基于加权证据权模型的丽江—小金河断裂中南段滑坡易发性评价

丽江—小金河断裂全新世活动强烈、地震频发,沿断裂带的滑坡地质灾害极为发育。以断裂带中南段两侧10 km为研究区,根据地质地理环境和滑坡发育特征,选取高程、坡度、坡向、距活动断裂距离、距河流水系距离、距道路距离、工程地质岩组、降雨量、土地利用类型以及地震动峰值加速度10个影响因子为评价指标,运用加权证据权模型,开展丽江—小金河断裂中南段滑坡易发性评价,基于自然断点法将滑坡易发程度划分为高易发、中等易发、低易发和非易发4个级别,评价结果AUC值为0.81。结果显示：(1)研究区内滑坡受坡度、断裂、水系、岩性因素的影响程度更高;(2)高易发区和中等易发区主要沿断裂带和金沙江等主要河流水系两侧分布,在玉龙县、松坪乡、大东乡等周边区域较集中;(3)西川乡处于高易发区,但目前滑坡灾害点较少,应加强关注。     

基于标签传播的岩性预测半监督学习算法研究

低渗透油气藏、致密油气藏、页岩油气藏等非常规油气藏的开发已成为全球油气开发的热点,也为测井解释带来新的挑战.为了提高测井解释精度,本文研究了岩性预测的半监督学习问题,提出了"聚类—人工标注—伪标注—分类"的岩性预测框架.首先,利用聚类算法选取待标注样本;然后,基于数据在特征空间和地理空间的相似性,利用图半监督学习方法实现人工标注样本到无标注样本的标注传播;最后,基于伪标注的置信度,采用加权支持向量机算法实现分类模型的设计与训练.本文在真实的测井数据上进行了大量的实验,发现半监督学习算法通过挖掘有标注数据和无标注数据中蕴含的分布特性,可获得更精确的岩性预测效果,即使对于不均衡的数据集,也能大幅提高分类模型在各类别上的准确率.进一步引入地理空间相似性,半监督的岩性预测模型在样本数量少的类别上的准确率得到了较大提高,从而验证了本文所提出方法的有效性.     

基于不同核函数的2010年玉树地震滑坡空间预测模型研究

基于统计学习理论与地理信息系统(GIS)技术的地震滑坡灾害空间预测是一个重要的研究方向,其可以对相似地震条件下地震滑坡的发生区域进行预测.2010年4月14日07时49分(北京时间),青海省玉树县发生了M_w6.9级大地震,作者基于高分辨率遥感影像解译与现场调查验证的方法,圈定了2036处本次地震诱发滑坡,这些滑坡大概分布在一个面积为1455.3 km²的矩形区域内.本文以该矩形区域为研究区,以GIS与支持向量机(SVM)模型为基础,开展基于不同核函数的地震滑坡空间预测模型研究.应用GIS技术建立玉树地震滑坡灾害及相关滑坡影响因子空间数据库,选择高程、坡度、坡向、斜坡曲率、坡位、水系、地层岩性、断裂、公路、归一化植被指数(NDVI)、同震地表破裂、地震动峰值加速度(PGA)共12个因子作为地震滑坡预测因子.以SVM模型为基础,基于线性核函数、多项式核函数、径向基核函数、S形核函数等4类核函数开展地震滑坡空间预测研究,分别建立了玉树地震滑坡危险性指数图、危险性分级图、预测结果图.4类核函数对应的模型正确率分别为79.87%,83.45%,84.16%,64.62%.基于不同的训练样本开展模型训练与讨论工作,表明径向基核函数是最适用于该地区的地震滑坡空间预测模型.本文为地震滑坡空间预测模型中核函数的科学选择提供了依据,也为地震区的滑坡防灾减灾工作提供了参考.     

