Applying different scenarios for landslide spatial modeling using computational intelligence methods

Landslides every year impose extensive damages to human beings in various parts of the world; therefore, identifying prone areas to landslides for preventive measures is essential. The main purpose of this research is applying different scenarios for landslide susceptibility mapping by means of combination of bivariate statistical (frequency ratio) and computational intelligence methods (random forest and support vector machine) in landslide polygon and point formats. For this purpose, in the first step, a total of 294 landslide locations were determined from various sources such as aerial photographs, satellite images, and field surveys. Landslide inventory was randomly split into a testing dataset 70% (206 landslide locations) for training the different scenarios, and the remaining 30% (88 landslides locations) was used for validation purposes. To providing landslide susceptibility maps, 13 conditioning factors including altitude, slope angle, plan curvature, slope aspect, topographic wetness index, lithology, land use/land cover, distance from rivers, drainage density, distance from fault, distance from roads, convergence index, and annual rainfall are used. Tolerance and the variance inflation factor indices were used for considering multi-collinearity of conditioning factors. Results indicated that the smallest tolerance and highest variance inflation factor were 0.31 and 3.20, respectively. Subsequently, spatial relationship between classes of each landslide conditioning factor and landslides was obtained by frequency ratio (FR) model. Also, importance of the mentioned factors was obtained by random forest (RF) as a machine learning technique. The results showed that according to mean decrease accuracy, factors of altitude, aspect, drainage density, and distance from rivers had the greatest effect on the occurrence of landslide in the study area. Finally, the landslide susceptibility maps were produced by ten scenarios according to different ensembles. The receiver operating characteristics, including the area under the curve (AUC), were used to assess the accuracy of the models. Results of validation of scenarios showed that AUC was varying from 0.668 to 0.749. Also, FR and seed cell area index indicators show a high correlation between the susceptibility classes with the landslide pixels and field observations in all scenarios except scenarios 10_RF and 10_SVM. The results of this study can be used for landslides management and mitigation and development activities such as construction of settlements and infrastructure in the future.     

National-scale data-driven rainfall induced landslide susceptibility mapping for China by accounting for incomplete landslide data

China is one of the countries where landslides caused the most fatalities in the last decades.The threat that landslide disasters pose to people might even be greater in the future,due to climate change and the increasing urbanization of mountainous areas.A reliable national-scale rainfall induced landslide suscep-tibility model is therefore of great relevance in order to identify regions more and less prone to landslid-ing as well as to develop suitable risk mitigating strategies.However,relying on imperfect landslide data is inevitable when modelling landslide susceptibility for such a large research area.The purpose of this study is to investigate the influence of incomplete landslide data on national scale statistical landslide susceptibility modeling for China.In this context,it is aimed to explore the benefit of mixed effects mod-elling to counterbalance associated bias propagations.Six influencing factors including lithology,slope,soil moisture index,mean annual precipitation,land use and geological environment regions were selected based on an initial exploratory data analysis.Three sets of influencing variables were designed to represent different solutions to deal with spatially incomplete landslide information:Set 1(disregards the presence of incomplete landslide information),Set 2(excludes factors related to the incompleteness of landslide data),Set 3(accounts for factors related to the incompleteness via random effects).The vari-able sets were then introduced in a generalized additive model(GAM:Set 1 and Set 2)and a generalized additive mixed effect model(GAMM:Set 3)to establish three national-scale statistical landslide suscep-tibility models:models 1,2 and 3.The models were evaluated using the area under the receiver operating characteristics curve(AUROC)given by spatially explicit and non-spatial cross-validation.The spatial pre-diction pattern produced by the models were also investigated.The results show that the landslide inven-tory incompleteness had a substantial impact on the outcomes of the statistical landslide susceptibility models.The cross-validation results provided evidence that the three established models performed well to predict model-independent landslide information with median AUROCs ranging from 0.8 to 0.9.However,although Model 1 reached the highest AUROCs within non-spatial cross-validation(median of 0.9),it was not associated with the most plausible representation of landslide susceptibility.The Model 1 modelling results were inconsistent with geomorphological process knowledge and reflected a large extent the underlying data bias.The Model 2 susceptibility maps provided a less biased picture of landslide susceptibility.However,a lower predicted likelihood of landslide occurrence still existed in areas known to be underrepresented in terms of landslide data(e.g.,the Kuenlun Mountains in the northern Tibetan Plateau).The non-linear mixed-effects model(Model 3)reduced the impact of these biases best by introducing bias-describing variables as random effects.Among the three models,Model 3 was selected as the best national-scale susceptibility model for China as it produced the most plausible portray of rainfall induced landslide susceptibility and the highest spatially explicit predictive perfor-mance(median AUROC of spatial cross validation 0.84)compared to the other two models(median AUROCs of 0.81 and 0.79,respectively).We conclude that ignoring landslide inventory-based incomplete-ness can entail misleading modelling results and that the application of non-linear mixed-effect models can reduce the propagation of such biases into the final results for very large areas.     

