A hybrid model using machine learning methods and GIS for potential rockfall source identification from airborne laser scanning data

The main objectives of this paper are to design and evaluate a hybrid approach based on Gaussian mixture model (GMM) and random forest (RF) for detecting rockfall source areas using airborne laser scanning data. The former model was used to calculate automatically slope angle thresholds for different type of landslides such as shallow, translational, rotational, rotational-translational, complex, debris flow, and rockfalls. After calculating the slope angle thresholds, a homogenous morphometric land use area (HMLA) was constructed to improve the performance of the model computations and reduce the sensitivity of the model to the variations in different conditioning factors. After that, the support vector machine (SVM) was applied in addition to backward elimination (BE) to select and rank the conditioning factors considering the type of landslides. Then, different machine learning methods [artificial neural network (ANN), logistic regression (LR), and random forest (RF) were trained with the selected best factors and previously prepared inventory datasets. The best fit method (RF) was then used to generate the probability maps and then the source areas were detected by combining the slope raster (reclassified according to the thresholds found by the GMM model) and the probability maps. The accuracy assessment shows that the proposed hybrid model could detect the potential rockfalls with an accuracy of 0.92 based on training data and 0.96 on validation data. Overall, the proposed model is an efficient model for identifying rockfall source areas in the presence of other types of landslides with an accepted generalization performance.     

Sinkhole characterization in the Dead Sea area using airborne laser scanning

Since the early 1980s, the Dead Sea coast has undergone a near catastrophic land deterioration as a result of a rapid lake-level drop. One conspicuous expression of this deterioration is the formation of sinkholes fields that puncture the coastal plains. The evolution of sinkholes along nearly 70-km strip has brought to a halt the regional development in this well-known and toured area and destroyed existing infrastructures. Great efforts are being invested in understanding the phenomena and in development of monitoring techniques. We report in this paper the application of airborne laser scanning for characterization of sinkholes. We demonstrate first the appropriateness of laser scanning for this task and its ability to provide detailed 3D information on this phenomenon. We describe then an autonomous means for their extraction over large regions and with high level of accuracy. Extraction is followed by their detailed geometric characterization. Using this high-resolution data, we show how sinkholes of 0.5 m radius and 25 cm depth can be detected from airborne platforms as well as the geomorphic features surrounding them. These sinkhole measures account for their embryonic stage, allowing tracking them at an early phase of their creation.     

Landslide vulnerability and risk assessment for multi-hazard scenarios using airborne laser scanning data (LiDAR)

Landslide hazard, vulnerability, and risk-zoning maps are considered in the decision-making process that involves land use/land cover (LULC) planning in disaster-prone areas. The accuracy of these analyses is directly related to the quality of spatial data needed and methods employed to obtain such data. In this study, we produced a landslide inventory map that depicts 164 landslide locations using high-resolution airborne laser scanning data. The landslide inventory data were randomly divided into a training dataset: 70 % for training the models and 30 % for validation. In the initial step, a susceptibility map was developed using logistic regression approach in which weights were assigned to every conditioning factor. A high-resolution airborne laser scanning data (LiDAR) was used to derive the landslide conditioning factors for the spatial prediction of landslide hazard areas. The resultant susceptibility was validated using the area under the curve method. The validation result showed 86.22 and 84.87 % success and prediction rates, respectively. In the second stage, a landslide hazard map was produced using precipitation data for 15 years. The precipitation maps were subsequently prepared and show two main categories (two temporal probabilities) for the study area (the average for any day in a year and abnormal intensity recorded in any day for 15 years) and three return periods (15-, 10-, and 5-year periods). Hazard assessment was performed for the entire study area. In the third step, an element at risk map was prepared using LULC, which was considered in the vulnerability assessment. A vulnerability map was derived according to the following criteria: cost, time required for reconstruction, relative risk of landslide, risk to population, and general effect to certain damage. These criteria were applied only on the LULC of the study area because of lack of data on the population and building footprint and types. Finally, risk maps were produced using the derived vulnerability and hazard information. Thereafter, a risk analysis was conducted. The LULC map was cross-matched with the results of the hazard maps for the return period, and the losses were aggregated for the LULC. Then, the losses were calculated for the three return periods. The map of the risk areas may assist planners in overall landslide hazard management.     