基于长时间尺度下滑坡易发性评价研究———以甘肃省永靖县为例

利用决策树模型,基于五期土地利用评价因子,对甘肃省永靖县进行近40年的长时间尺度下的滑坡易发性评价,五期评价结果均显示研究区内滑坡灾害的极高和高易发性区域主要集中在中部黄河流域(盐锅峡镇至刘家峡水电站段)周边、西南部川城村—红泉镇—王台乡周边区域以及中部偏东的三条岘乡,该区域人口密集,人类活动较多.研究结果与前人研究结果类似,且通过受试者工作特征曲线的精度检验,说明五期评价结果均具有较高的可靠性.另外,研究区内的自然植被和裸土地与滑坡易发性指标之间具有负相关关系,而旱地、水域和城乡建设用地等人类活动频繁的区域则更容易导致滑坡灾害的发生.从时间尺度上来看,极高和高易发性分区面积逐年下降,但自 2000年,极高和高易发性分区面积减少速度出现显著减缓,同期,该区域内的土地利用变化为城乡建设用地面积增加而植被面积减少,这使得区域内边坡稳定性下降,使部分防灾工程措施的减灾能力下降.本研究为该地区的灾害预防、预测和城乡土地规划提供了参考.     

集镇地质灾害易发性分区研究——以梅州市大埔县银江镇为例

基于梅州市大埔县银江镇的地质灾害勘查资料,结合广东典型山地丘陵地区的地质环境条件,选取地质灾害点的点密度、面密度、体密度、坡度、岩土类型、断裂密度、年降雨量和人类工程活动强弱情况等8个因子,建立镇域地质灾害易发性分区的指标体系。在GIS平台上应用综合指数法,计算出833个基本评价单元（0.5 km×0.5 km）的地质灾害综合易发性指数,并据此进行了银江镇地质灾害易发性区划。研究发现：银江镇地质灾害分区以中、高易发区为主,分别占总面积的36.63%和36.67%,其中高易发区灾害点数量占到总数的74.08%。研究结果可以为合理减少银江镇的地质灾害危害性、风险性提供基础的地质依据。     

基于粒子群算法的地震预报方法研究

针对地震预测中定量计算的困难性,利用地震前兆异常高维数据特征,研究一种基于粒子群聚类算法的地震预报模型。该模型输入为条带、空区、短水准等14项异常指标数据,输出为震级分类。模型设定聚类平均距离为粒子群算法的评价函数,发掘分析地震前兆数据与地震震级的关系。结果表明该模型能有效地根据地震前兆数据预测地震震级,与传统聚类k-means算法模型相比,稳定性强,预报准确性更高。历史地震数据实例研究表明,本文提出的模型充分利用了粒子群算法的高鲁棒性、高适应性和群体智能的协同策略,是改进地震预报效能的途径之一。     

基于GIS的2005年克什米尔地震区滑坡易发性区划

2005年10月8日发生的MW7.6克什米尔地震在喜马拉雅地区的巴基斯坦与印度的北部触发了数千个滑坡。这些滑坡密集分布于6个地貌—地质—人类活动环境区内。基于ASTER卫星影像和GIS技术,构建并分析了一个包含2252个滑坡的空间数据库。应用滑坡多元评估判据确定单次地震事件触发滑坡各个地震滑坡控制参数的重要性。这些控制参数包括岩性、断裂、坡度、坡向、高程、土地覆盖类型、河流与公路。结果展示了4类级别的滑坡易发区域。此外,结果还表明了岩性对滑坡的影响作用最大,尤其在岩石高度破碎区域,例如页岩、板岩、碎屑岩、石灰岩与白云岩。还有,距离断层、河流与公路较近也是滑坡发生的一个重要因素。滑坡也常常发生在坡向朝南的中海拔斜坡上。灌木林地、草原、农业用地也是滑坡易发区域,而森林覆盖的斜坡较少发育滑坡。研究区的1/3被滑坡极高易发区或高易发区所覆盖,需要开展快速的滑坡减灾措施。其余的区域为滑坡低易发区与中易发区,相对稳定。本文研究支持以下观点:(1)地震触发滑坡往往发生在地震滑坡控制参数相关的特定区域内;(2)在西喜马拉雅地区,森林砍伐与公路建设往往是同震滑坡或地震后短期内滑坡发生的重要控制因素。     

