汶川地震前后崩塌和滑坡分布特征与敏感性对比分析

以汶川地震活动断裂龙门山断裂带穿过的12个县(市)为研究区,分析了研究区汶川地震前后崩塌和滑坡空间分布特征,并使用统计学方法Logistic回归模型分别对震前震后的崩塌和滑坡敏感性进行评价,结合崩塌和滑坡敏感性变化矩阵,分析了汶川地震后崩塌和滑坡敏感性的空间变化特征。震前崩塌和滑坡高敏感区主要沿河谷分布,而震后崩塌和滑坡高敏感区主要沿龙门山断裂带分布。震前的极高敏感区震后仍表现为极高敏感区,而靠近断裂带的区域,震前的低敏感区转换为高敏感和极高敏感区。     

基于GIS的城市滑坡灾害易发性评价——以湖北省宜昌市城区为例

[目的]对城市滑坡灾害进行易发性分区评价,为城市规划与防灾减灾工程提供理论依据。[方法]以湖北省宜昌市城区为研究区,通过GIS平台选取高程、坡度、地层岩性、归一化植被指数(NDVI)、与水系的距离和道路密度等6个评价因子,采用似然比方法分析评价因子和滑坡发育的关系,并以归一化似然比值将评价因子参数分类量化;以量化值作为Logistic回归模型的自变量,抽取样本数据建立滑坡易发性评价回归模型。[结果]评价因子结果显著,模型的整体准确率达到79.2%,ROC曲线下面积达0.871;极低易发区和低易发区占全区面积的61.59%,包含滑坡灾害的11.29%;高易发区和极高易发区虽仅占全区面积的17.88%,却发育了68.55%的滑坡灾害,结果与滑坡灾害分布特征相符合。[结论]对宜昌市城区的滑坡易发性进行了等级划分。采用GIS和Logistic回归相结合的滑坡易发性评价方法,结果准确可靠。     

基于GBDT-LR和信息量模型耦合的滑坡易发性评价

[目的]探索准确、快速的滑坡易发性区划方法,为区域安全监测提供参考,为政府治理滑坡灾害提供科学依据。[方法]以安徽省池州市贵池区为研究区域,采用梯度提升决策树—逻辑回归(GBDT-LR)和信息量(I)模型耦合的方法,实现区域滑坡易发性评价。该方法通过对原样本地学习,组合产生新的模拟样本,从而增强易发性评价模型对滑坡的拟合能力;采用Borderline-Smote算法解决样本数据不对称的问题。选用r.slopeunits软件划分的斜坡单元作为最小评价单元,选取坡度、坡向、地形曲率、剖面曲率、平面曲率、地形湿度指数(TWI)、地形起伏度、归一化植被指数(NDVI)、距断裂距离和距水系距离总计10个评价因子。分别从频率比、滑坡灾害点及隐患点密度、ROC曲线3个方面对构建的滑坡易发性模型进行评价。[结果]试验结果表明：耦合模型I-GBDT-LR分别比I,LR,I-LR模型的高易发区频率比所占比例提升约10%,13%,7%,高易发区滑坡灾害点及隐患点密度分别提升约9,11,7,ROC精度提升约10%,9%,5%。[结论]从检验指标综合来看,耦合模型的精度均高于单一模型,所提出耦合模型精度又高于I...     

四川省降雨诱发滑坡灾害的气象预警模型

[目的]分析四川省2008—2014年的滑坡灾害与前期降雨量间的关系,构建降雨诱发滑坡灾害气象预警模型,优化模型的权重参数,为四川省滑坡灾害预警工作提供有效参考。[方法]基于降雨信息资料,对四川省滑坡灾害的前期雨量进行统计分析,并采用逻辑回归分析分别计算和优化灾害的前期综合日降雨量间的权重关系和气象预警模型中地质环境背景值与降雨诱发灾害概率值的权重系数。[结果]滑坡灾害的当日降雨量、前一日降雨量和前两日降雨量对灾害的影响权重分别为0.587,0.220,0.189,气象预警模型的地质背景概率和综合雨量概率的权重参数分别为0.394,0.606。根据以上研究构建灾害气象预警模型,利用已有灾害点对其进行验证,证得模型的整体准确率为78.36%,进而通过2013年7月10日的群发滑坡灾害个例对模型进行检验,检验结果良好。[结论]该模型预警效果良好,精确度较高,能为相关部门提供的防灾减灾参考。     