基于GIS 及聚类分析法的抚顺地区地质灾害综合评价

采用GIS 手段建立研究区地质灾害的多影响因素空间数据库,确定本区域引发地质灾害的因素评价体系。运用GIS 技术空间数据分析方法,结合评价分级准则,对地质灾害影响因素进行处理,采用聚类分析法初步确定4 类地质灾害聚类区。Ⅰ类区为抚顺市区工程活动主导因素型地质灾害区; Ⅱ 类区为沟谷盆地及其边缘工程活动诱发因素型地质灾害区; Ⅲ类区为北部及中部丘陵工程活动微弱因素型地质灾害区; Ⅳ类区为南部低山丘陵自然因素主导型地质灾害区。     

Semi-automatic shallow landslide detection by the integration of airborne imagery and laser scanning data

Landslide mapping is essential for effective watershed management. In Taiwan, a typhoon or earthquake event can trigger hundreds, even thousands, of shallow landslides in mountainous watersheds. Thus, improving the efficiency of landslide mapping by means of remote sensing techniques is an important issue. This study proposes a new method that uses concurrent aerial laser scanning (ALS) data and color ortho-imagery as input data: the topographic indices of slope, surface roughness, and object height model can be derived from the ALS data and the Green–Red Vegetation Index from the ortho-images. The method first uses these topographic and spectral indices in a global, semi-automatic algorithm to separate landslide from non-landslide pixels. It then offers a region growing tool and a 3D Eraser/Painter to edit detected landslides locally. These global and local operations are designed with a user interface, which is intuitive and user-friendly. Results from four test sites in a mountainous watershed prove that the method is easy, accurate, and suitable for landslide mapping in Taiwan.     

基于GIS的小江断裂中北段滑坡灾害危险性评价

小江断裂中北段作为金沙江下游重大水电工程库区内最重要的活动断裂构造,其沿线滑坡地质灾害发育,严重制约了当地的经济和社会发展。在详细调查了小江断裂中北段沿线滑坡地质灾害的基础上,选取了坡度、坡向、工程地质岩组、岸坡结构、断裂构造等9个影响因子作为评价指标,对研究区滑坡灾害的发育分布规律进行了分析研究,并采用AHP-CF法建立了滑坡灾害危险性评价模型,利用GIS空间分析功能得出了研究区滑坡地质灾害危险性分区。通过检验曲线验证滑坡地质危险性评价结果可知,AUC值为80%。     

Manifestation of remote sensing data and GIS on landslide hazard analysis using spatial-based statistical models

This paper presents landslide hazard analysis at Cameron area, Malaysia, using a geographic information system (GIS) and remote sensing data. Landslide locations were identified from interpretation of aerial photographs and field surveys. Topographical and geological data and satellite images were collected, processed, and constructed into a spatial database using GIS and image processing. The factors chosen that influence landslide occurrence are topographic slope, topographic aspect, topographic curvature, and distance to rivers, all from the topographic database; lithology and distance to faults were taken from the geologic database; land cover from TM satellite image; the vegetation index value was taken from Landsat images; and precipitation distribution from meteorological data. Landslide hazard area was analyzed and mapped using the landslide occurrence factors by frequency ratio and bivariate logistic regression models. The results of the analysis were verified using the landslide location data and compared with the probabilistic models. The validation results showed that the frequency ratio model (accuracy is 89.25%) is better in prediction of landslide than bivariate logistic regression (accuracy is 85.73%) model.     

机载LiDAR与地面三维激光扫描在贵州水城独家寨崩塌地质灾害风险调查中的应用

在贵州岩溶山区开展地质灾害风险调查时,由于局部存在高差较大的复杂山体,传统的地面调查手段往往具有局限性。为有效地识别及测量潜在的高位隐蔽性地质灾害隐患,采用无人机载LiDAR和地面三维激光扫描的方法,通过“俯视”数据与“正视”数据相融合,可完整、精确地获取崩塌危岩带的高精度点云及三维模型等数据。以贵州省水城区鸡场镇独家寨崩塌为例,通过野外数据采集-原始点云处理-不同数据融合-整体着色修复等步骤,最后得到研究区高精度三维模型数据,在此基础上对岩体结构和裂隙进行提取,解译出层面和两组节理裂隙产状,进而有效识别危岩体的空间分布,并基于网格划分方法计算出危岩带的规模约6.6×104 m3。结果表明无人机载LiDAR与地面三维激光扫描相融合的方法可优势互补,具有可操作性强、精度高、识别准等特点,可有效地获取解译并识别危岩体的分布及规模,为后续稳定性分析及风险评价提供基础。     

基于三维激光扫描的滑坡变形监测与数据处理

地面三维激光扫描技术可以高精度、高密度、高速度地测量物体表面三维空间坐标,从而详细描述表面细部状况。它已经在静态形状测量中获得了成功的应用。目前,滑坡地表变形分析主要是采用GPS或全站仪测量的少量特征点来进行,缺乏整体变形资料。但滑坡变形体的诸多细节变化对正确的变形分析有着重要的作用,将该技术引入到滑坡变形监测与分析中,可充分利用滑坡体上的大量点自然地物作为监测点,来完整监测和分析其变形。作为一项新的研究内容,文章对此进行了相应的理论分析与实际测量,获得了初步满意的结果。     