地震动强度参数相关性对地震滑坡危险性影响分析

本文基于全概率地震滑坡危险性分析方法,利用蒙特卡罗模拟研究在不同临界屈服加速度a_c、永久位移模型、场地类别和断层距情况下,地震动强度参数相关性对地震滑坡危险性结果的影响规律。主要结果表明:在进行滑坡危险性分析时,不考虑多地震动强度参数相关性会造成预测位移值偏小,滑坡风险被低估。因此,考虑地震动强度参数相关性对滑坡危险性评价很有必要,这能使预测结果反映地震动参数样本作为输入时的实际相关性特征,为合理进行滑坡防护提供理论依据和参考。     

基于逻辑回归模型的九寨沟地震滑坡危险性评估

发生于2017年8月8日的四川九寨沟M_S7. 0地震触发了大量的同震滑坡。基于Geoeye-1震后0. 5m分辨率的遥感影像开展极震区同震滑坡解译,圈定了4 834处滑坡。选择高程、坡度、坡向、水平断层距离、垂直断层距离、震中距离、河流距离、道路距离、TPI指数以及岩性共10个因子作为地震滑坡的影响因子,应用逻辑回归(Logistic Regression,LR)模型开展九寨沟地震滑坡危险性评价,并对评价结果的合理性进行检验。结果表明,基于LR模型的滑坡危险性评价图与实际滑坡发育情况十分吻合,其中五花海—夏莫段、火花海和九寨天堂洲际大饭店—如意坝段均为滑坡危险性极高的区域。采用ROC曲线对危险性评价结果进行模型成功率与预测率的定量评价,结果显示,LR模型的预测精度较为理想,训练样本集和验证样本集的AUC值分别为0. 91和0. 89。文中结论为震区恢复重建工作中地震滑坡的防灾减灾提供了科学参考。     

基于ACCRBF网络的多层砖房震害预测

针对传统震害预测方法逐栋抽样计算建筑物抗震性能的不足,本文提出了一种基于蚁群聚类径向基(ACCRBF)网络模型的建筑物震害预测方法。依据不同地震动峰值加速度下多层砖房的实际震害资料,对模型进行训练,在模型的输入和输出之间建立映射关系,并利用这种映射关系对未知样本进行分类,实现对多层砖房的震害分析和预测。模型的输入为反映结构的震害影响因子,输出为给定的地震动峰值加速度下结构震害等级。研究表明,基于ACCRBF网络模型的多层砖房震害预测结果与震害实例基本吻合,具有推广应用价值。     

APPLICATION OF LOGISTIC REGRESSION MODEL FOR HAZARD ASSESSMENT OF EARTHQUAKE-TRIGGERED LANDSLIDES: A CASE STUDY OF 2017 JIUZHAIGOU(CHINA)MS7.0 EVENT