Improving the accuracy of landslide susceptibility model using a novel region-partitioning approach

Landslide is a natural disaster that threatens human lives and properties worldwide. Numerous have been conducted on landslide susceptibility mapping (LSM), in which each has attempted to improve the accuracy of final outputs. This study presents a novel region-partitioning approach for LSM to understand the effects of partitioning a focused region into smaller areas on the prediction accuracy of common regression models. Results showed that the partitioning of the study area into two regions using the proposed method improved the prediction rate from 0.77 to 0.85 when support vector machine was used, and from 0.87 to 0.88 when logistic regression model was utilized. The spatial agreements of the models were also improved after partitioning the area into two regions based on Shannon entropy equations. Our comparative study indicated that the proposed method outperformed the geographically weighted regression model that considered the spatial variations in landslide samples. Overall, the main advantages of the proposed method are improved accuracy and the reduction of the effects of spatial variations exhibited in landslide-conditioning factors.     

A new region-based preparatory factor for landslide susceptibility models: the total flux

This paper presents a new region-based preparatory factor, total flux (TF), for landslide susceptibility models (LSMs). TF takes into account the topography and hydrology conditions upstream of each gridded data cell and represents the total flux of water in the stream. The results show that TF is strongly associated with the occurrence of landslides and is a good preparatory factor for LSM. Using TF instead of a drainage distance factor in I-Lan region in Taiwan shows an improvement in the accuracy of the cumulative percentage of landslide occurrence of 44 and 14 % for the top 1 and 10 % susceptible areas, respectively. This significant improvement in accuracy in these high-risk areas is critical for preventing and mitigating the economic and human losses due to landslides.     

Comparing GIS-based support vector machine kernel functions for landslide susceptibility mapping

This study compares the predictive performance of GIS-based landslide susceptibility mapping (LSM) using four different kernel functions in support vector machines (SVMs). Nine possible causal criteria were considered based on earlier similar studies for an area in the eastern part of the Khuzestan province of southern Iran. Different models and the resulting landslide susceptibility maps were created using information on known landslide events from a landslide inventory dataset. The models were trained using landslide inventory dataset. A two-step accuracy assessment was implemented to validate the results and to compare the capability of each function. The radial basis function was identified as the most efficient kernel function for LSM with the resulting landslide susceptibility map showing the highest predictive accuracy, followed by the polynomial kernel function. According to the obtained results, it concluded that using SVMs can generally be considered to be an effective method for LSM while it demands careful consideration of kernel function. The results of the present research will also assist other researchers to select the best SVM kernel function to use for LSM.     