高斯一牛顿法在滑坡预测预报模型参数优化中的应用

对滑坡预测预报的非线性模型,在估计其参数时,传统的方法是将非线性模型在参数的近似值处展开成泰勒级数,并仅取至一次项,然后再应用线性模型参数估计理论进行参数估计。因线性化时略去了二阶及二阶以上的各高次项,所以必然会产生模型误差。介绍了高斯牛顿法的基本原理,并以洒勒山新滑坡为例,在建立该滑坡灰色GM（1,1）模型和Verhulst模型的基础上,运用高斯-牛顿法对两个非线性模型的参数进行优化。计算结果表明,参数优化后各模型的预测精度比优化前各模型的精度有显著提高。说明采用高斯-牛顿法优化非线性模型参数是提高滑坡预测预报精度的一种有效且切实可行的方法。     

高斯-牛顿法在滑坡预测预报模型参数优化中的应用

对滑坡预测预报的非线性模型,存估计其参数时,传统的方法是将非线性模型在参数的近似值处展开成泰勒级数.并仅取至一次项,然后冉应用线性模型参数估计理论进行参数估计.因线性化时略去了,二阶及二阶以上的各高次项.所以必然会产生模型误差.介绍了高斯-牛顿法的基本原理,并以洒勒山新滑坡为例,在建立该滑坡厌色GM(1,1)模型和Verhulst模型的基础上,运用高斯牛顿法对两个非线性模型的参数进行优化.计算结果表明,参数优化后各模型的预测精度比优化前各模型的精度有显著提高.说明采用高斯-牛顿法优化非线性模型参数是提高滑坡预测预报精度的一种有效且切实可行的方法.     

黄土滑坡稳定性评价的集合卡尔曼滤波同化方法

[目的] 为提升区域滑坡稳定性评价模型的预测精度,解决传统滑坡稳定性分析基于静态的物理模型过度简化滑坡发生机理与力学机制,导致过度预测的缺点,以及模型参数通常具有的时空变异性、不确定性的问题。[方法] 基于集合卡尔曼滤波的数据同化方法,以甘肃省兰州市北环路周边区域为例,构建了基于TRIGRS模型和SBAS-InSAR观测数据的区域滑坡数据同化方案,对模型中的安全系数(F_s)进行同化,更新模型参数内摩擦角,进而修正滑坡稳定性,并利用均方根偏差(RMSD)检验同化值的精度。[结果] 同化后研究区域滑坡安全系数明显高于模型预测的结果,不稳定区域的面积比例由12 %降低至7 %,与实际观测更为接近;试验使内摩擦角参数逐渐向观测值方向改正,实现了模型参数的动态更新;均方根偏差从0.33减小到0.04左右。[结论] 基于集合卡尔曼滤波的数据同化方法有效修正了模型稳定性预测结果,可以更准确体现当前区域滑坡实际情况,具有更高的预测精度。     

汶川震区滑坡堆积体体积三维激光扫描仪测量与计算方法

2008年5.12汶川8级地震形成的大量松散滑坡堆积体,破坏农用地,损毁农田基础设施,对震区农业经济造成重大损失,严重威胁震区农业生产安全。本研究所选震区典型滑坡堆积体位于汶川县草坡乡,采用三维激光扫描仪实地测量滑坡堆积体,构建滑坡堆积体几何模型,计算滑坡堆积体体积。扫描仪到滑坡堆积体前、后缘的距离分别为25.74和79.00 m,距离扫描仪200 m处的扫描精度为20.0 cm×20.0 cm。根据扫描得到的滑坡堆积体点云数据,采用特征值法拟合近似滑坡堆积体的坡面、滑动面,构建滑坡堆积体几何模型。计算得到该滑坡堆积体体积为19 018 m3。建立了使用三维激光扫描仪测量和计算滑坡堆积体体积的方法。这种测量与计算方法可为地震受损农用地灾后的土地整理提供重要依据。     

三角带滑坡稳定性对极端气候的响应

三角带滑坡是典型的久治不愈滑坡,随着三峡库区蓄水,三角带滑坡呈现古滑坡的复活和新滑坡的形成双重变形,极端气候必定会对其稳定性产生影响。本次研究选取三角带滑坡的深层滑面、浅层滑面及高焦炉古滑坡的滑面为对象,对4种工况下的滑坡稳定性进行评价。评价结果表明:(1)在洪水和暴雨极端条件下,高焦炉古滑坡各剖面均处于稳定状态;三角带滑坡浅层滑面由基本稳定向欠稳定状态转变,局部出现不稳定状态;三角带滑坡深层滑面由基本稳定向欠稳定状态变化;(2)通过φ值与c值参数敏感性比较分析可以看出,φ值对滑坡稳定系数的敏感性较大;(3)综合分析结果表明,三个滑坡稳定性评价结果为:三角带滑坡深层滑面<三角带滑坡浅层滑面<高焦炉古滑坡,且在暴雨和洪水极端气候的双重影响下均发生一定的稳定性下降现象。     