The M_S7.0 Jiuzhaigou earthquake in Sichuan Province of 8 August 2017 triggered a large number of landslides. A comprehensive and objective panorama of these landslides is of great significance for understanding the mechanism, intensity, spatial pattern and law of these coseismic landslides, recovery and reconstruction of earthquake affected area, as well as prevention and mitigation of landslide hazard. The main aim of this paper is to present the use of remote sensing images, GIS technology and Logistic Regression(LR)model for earthquake triggered landslide hazard mapping related to the 2017 Jiuzhaigou earthquake. On the basis of a scene post-earthquake Geoeye-1 satellite image(0.5m resolution), we delineated 4834 co-seismic landslides with an area of 9.63km². The ten factors were selected as the influencing factors for earthquake triggered landslide hazard mapping of Jiuzhaigou earthquake, including elevation, slope angle, aspect, horizontal distance to fault, vertical distance to fault, distance to epicenter, distance to roads, distance to rivers, TPI index, and lithology. Both landsliding and non-landsliding samples were needed for LR model. Centroids of the 4834 initial landslide polygons were extracted for landslide samples and the 4832 non-landslide points were randomly selected from the landslide-free area. All samples(4834 landslide sites and 4832 non-landslide sites)were randomly divided into the training set(6767 samples)and validation set(2899 samples). The logistic regression model was used to carry out the landslide hazard assessment of the Jiuzhaigou earthquake and the results show that the landslide hazard assessment map based on LR model is very consistent with the actual landslide distribution. The areas of Wuhuahai-Xiamo, Huohuahai and Inter Continental Hotel of Jiuzhai-Ruyiba are high hazard areas. In order to quantitatively evaluate the prediction results, the trained model calculated with the training set was evaluated by training set and validation set as the input of the model to get the output results of the two sets. The ROC curve was used to evaluate the accuracy of the model. The ROC curve for LR model was drawn and the AUC values were calculated. The evaluation result shows good prediction accuracy. The AUC values for the training and validation data set are 0.91 and 0.89, respectively. On the whole, more than 78.5% of the landslides in the study area are concentrated in the high and extremely high hazard zones. Landslide point density and landslide area density increase very rapidly as the level of hazard increases. This paper provides a scientific reference for earthquake landslides, disaster prevention and mitigation in the earthquake area.     

基于SVM测井数据的火山岩岩性识别——以辽河盆地东部坳陷为例

辽河盆地东部坳陷储集层由火山多期喷发形成,岩相岩性复杂,岩性以中、基性火山岩为主.本文将火山岩的岩心及岩矿鉴定资料与测井数据进行整合,应用测井数据建立支持向量机(SVM)两分类和多分类岩性识别模式.首先,深入研究支持向量机二分类及"一对一"、"一对多"和有向无环图三种经典多分类算法的基本原理及结构;然后,总结研究区域火山岩岩石特征,分析测井数据的测井响应组合特征,选择40口井中岩心分析和薄片鉴定资料完整、常规五种测井曲线(RLLD,CNL,DEN,AC,GR)齐全的1200个测井数据作为训练样本,构造三种支持向量机岩性识别模式;最后,对4测试井中800个测井数据进行岩性识别,识别结果与取心段岩心描述和岩心/岩屑薄片鉴定资料对比,实验结果表明有向无环图更适合辽河盆地火山岩的识别,识别正确率达到82.3%.     

基于GIS的白龙江流域舟曲—武都段的滑坡危险性评价

本研究以围绕着白龙江流域的甘肃省南部的宕昌县、舟曲县和武都区部分地区为研究区,根据全国滑坡编目中得到的272个历史滑坡数据以及选取的高程、坡度、坡向、平面曲率、剖面曲率、归一化植被指数（NDVI）、降雨、岩性、距道路距离和距河流距离10种影响因子,利用三种具有代表性的定量方法：信息量模型、以及基于频率比模型的逻辑回归模型和人工神经网络模型对研究区内滑坡灾害危险性进行评价。三种评价结果均显示研究区内滑坡灾害的极高和高危险区主要沿白龙江河谷地区呈带状分布。从危险性分区图可看出,人工神经网络模型得到的分区图较为合理,既表现出沿河谷地区集中分布的趋势,也呈现出对滑坡历史数据较为独立的特征,这一研究结果与前人研究结果一致。根据受试者工作特征曲线（ROC曲线）对三种模型的精度进行检验,检验得到的AUC值分别为0.818、0.829和0.837,说明三种评价结果均具有较高的可靠性,基于频率比模型的人工神经网络模型相比其他两个模型具有更好的评价精度,能更好地进行滑坡危险性的预测和评价,其中高程、降雨、岩性以及距道路距离对评价结果影响更大,这四种影响因子重要性值占比为52.1%。为该地区的城市扩建与灾害预防预测提供了参考。     