Centrifuge modelling of landslides and landslide hazard mitigation: A review

Landslides are serious geohazards that occur under a variety of climatic conditions and can cause many casualties and significant economic losses. Centrifuge modelling, as a representative type of physical modelling, provides a realistic simulation of the stress level in a small-scale model and has been applied over the last 50 years to develop a better understanding of landslides. With recent developments in this technology, the application of centrifuge modelling in landslide science has significantly increased. Here, we present an overview of physical models that can capture landslide processes during centrifuge modelling. This review focuses on (i) the experimental principles and considerations, (ii) landslide models subjected to various triggering factors, including centrifugal acceleration, rainfall, earthquakes, water level changes, thawing permafrost, excavation, external loading and miscellaneous conditions, and (iii) different methods for mitigating landslides modelled in centrifuge, such as the application of nails, piles, geotextiles, vegetation, etc. The behaviors of all the centrifuge models are discussed, with emphasis on the deformation and failure mechanisms and experimental techniques. Based on this review, we provide a best-practice methodology for preparing a centrifuge landslide test and propose further efforts in terms of the seven aspects of model materials, testing design and equipment, measurement methods, scaling laws, full-scale test applications, landslide early warning, and 3D modelling to better understand the complex behaviour of landslides.     

Innovative landslide susceptibility mapping supported by geomorphon and geographical detector methods

Landslides are among the most common and dangerous natural hazards in mountainous regions that can cause damage to properties and loss of lives. Landslide susceptibility mapping (LSM) is a critical tool for preventing or mitigating the negative impacts of landslides. Although many previous studies have employed various statistical methods to produce quantitative maps of the landslide susceptibility index (LSI) based on inventories of past landslides and contributing factors, they are mostly ad hoc to a specific area and their success has been hindered by the lack of a methodology that could produce the right mapping units at proper scale and by the lack of a general framework for objectively accounting for the differing contribution of various preparatory factors. This paper addresses these issues by integrating the geomorphon and geographical detector methods into LSM to improve its performance. The geomorphon method, an innovative pattern recognition approach for identifying landform elements based on the line of sight concept, is adapted to delineate ridge lines and valley lines to form slope units at self-adjusted spatial scale suitable for LSM. The geographical detector method, a spatial variance analysis method, is integrated to objectively assign the weights of contributing factors for LSM. Applying the new integrated approach to I-Lan, Taiwan produced very significant improvement in LSI mapping performance than a previous model, especially in highly susceptible areas. The new method offers a general framework for better mapping landslide susceptibility and mitigating its negative impacts.     

滑坡灾害预测分区的计算机制图

本文在滑坡灾害预测分区的信息模型基础上,重点讨论了灾害预测的计算机制图化的主要过程:因素的数值化,单元边界的确定和彩色图件的绘制。运用中国地质大学计算机系开发的Mapcad系统,在Mv/10000计算机上较好地处理了不规则图幅边界的自然裁剪,不规则单元的输入,以及彩色图件的绘制等问题。     

A review of assessing landslide frequency for hazard zoning purposes

The probability of occurrence is one of the key components of the risk equation. To assess this probability in landslide risk analysis, two different approaches have been traditionally used. In the first one, the occurrence of landslides is obtained by computing the probability of failure of a slope (or the reactivation of existing landslides). In the second one, which is the objective of this paper, the probability is obtained by means of the statistical analysis of past landslide events, specifically by the assessment of the past landslide frequency. In its turn, the temporal frequency of landslides may be determined based on the occurrence of landslides or from the recurrence of the landslide triggering events over a regional extent. Hazard assessment using frequency of landslides, which may be taken either individually or collectively, requires complete records of landslide events, which is difficult in some areas. Its main advantage is that it may be easily implemented for zoning. Frequency assessed from the recurrence of landslide triggers, does not require landslide series but it is necessary to establish reliable relations between the trigger, its magnitude and the occurrence of the landslides. The frequency of the landslide triggers can be directly used for landslide zoning. However, because it does not provide information on the spatial distribution of the potential landslides, it has to be combined with landslide susceptibility (spatial probability analysis) to perform landslide hazard zoning. Both the scale of work and availability of data affect the results of the landslide frequency and restrict the spatial resolution of frequency zoning as well. Magnitude–frequency relationships are fundamental elements for the quantitative assessment of both hazard and risk.     

