基于逻辑回归与GIS的青海省尖扎县区域滑坡危险性评价

发布时间：2018-03-20 20:37

本文选题：尖扎县　切入点：滑坡　出处：《中国地质大学(北京)》2017年硕士论文　论文类型：学位论文

【摘要】：本论文在进行青海省尖扎县滑坡地质灾害调查的基础上,对尖扎县典型滑坡拉萨滑坡和康杨滑坡进行力学实验和矿物学实验,分析滑坡成因机理和内在微观矿物组成,发现红色泥岩地层对滑坡发育具有控制性作用。泥岩地层中的导水通道从内部湿润泥岩,红色泥岩层遇水后力学强度骤减,是滑坡启动的内在原因。而泥岩中高比例的粘土矿物组成尤其具有膨胀性质粘土矿物是泥岩具有特殊力学性质的重要原因,在两个典型滑坡的滑带位置,其岩层粘土矿物含量比滑坡其他部位相对较高,说明粘土矿物含量高低与岩层力学性能呈反比,高含量粘土矿物岩层成为潜在的软弱带。根据收集的资料和野外调查结果,通过ArcGIS软件处理,初步选取“高差”、“坡度”、“高程”、“居民地距离”、“公路距离”、“河流距离”、“坡向”、“地层”八个影响因素,形成空间地理数据库。基于因子分析和逻辑回归的统计分析方法在GIS平台对尖扎县进行区域滑坡危险性评价。利用因子分析的方法去除影响因素的相关性,并且将原先初选的八个影响因素降维处理形成四个主成分影响因素,在ArcGIS中形成“主成分空间地理数据库”,然后利用逻辑回归的方法,对选取的73个滑坡样本,200个非滑坡样本,以因子分析得来的四个主成分为自变量进行逻辑回归分析,具体操作在SPSS软件中执行,利用逻辑回归结果在ArcGIS上形成滑坡危险性区划图。通过对评价结果分析,得到如下结论:(1)尖扎县滑坡发育高危险性区域主要集中在县区东北部及东南部和西南部低洼处,危险程度总体上呈现由东北向西南逐渐降低的趋势,这与野外调查基本一致;(2)小尺度上高危险区的分布与人类活动和特定地层密切相关,主要分布在人类活动频繁的地区和K+N一套半胶结泥岩地层中,该套地层具有软硬岩层相间分布和透水地层和不透水地层相间分布的特点;(3)河流的分布及地形起伏度对滑坡的发育也有重要影响,但影响力不及地层因素和人类活动因素;(4)本文创造性地将利用因子分析改进的逻辑回归方法用于区域灾害评价中,改进后的逻辑回归方法,大大的提高了评价结果的可信度和精度。
[Abstract]:Based on the investigation of landslide geological hazard in Jianza County, Qinghai Province, this paper carries out mechanical and mineralogical experiments on the typical landslide in Jianza County, Lhasa landslide and Kangyang landslide, and analyzes the genetic mechanism and internal microscopic mineral composition of the landslide. It is found that the red mudstone strata have a controlling effect on the development of the landslide. The high proportion of clay minerals in mudstone, especially the swelling clay mineral, is an important reason for the special mechanical properties of mudstone, which is located in the slip zone of two typical landslides. The clay mineral content of the rock layer is relatively higher than that of other parts of the landslide, indicating that the clay mineral content is inversely proportional to the mechanical properties of the rock layer, and the high content clay mineral rock formation becomes a potential weak zone. Through ArcGIS software processing, eight influencing factors, such as "height difference", "slope degree", "elevation", "residential land distance", "highway distance", "river distance", "slope direction" and "stratum" are selected preliminarily. The method of statistical analysis based on factor analysis and logical regression was used to evaluate the risk of regional landslide in GIS platform. The correlation of influencing factors was removed by the method of factor analysis. And the original primary eight influencing factors are reduced to form four principal component influencing factors, and "principal component spatial geographical database" is formed in ArcGIS. Then, using the method of logical regression, the selected 73 landslide samples and 200 non-landslide samples are selected. The four principal components obtained by factor analysis are used as independent variables for logical regression analysis. The concrete operation is performed in SPSS software. The landslide hazard map is formed on ArcGIS by using the result of logical regression. The conclusions are as follows: (1) the high risk area of landslide development in Jianza County is mainly located in the northeast, southeast and southwest of the county, and the risk degree is gradually decreasing from northeast to southwest. This is basically consistent with field investigations.) the distribution of high risk areas on a small scale is closely related to human activities and specific strata, mainly in areas where human activities are frequent and in a set of K N semi-cemented mudstone formations. This set of strata has the characteristics of interphase distribution of hard and soft strata and interphase distribution of permeable and impermeable strata.) the distribution of rivers and the degree of topographic fluctuation also have an important influence on the development of landslides. But the influence is less than that of stratigraphic factor and human activity factor. (4) this paper creatively applies the improved logic regression method to regional disaster evaluation, and the improved logic regression method is applied to the evaluation of regional disaster. The reliability and accuracy of the evaluation results are greatly improved.
【学位授予单位】：中国地质大学(北京)
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P642.22;P208

【相似文献】