汶川地震滑坡危险性评价——以武都区和文县为例

利用GIS技术详细研究汶川地震在甘肃省陇南市武都区和文县触发的滑坡地质灾害的分布规律及其与地震烈度、地形坡度、断层、高程、地层岩性的相关关系,采用基于GIS的加权信息量模型的崩塌滑坡危险性评价方法,对研究区的地震滑坡危险性进行学科分析。结果表明：极高危险区在高程上主要分布在集水高程区,高度危险区主要沿白水江、白龙江等主干河流两侧极高易发区的边界向两侧扩展,轻度和极轻度危险区面积占比较小,主要分布在低烈度、活动断裂不发育、人类活动微弱的高海拔地区,另外国道G215沿极高危险性区域分布明显;利用危险性等级分区结果统计人口公里格网数据,得到武都区和文县潜在影响人口,发现研究区约78万人将受到地震滑坡灾害的潜在影响。     

基于改进EWT和LogitBoost集成分类器的地震事件分类识别算法

快速、准确地识别天然地震和人工爆破事件是地震台网监测的重要工作之一,也是提高地震观测记录质量、开展地震研究工作的重要基础.针对反向传播神经网络、支持向量机等主流分类识别方法在地震事件分类识别应用上的不足,提出一种基于改进 EWT 和 LogitBoost集成分类器的地震事件分类识别算法.首先,基于S谱能量曲线对传统经验小波变换进行改进,将信号自适应分解为按频率和能量分布的本征模函数;其次,提取 P波与S波最大振幅比,前4个本征模函数的香农熵、对数能量熵,以及去噪后重构信号主频等特征;最后,采用基于集成学习 LogitBoost的决策树集成分类器进行分类.实验结果表明,所提算法具有较高的鲁棒性,能有效解决样本不足的问题,识别准确率达93．1%以上,比集成学习 AdaBoost、反向传播神经网络和支持向量机等方法提高了1%以上,且分类识别效果好.     

Landslide intelligent prediction using object-oriented method

Landslides are one of the most serious geological disasters in the world and happen quite frequently in the Three Gorges. Landslide prediction is a very important measure of landslide prevention and cure in the Three Gorges. Traditional methods lack in sufficiently mining the various complex information from a landslide system. They often need much manual intervention and possess poor intelligence and accuracy. An intelligent method proposed in this paper for landslide prediction based on an object-oriented method and knowledge driving is hopeful to solve the above problem. The method adopted Landsat ETM+ images, 1:50,000 geological map and 1:10,000 relief map in the Three Gorges as the data origins. It firstly produced the key factors influencing landslide development and used multi-resolution segmentation algorithm to segment the image objects based on the key landslide factors of engineering rock group, reservoir water fluctuation, slope structure and slope level. Secondly, the method chose some sample objects and adopted the decision tree algorithm C5.0 to mine the landslide forecast criteria according to the factor values of each sample object. Finally, under knowledge driving the method classified the image objects and realized landslide susceptibility analysis and intelligent prediction in the Three Gorges. The method proposed in this paper is object-oriented. Results of a real-world example show that: (1) the object-oriented method possesses much more compact knowledge representation, higher efficiency, more continuous classifying result and higher prediction accuracy compared with the pixel-oriented method; (2) it possesses the overall accuracy of 87.64% and kappa coefficient of 0.8305 and is more accurate than the other seven methods (such as the pixel-oriented methods of Parallelpiped, Minimum Distance, Maximum Likelihood, Mahalanobis Distance, K-means and Isodata and the object-oriented method of Nearest Neighbor); (3) about 46.97% landslides lie in the high susceptibility region, 24.24% landslides lie in the moderate susceptibility region, 27.27% landslides lie in the low susceptibility region and 1.52% landslides lie in the very low susceptibility region. Therefore the method can effectively realize landslide susceptibility analysis and provides a new idea for landslide intelligent and accurate prediction.     