Shallow landslide hazard assessment using a physically based model and digital elevation data

 A model for the analysis of topographic influence on shallow landslide initiation is applied to an experimental mountain basin where high-resolution digital elevation data are available: the Cordon catchment (5 km²) located in northern Italy. The model delineates those areas most prone to shallow landsliding due to surface topographic effects on hydrologic response. The model is composed of two parts: a steady-state model for shallow sub-surface runoff and an infinite-slope Coulomb failure model which assumes that the soil is cohesionless at failure. An inventory of landslide scars is used to document sites of instability and to provide a test of model performance by comparing observed landslide locations with model predictions. The model reproduces the observed distribution of landslide locations in a consistent way, although spatial variations in soil strength and transmissivity, which are not accounted for in the model, influence specific distribution of landslide areas within regions of similar topographic control. Received: 15 October 1996 · Accepted: 25 June 1997     

Landslide and debris-flow hazard analysis and prediction using GIS in Minamata–Hougawachi area, Japan

On July 20, 2003, following a short duration of heavy rainfall, a debris-flow disaster occurred in the Minamata–Hougawachi area, Kumamoto Prefecture, Japan. This disaster was triggered by a landslide. In order to assess the landslide and debris-flow hazard potential of this mountainous region, the study of historic landslides is critical. The objective of the study is to couple 3D slope-stability analysis models and 2D numerical simulation of debris flow within a geographical information systems in order to identity the potential landslide-hazard area. Based on field observations, the failure mechanism of the past landslide is analyzed and the mechanical parameters for 3D slope-stability analysis are calculated from the historic landslide. Then, to locate potential new landslides, the studied area is divided into slope units. Based on 3D slope-stability analysis models and on Monte Carlo simulation, the spots of potential landslides are identified. Finally, we propose a depth-averaged 2D numerical model, in which the debris and water mixture is assumed to be a uniform continuous, incompressible, unsteady Newtonian fluid. The method accurately models the historic debris flow. According to the 2D numerical simulation, the results of the debris-flow model, including the potentially inundated areas, are analyzed, and potentially affected houses, river and road are mapped.     

基于高分二号卫星数据的煤矿区地质灾害信息提取研究

随着遥感数据获取技术和能力的全面提高,遥感数据呈现出明显的大数据特征。发展适应于遥感大数据的智能分析和信息挖掘技术,成为当前遥感技术研究的前沿。高分二号（GF-2）卫星数据是我国首颗自主研发的亚米级高分辨率卫星数据,具有观测幅宽、重访周期短、高辐射精度、高定位精度等优势,为未来我国地质灾害的长期、动态地监测和研究提供了高精度、稳定可靠的数据源。本文选取安徽谢桥煤矿2015年1月8日的GF-2卫星影像为研究数据,在对煤矿区主要地质灾害遥感地学分析的基础上,采用面向对象的影像分析方法对研究区由采煤活动所诱发的地质灾害信息进行自动提取。结果表明：利用GF-2卫星数据能够有效地识别地质灾害体的位置、范围、形态等空间分布特征;面向对象的自动提取方法对于煤矿区大面积的积水塌陷盆地、小规模的塌陷坑和线性的地裂缝都有很高的提取精度,识别精度达90% 以上;基于逐层剔除的思路构建的提取规则,为GF-2数据在地质灾害调查和大数据分析中的应用提供了很好的技术支持,也为其它地物目标的提取提供了参考,但在特征的选择和阈值的设定上需要具体分析。     

基于GIS的重庆万州区滑坡灾害危险性评价

在充分调查万州区地质环境及滑坡灾害基本特征的基础上,根据资料的有效性和可获得性,选取地表高程、坡度、地层岩性、地质构造、土地利用类型、区域交通建设及河流侵蚀冲刷7个影响滑坡发生的因素作为评价指标,采用AHP法确定各个指标权重并建立滑坡灾害危险性指数模型,通过GIS系统的空间分析功能进行栅格运算,得出研究区滑坡灾害危险性分区。采用上述指标和方法将重庆市万州区的滑坡灾害划分为极高危险区、高危险区、中危险区、低危险区和极低危险区,划分结果符合该区滑坡灾害的实际情况。     

