019 年4 月的299 幅Sentinel-1A升降轨影像获取研究区时间序列地表形变信息,然后,分别利用斜坡单元和网格单元为评价单元进行对比分析,对初步提取的坡度、坡向、高程等8 个评价因子进行皮尔逊相关性分析和多重共线性分析,完善评价因子体系.最后,利用BP神经网络模型对滑坡危险性等级划分.结果显示,考虑滑坡形状的斜坡单元评价效果优于网格单元评价结果,斜坡单元更能反映单体滑坡危险区域的空间分布,且模型AUC值为0.724,表明该模型可适用于单体滑坡.其中极高危险区和高危险区主要分布在滑坡后壁和滑坡侧壁部分区域,极低风险区和低风险区基本处于滑坡面和坡脚区域.本研究结果可为单体高位远程滑坡灾害前期预警和防灾减灾工作提供技术支撑.
Research on High-Altitude and Long-Distance Landslide Risk Analysis Based on the Combination of SBAS-InSAR Technology and BP Neural Network
High-position and long-distance landslides are characterized by strong suddeness and extensive dis-aster-causing areas,posing a significant threat to the lives and property of individuals.In response to the issue of the lack of consideration for landslide shapes in the division of evaluation units during landslide hazard assess-ment,this paper studies the hazard assessment of the Touzhai landslide in Yunnan by combining SBAS-InSAR and BP neural network models.Firstly,299 Sentinel-1A ascending and descending orbit images from April 2018 to April 2019 were used to obtain time-series surface deformation information in the study area.Then,slope u-nits and grid units were used as evaluation units for comparative analysis.Pearson correlation analysis and multi-collinearity analysis were performed on the initially extracted nine evaluation factors such as slope,aspect,and el-evation to improve the evaluation factor system.Finally,the BP neural network model was used to classify the landslide hazard level.The results show that the evaluation effect of slope units considering the shape of the land-slide is superior to the evaluation results of grid units.Slope units can better reflect the spatial distribution of dan-gerous areas of ndividual landslides,and the AUC value of the model is 0.724,indicating that the model can be iapplied to individual landslides.Among them,the extremely high-risk and high-risk areas are mainly distribu-ted in the rear wall and partial side wall of the landslide,while the extremely low-risk and low-risk areas are basically located in the landslide surface and toe areas.The findings of this research can offer technical assistance for enhancing early warning systems and disaster prevention and reduction measures pertaining to individual high-risk and long-distance landslide hazards.

high-altitude and long-distance landslideSBAS-InSAR technologyBP neural networkevalu-ation unit divisionlandslide risk assessment

喜文飞、成鑫、杨志全、刘美杉、郭峻杞

展开 >

昆明理工大学 公共安全与应急管理学院,云南 昆明 650093

云南师范大学 地理学部,云南 昆明 650500

昆明理工大学 应急管理部地质灾害风险防控与应急减灾重点实验室,云南 昆明 650093

云南省高校高烈度地震山区交通走廊工程地质病害早期快速判识与防控重点实验室,云南 昆明 650093

展开 >

高位远程滑坡 SBAS-InSAR技术 BP神经网络 评价单元划分 滑坡危险性评价

国家自然科学基金项目云南省自然科学基金杰青项目云南省重点研发计划项目云南省重大科技专项

42361144880202301AV070066202003AC10002202202AD080010

2024

10.16112/j.cnki.53-1223/n.2024.03.521

昆明理工大学学报(自然科学版)
昆明理工大学

昆明理工大学学报(自然科学版)

CSTPCD北大核心
影响因子：0.516
ISSN：1007-855X
年,卷(期)：2024.49(3)
  • 7