A spatiotemporal probabilistic modelling of storm‐induced shallow landsliding using aerial photographs and logistic regression

Landslides constitute one of the major natural hazards that could cause significant losses of life and property. Mapping or delineating areas prone to landsliding is therefore essential for land‐use activities and management decision making in hilly or mountainous regions. A landslide hazard map can be constructed by a qualitative combination of maps of site conditions, including geology, topography and geomorphology, by statistical methods through correlating landslide occurrence with geologic and geomorphic factors, or by using safety factors from stability analysis. A landslide hazard map should provide information on both the spatial and temporal probabilities of landsliding in a certain area. However, most previous studies have focused on susceptibility mapping, rather than on hazard mapping in a spatiotemporal context. This study aims at developing a predictive model, based on both quasi‐static and dynamic variables, to determine the probability of landsliding in terms of space and time. The study area selected is about 13 km² in North Lantau, Hong Kong. The source areas of the landslides caused by the rainstorms of 18 July 1992 and 4–5 November 1993 were interpreted from multi‐temporal aerial photographs. Landslide data, lithology, digital elevation model data, land cover, and rainfall data were digitized into a geographic information system database. A logistic regression model was developed using lithology, slope gradient, slope aspect, elevation, slope shape, land cover, and rolling 24 h rainfall as independent variables, since the dependent variable could be expressed in a dichotomous way. This model achieved an overall accuracy of 87·2%, with 89·5% of landslide grid cells correctly classified and found to be performing satisfactorily. The model was then applied to rainfalls of a variety of periods of return, to predict the probability of landsliding on natural slopes in space and time. It is observed that the modelling techniques described here are useful for predicting the spatiotemporal probability of landsliding and can be used by land‐use planners to develop effective management strategies. Copyright © 2003 John Wiley & Sons, Ltd.     

Hybrid meta-heuristic machine learning methods applied to landslide susceptibility mapping in the Sahel-Algiers

Landslides are considered as one among many phenomena jeopardizing human beings as well as their constructions. To prevent this disastrous problem, researchers have used several approaches for landslide susceptibility modeling, for the purpose of preparing accurate maps marking landslide prone areas. Among the most frequently used approaches for landslide susceptibility mapping is the Artificial Neural Network (ANN) method. However, the effectiveness of ANN methods could be enhanced by using hybrid metaheuristic algorithms, which are scarcely applied in landslide mapping. In the current study, nine hybrid metaheuristic algorithms, genetic algorithm (GA)-ANN, evolutionary strategy (ES)-ANN, ant colony optimization (ACO)-ANN, particle swarm optimization (PSO)-ANN, biogeography based optimization (BBO)-ANN, gravitational search algorithm (GHA)-ANN, particle swarm optimization and gravitational search algorithm (PSOGSA)-ANN, grey wolves optimization (GWO)-ANN, and probability based incremental learning (PBIL)-ANN have been used to spatially predict landslide susceptibility in Algiers’ Sahel, Algeria. The modeling phase was done using a database of 78 landslides collected utilizing Google Earth images, field surveys, and six conditioning factors (lithology, elevation, slope, land cover, distance to stream, and distance to road). Initially, a gamma test was used to decrease the input variable numbers. Furthermore, the optimal inputs have been modeled by the mean of hybrid metaheuristic ANN techniques and their performance was assessed through seven statistical indicators. The comparative study proves the effectiveness of the co-evolutionary PSOGSA-ANN model, which yielded higher performance in predicting landslide susceptibility compared to the other models. Sensitivity analysis using the step-by-step technique was done afterward, which revealed that the distance to the stream is the most influential factor on landslide susceptibility, followed by the slope factor which ranked second. Lithology and the distance to road have demonstrated a moderate effect on landslide susceptibility. Based on these findings, an accurate map has been designed to help land-use managers and decision-makers to mitigate landslide hazards.