焦作市典型煤矿塌陷区地质灾害危险性分析及防治对策

由于大规模地开发矿产资源,使焦作市成为地质灾害的易发区,引起的地质灾害分布范围广,危害程度大,已造成巨大的经济损失和人员伤亡。本文以焦作市某城中村改造建设场地地质灾害危险性评估为例,用概率积分法等数理统计的方法对采煤塌陷区的采深采厚比和地表移动变形值进行了预测,对焦作市典型煤矿区形成的地面塌陷、地裂缝等地质灾害的危险性进行了分析,并提出了相应的防治对策。为今后矿山的设计和地质灾害防治工作提供一定的科学依据。     

Assessment of soil erosion in a tropical mountain river basin of the southern Western Ghats,India using RUSLE and GIS

Revised Universal Soil Loss Equation(RUSLE) model coupled with transport limited sediment delivery(TLSD) function was used to predict the longtime average annual soil loss, and to identify the critical erosion-/deposition-prone areas in a tropical mountain river basin, viz., Muthirapuzha River Basin(MRB; area=271.75 km~2), in the southern Western Ghats, India. Mean gross soil erosion in MRB is 14.36 t ha~(-1) yr~(-1), whereas mean net soil erosion(i.e., gross erosion-deposition) is only 3.60 t ha~(-1) yr~(-1)(i.e., roughly 25% of the gross erosion). Majority of the basin area(~86%) experiences only slight erosion(5 t ha~(-1) yr~(-1)), and nearly 3% of the area functions as depositional environment for the eroded sediments(e.g., the terraces of stream reaches, the gentle plains as well as the foot slopes of the plateau scarps and the terrain with concordant summits). Although mean gross soil erosion rates in the natural vegetation belts are relatively higher, compared to agriculture, settlement/built-up areas and tea plantation, the sediment transport efficiency in agricultural areas and tea plantation is significantly high,reflecting the role of human activities on accelerated soil erosion. In MRB, on a mean basis, 0.42 t of soil organic carbon(SOC) content is being eroded per hectare annually, and SOC loss from the 4th order subbasins shows considerable differences, mainly due to the spatial variability in the gross soil erosion rates among the sub-basins. The quantitative results, on soil erosion and deposition, modelled using RUSLE and TLSD, are expected to be beneficial while formulating comprehensive land management strategies for reducing the extent of soil degradation in tropical mountain river basins.     

Assessment of earthquake hazard in Panama based on seismotectonic regionalization

Assessment of seismic hazard in Panama is made using a seismotectonic regionalization model. The coefficients of Gumbel's Type-I distribution are calculated and return periods for several magnitudes are found. From these coefficients intensities, peak ground acceleration and earthquake hazard for a set of return periods and epicentral distances are estimated and substantial variations in the probability of occurrence are noted. The Panama Fracture Zone (PFZ) and the Panama-South America Suture Zone (PSZ) provinces are the most active in producing earthquakes with a magnitude of about 7.0 in less than 16 yr. Magnitude 7.0 earthquakes in the Azuero province have a return period of about 160 yr, whereas in the Panama Deformed Belt (PDB) province the return period for magnitude 7.5 events is about 175 yr.     

Assessment of earthquake hazard in Panama based on seismotectonic regionalization

Assessment of seismic hazard in Panama is made using a seismotectonic regionalization model. The coefficients of Gumbel's Type-I distribution are calculated and return periods for several magnitudes are found. From these coefficients intensities, peak ground acceleration and earthquake hazard for a set of return periods and epicentral distances are estimated and substantial variations in the probability of occurrence are noted. The Panama Fracture Zone (PFZ) and the Panama-South America Suture Zone (PSZ) provinces are the most active in producing earthquakes with a magnitude of about 7.0 in less than 16 yr. Magnitude 7.0 earthquakes in the Azuero province have a return period of about 160 yr, whereas in the Panama Deformed Belt (PDB) province the return period for magnitude 7.5 events is about 175 yr.     

Development and parameterisation of a complex hydrogeological model based on high-resolution direct-push data

Ongoing developments in geological and hydrogeological investigation techniques, especially direct-push methods, have led to an increase in the quality, density and spatial resolution of data available from such investigations. This has created new challenges in the development of numerical models in terms of accurately and efficiently translating detailed and complex conceptual models into effective numerical models. Suitable geometrical and numerical modelling tools are essential in order to meet these challenges. This paper describes the development of a three-dimensional hydrogeological flow model for a contaminated site near Berlin, Germany, based on high-resolution geological data obtained principally using direct-push methods. The available data were first interpreted to construct a detailed GIS-based geological model, which formed the basis of the conceptual site model. The conceptual model was then translated into a geometrical model, which was used to create a finite element numerical model. An innovative geometry object-based approach enabled the complex structural details of the conceptual model to be accurately reproduced in the numerical model domain. The resulting three-dimensional steady-state unconfined flow model was successfully calibrated using external automated calibration software, whereby parameter values for groundwater recharge and hydraulic conductivity were determined.