川鄂褶皱山地溪洪-滑坡灾害与主要自然因子的关系——以香溪河流域为例

[目的]明确影响香溪河流域溪洪—滑坡灾害分布的瓶颈性自然因子,为其代表的三峡库区东部川鄂褶皱山地山洪地质灾害防治提供科学依据,并探讨地理探测器模型的应用表现。[方法]应用GIS技术、逐步回归和地理探测器模型,分析研究流域溪洪—滑坡灾害分布与主要自然因子的定量关系。[结果]进入回归模型的5项因子总解释度80.6%,贡献率依次为岩性脆弱性28.6%,断层缓冲区19.8%,NDVI18.9%,暴雨极值16.4%,岩性软硬程度16.3%;地理探测中对因变量解释度较大的5项因子分别为岩性脆弱性24.5%,断层缓冲区20.9%,NDVI 18.3%,土壤入渗性16.3%,岩性软硬程度14.8%;岩性脆弱性与NDVI值、坡度的交互作用强度分别达68.8%和62.0%。[结论]地质要素对流域溪洪—滑坡灾害分布具有绝对的控制作用,植被的影响其次,土壤和降水的影响有限,地形的作用体现在协同方面;地理探测器模型为定量分析非线性变量间的关联性提供了新的视角,但它消除了自变量的数值属性,在变量影响方向和重复性侦测方面有局限。     

基于IOE和SVM模型的府谷镇滑坡易发性分区

将陕西省府谷县府谷镇作为研究区,通过野外实地调查,圈定了47个滑坡点,制作了研究区滑坡编录图。以GIS软件和统计分析模型为基础,开展研究区滑坡易发性分区研究。首先通过GIS软件将滑坡点随机分成训练样本（70%）和测试样本（30%）两组。然后选择坡度、坡向、高程、距断层的距离、距道路的距离、距河流的距离、岩性、土地利用、NDVI、降雨量作为影响因子,提取因子图层。分别应用熵权模型（IOE）和支持向量机模型（SVM）计算滑坡易发性指数,利用自然间断点法将研究区划分为低易发区、中易发区、高易发区和极高易发区。最后利用ROC敏感度曲线下的面积（AUC）分别检验两种模型所得到的分区结果,结果表明,成功率曲线和预测度曲线的AUC值均在0.70～0.90,表明两种模型所得到的分区结果具有较高的精度,都可以为研究区的滑坡防治提供参考。在训练样本和测试样本中SVM模型的AUC值均最高,说明SVM模型比IOE模型适合在研究区开展滑坡预测研究。     

陇东地区黄土滑坡稳定性分析

在分析陇东地区地质环境背景的基础上,选择在坡度、坡高、滑坡体厚度等方面具有代表性的直线型、凸型、凹型和阶梯型黄土滑坡为计算模型,设定天然、饱水、地震和饱水+地震等4种工况,对不同类型、不同坡度、不同坡高、不同滑坡体厚度的黄土滑坡在各工况条件下的稳定性进行了分析。结果表明:从影响因素看,对滑坡稳定性影响最大的因素是坡度,其次是坡高;从工况条件看,诱发滑坡的主要因素是降雨入渗,其次是地震;从滑坡类型看,直线型滑坡稳定性相对较好,阶梯型滑坡和凸型滑坡次之,凹型滑坡稳定性最差。     

强震区滑坡活动强度的演变及敏感性评价

[目的]探讨汶川地震强震区滑坡的活动强度演变与敏感性,为滑坡、泥石流等地质灾害风险管理提供依据。[方法]选取四川省都江堰市龙池镇龙溪河12条泥石流流域为研究区,通过对研究区震后4期遥感影像的滑坡解译分析,研究震后滑坡的活动演化特征;同时利用概率综合判别法—层次分析法对滑坡物源进行多期敏感性评价。[结果]地震后该区域产生了825个强震滑坡;2009—2017年,在强震滑坡区域外新增376个滑坡,至2017年,仍活动的滑坡数量减少到368个,占滑坡总数的30.6%,同时利用曲线下面积(AUC)检验多期敏感性评价结果,准确率为75.6%～81.4%,评价效果较好。[结论]强震区震后活动滑坡数量及高敏感性区域面积整体表现降低趋势,表明震后滑坡处于逐步恢复过程,强震区的地质灾害活动但活动强度仍远远高于震前。     

