【作者】 孙海涛；

【导师】 吕岩； 林森；

【作者基本信息】 吉林大学 ， 工程硕士（专业学位）， 2024， 硕士

【摘要】 集安为吉林省县级市,由通化地级市代管,位于吉林省东南部山区,处在老岭山脉与长白山尾端。集安市内地质环境总体较好,但局部存在着自然因素与不科学的人类工程建设、山地开发引起的滑坡、泥石流与崩塌等地质灾害,尤其以崩塌灾害居多。研究区内数量众多的崩塌灾害对当地的生态环境造成了恶劣影响,更加威胁着人民的生命财产安全,也制约了当地的经济发展。因此以集安市为研究区,开展集安市崩塌灾害敏感性评价,希望可以为当地崩塌灾害预防与治理提供一定的参考。以集安市崩塌灾害为背景,调查得到了研究区内崩塌灾害点和区域地质条件的基本资料,分析其规模、发育类型、破坏形式以及分布情况,发现了灾害点空间分布的分形特征和空间聚集特征。选取能够反映研究区崩塌灾害产生的静态控制因素和动态诱发因素,采用频率比模型、确定性系数模型和随机森林模型对集安市崩塌灾害敏感性进行评价并得出了敏感性分区图。讨论了空间尺度改变与因子的子区间划分数量发生变化时,灾害点的空间聚集性与它们产生的双重作用,采用上述3种模型探究了不同空间尺度和子区间划分数量下AUC值的波动,以此分析了空间尺度和子区间划分对不同模型产生的影响,并选取了集安市崩塌灾害敏感性评价的最优组合工况。最后采用空间自相关分析,发现集安市崩塌灾害敏感性表现出显著的空间正相关性。主要研究内容和成果如下:1、查阅了集安市政府发布的地质灾害专题文件,结合对集安市崩塌灾害进行的现场调查,整理了集安区域内崩塌灾害和地质环境、人类活动等基本资料,阐述了研究区灾害的规模、发育类型、破坏形式以及分布情况,分析发现集安市崩塌灾害空间分布具有分形和空间聚集特征,为进一步的灾害敏感性评价打下基础。2、选取能够反映集安市崩塌灾害发育产生的静态控制因素(坡度、坡向、高程、水系密度、地形湿度指数、归一化植被指数、断裂密度、工程岩组)和动态诱发条件(道路密度、土地覆盖类型),共计10个评价因子,按自然断点法将各个连续型因子进行重分类,以各子区间的灾害分布密度为标准得到了各评价因子的灾害易发范围。3、采用频率比模型、确定性系数模型和随机森林模型对集安市崩塌灾害的敏感性进行评价,以受试者工作特征曲线及曲线下面积衡量和比较各模型在研究区的表现效果,并生成了基于3个模型的集安市崩塌灾害敏感性分区图。4、考虑到空间尺度对地质灾害敏感性评价的影响,使用频率比模型、确定性系数模型和随机森林模型在3个尺度上进行崩塌灾害的敏感性建模,为避免因子的子区间划分数量不同带来的偶然性,对连续型评价因子采用自然断点4、5、6类的划分方式。在27个组合工况下进行敏感性评价,并分析了空间尺度和子区间划分数量对不同模型带来的影响。利用ROC曲线选出预测效果最佳的组合,并对该组合得到的敏感性指数图进行了空间自相关分析,发现集安崩塌灾害敏感性呈现显著的空间正相关格局。更多还原

【Abstract】 Ji’an,a county-level city within Jilin Province managed by Tonghua,is located in the mountainous southeastern part of the province,situated between the Laoling Mountains and the tail end of the Changbai Mountains.While the geological environment within Ji’an City is generally favorable,localized areas are susceptible to geological disasters such as landslides,debris flows,and collapses,primarily due to natural factors coupled with unscientific human engineering constructions and mountainous area development.Among these,collapses are particularly prevalent.The numerous collapse disasters within the research area have adversely affected the local ecological environment,posed threats to the safety of people’s lives and property,and have impeded local economic development.Therefore,conducting a susceptibility assessment of collapse disasters in Ji’an City as the research area aims to provide references for the prevention and management of local collapse disasters.Against the backdrop of collapse disasters in Ji’an City,an investigation was carried out to gather basic information on collapse disaster sites and regional geological conditions within the study area.An analysis of their scale,development types,damage forms,and distribution revealed the fractal characteristics and spatial clustering of disaster sites.Selecting static control factors and dynamic inducing factors that reflect the occurrence of collapse disasters in the research area,the Frequency Ratio model,Deterministic Coefficient model,and Random Forest model were employed for the susceptibility assessment of collapse disasters in Ji’an City,resulting in the generation of susceptibility zoning maps.The effects of changes in spatial scale and the number of subdivisions of factors on the spatial clustering of disaster sites and their dual influence were discussed.Using the three aforementioned models,the effectiveness of collapse disaster sensitivity evaluation under different spatial scales and numbers of subdivisions was explored,revealing the impact of spatial scale and subdivisions on the evaluation effectiveness of different models and selecting the optimal combination of conditions for the susceptibility assessment of collapse disasters in Ji’an City.Finally,spatial autocorrelation analysis was used to discover that the sensitivity of collapse disasters in Ji’an City exhibits significant spatial positive correlation.The main research contents and results are as follows:1.By consulting the geological disaster thematic documents issued by the Ji’an City government and combining them with on-site investigations of collapse disasters,basic information on collapse disasters,geological environments,and human activities within the Ji’an region was organized.The research elucidates the scale,development types,damage forms,and distribution of disasters in the study area,and analyses found the spatial distribution of collapse disasters in Ji’an City to exhibit fractal and spatial clustering characteristics,laying a foundation for further susceptibility assessment of disasters.2.Ten evaluation factors were selected to reflect the static control factors(slope,aspect,elevation,stream density,topographic wetness index,normalized difference vegetation index,fault density,engineering rock groups)and dynamic inducing conditions(road density,land cover types)that contribute to the development of collapse disasters in Ji’an City.Using the natural breakpoint method,these continuous factors were reclassified,and the disaster susceptibility range of each evaluation factor was determined based on the disaster distribution density of each subdivision.3.Used frequency ratio model,certainty factor model,and random forest model to evaluate the susceptibility of collapse disasters in Ji’an City.It measured and compared the performance of each model in the study area with the ROC and the AUC,and generated susceptibility zoning maps of collapse disasters in Ji’an City.4.Considering the impact of spatial scale on the evaluation of geological disaster susceptibility,collapse disaster susceptibility modeling was performed at three scales using the Frequency Ratio model,Deterministic Coefficient model,and Random Forest model.To avoid randomness caused by varying numbers of subdivisions of factors,a classification scheme of 4,5,and 6 categories was used for continuous evaluation factors based on natural breakpoints.Susceptibility assessment was conducted under 27 combination conditions,analyzing the impact of spatial scale and number of subdivisions on different models.The combination with the best predictive effect was selected using the ROC curve,and spatial autocorrelation analysis was performed on the susceptibility index map obtained from this combination,finding a significant spatial positive correlation pattern in the sensitivity of collapse disasters in Ji’an.更多还原