Prediction of spatial soil property information from ancillary sensor data using ordinary linear regression: Model derivations, residual assumptions and model validation tests

Geospatial measurements of ancillary sensor data, such as bulk soil electrical conductivity or remotely sensed imagery data, are commonly used to characterize spatial variation in soil or crop properties. Geostatistical techniques like kriging with external drift or regression kriging are often used to calibrate geospatial sensor data to specific soil or crop properties. More traditional statistical methods such as ordinary linear regression models are also commonly used. Unfortunately, some soil scientists see these as competing and unrelated modeling approaches and are unaware of their relationship. In this article we review the connection between the ordinary linear regression model and the more comprehensive geostatistical mixed linear model and describe when and under what conditions ordinary linear regression models represent valid spatial prediction models. The formulas for the ordinary linear regression model parameter estimates and best linear unbiased predictions are derived from the geostatistical mixed linear model under two different residual error assumptions; i.e., strictly uncorrelated (SU) residuals and effectively uncorrelated (EU) residuals. The theoretically optimal (best linear unbiased) and computable (linear unbiased) predictions and variance estimates derived under the EU error assumption are examined in detail. Statistical tests for detecting spatial correlation in LR model residuals are also reviewed, in addition to three LR model validation tests derived from classical linear modeling theory. Two case studies are presented that highlight and demonstrate the various parameter estimation, response variable prediction and model validation techniques discussed in this article.     

A GIS-based comparative study of frequency ratio, analytical hierarchy process, bivariate statistics and logistics regression methods for landslide susceptibility mapping in Trabzon, NE Turkey

Over the last few decades, many researchers have produced landslide susceptibility maps using different techniques including the probability method (frequency ratio), the analytical hierarchy process (AHP), bivariate, multivariate, logistics regression, fuzzy logic and artificial neural network In addition, a number of parameters such as lithology, slope, aspect, land cover, elevation, distance to stream, drainage density, distance to lineament, seismicity, and distance to road are recommended to analyze the mechanism of landslides. The data quality is a very important issue in landslide studies, and more accurate results will be achieved if the data is adequate, appropriate and drawn from a wide range of parameters. The aim of this study was to evaluate the susceptibility of the occurrence of landslides in Trabzon province, situated in north east Turkey. This was achieved using the following five methods the frequency ratio model, AHP, the statistical index (Wi), weighting factor (Wf) methods, and the logistics regression model, incorporating a Geographical Information System (GIS) and remote sensing techniques. In Trabzon province there has been an increasing occurrence of landslides triggered by rainfall. These landslides have resulted in death, significant injury, damage to property and local infrastructure and threat of further landslides continues. In order to reduce the effects of this phenomenon, it is necessary to scientifically assess the area susceptible to landslide. To achieve this, landslide susceptible areas were mapped the landslide occurrence parameters were analyzed using five different methods. The results of the five analyses were confirmed using the landslide activity map containing 50 active landslide zones. Then the methods giving more accurate results were determined. The validation process showed that the Wf method is better in prediction than the frequency ratio model, AHP, the statistical index (Wi), and logistics regression model.     

A review on parameterization and uncertainty in modeling greenhouse gas emissions from soil

The efficacy of mathematical modeling as a tool for estimating greenhouse gas (GHG) emissions from soil depends on the uncertainty. Systematic evaluation of various sources of uncertainties in GHG emission models is limited. This paper reviews the state-of-the-art knowledge on the parameterization and uncertainty analysis of soil GHG emission models. Major recommendations and conclusions from this work include: (a) uncertainties due to model parameters and structure can be quantified by combining the Bayesian theorem and the Markov Chain Monte Carlo (MCMC) method; (b) uncertainty due to event-based model input may also be assessed by regarding each event as a latent variable; however, the necessity of the simultaneous evaluation of uncertainties from model input, parameters, and structure might be negotiable because strong correlations may exist between input errors and model parameters; (c) uncertainty analysis is essential for a successful model parameterization by reducing both the number of undetermined parameters and the parameter space; and (d) model parameterization (calibration) should be conducted on multiple sites towards multiple objectives. Case studies were presented for comparing the model uncertainties of the denitrification components of four models, DAYCENT, DNDC, ECOSYS, and COMP. The methods discussed in this paper can help to evaluate model uncertainties and performances, and to offer a critical guidance for model selection and parameterization.