GIS-based weights-of-evidence modelling of rainfall-induced landslides in small catchments for landslide susceptibility mapping

Landslide susceptibility mapping is a vital tool for disaster management and planning development activities in mountainous terrains of tropical and subtropical environments. In this paper, the weights-of-evidence modelling was applied, within a geographical information system (GIS), to derive landslide susceptibility map of two small catchments of Shikoku, Japan. The objective of this paper is to evaluate the importance of weights-of-evidence modelling in the generation of landslide susceptibility maps in relatively small catchments having an area less than 4 sq km. For the study area in Moriyuki and Monnyu catchments, northeast Shikoku Island in west Japan, a data set was generated at scale 1:5,000. Relevant thematic maps representing various factors (e.g. slope, aspect, relief, flow accumulation, soil depth, soil type, land use and distance to road) that are related to landslide activity were generated using field data and GIS techniques. Both catchments have homogeneous geology and only consist of Cretaceous granitic rock. Thus, bedrock geology was not considered in data layering during GIS analysis. Success rates were also estimated to evaluate the accuracy of landslide susceptibility maps and the weights-of-evidence modelling was found useful in landslide susceptibility mapping of small catchments.     

Optimized object-based image classification: development of landslide knowledge decision support system

Landslide spatial decision support systems (LS-DSS) are computer-based systems that combine the geographic storage, search, and retrieval capabilities of geographic information systems with the decision models and optimizing algorithms used to support decision-making for landslide problems. This study proposes an optimization process of region object-oriented classification (ROC) to analyze the landslide image information. The surface information from the Wan Da reservoir area is collected and studied. We collected different spectrum with several texture information to analyze the surrounding area of the Wan Da reservoir. ROC is used to classify the landslide area. Entropy-based classification is used as a classifier in ROC to determine the landslide/nonlandslide area. The parameters of S (similarity) and A (area) are used and then the best combinations are found. An optimize algorithm is developed to access the above variables to perform the best classification outcomes. The relations of occurrence vs. non-occurrence of landslide which are linked to the attributes of land surface are studied. An improved translation model (Expert Knowledge Translation Platform) is also presented to increase the accuracy. This could be of help to manage/monitor the landslide area near the reservoir.     

A statistical approach to landslide risk modelling at basin scale: from landslide susceptibility to quantitative risk assessment

A procedure for landslide risk assessment is presented. The underlying hypothesis is that statistical relationships between past landslide occurrences and conditioning variables can be used to develop landslide susceptibility, hazard and risk models. The latter require also data on past damages. Landslides occurred during the last 50 years and subsequent damages were analysed. Landslide susceptibility models were obtained by means of Spatial Data Analysis techniques and independently validated. Scenarios defined on the basis of past landslide frequency and magnitude were used to transform susceptibility into quantitative hazard models. To assess vulnerability, a detailed inventory of exposed elements (infrastructures, buildings, land resources) was carried out. Vulnerability values (0–1) were obtained by comparing damages experienced in the past by each type of element with its actual value. Quantitative risk models, with a monetary meaning, were obtained for each element by integrating landslide hazard and vulnerability models. Landslide risk models showing the expected losses for the next 50 years were thus obtained for the different scenarios. Risk values obtained are not precise predictions of future losses but rather a means to identify areas where damages are likely to be greater and require priority for mitigation actions.     

A graph deep learning method for landslide displacement prediction based on global navigation satellite system positioning

The accurate prediction of displacement is crucial for landslide deformation monitoring and early warning. This study focuses on a landslide in Wenzhou Belt Highway and proposes a novel multivariate landslide displacement prediction method that relies on graph deep learning and Global Navigation Satellite System (GNSS) positioning. First model the graph structure of the monitoring system based on the engineering positions of the GNSS monitoring points and build the adjacent matrix of graph nodes. Then construct the historical and predicted time series feature matrixes using the processed temporal data including GNSS displacement, rainfall, groundwater table and soil moisture content and the graph structure. Last introduce the state-of-the-art graph deep learning GTS (Graph for Time Series) model to improve the accuracy and reliability of landslide displacement prediction which utilizes the temporal-spatial dependency of the monitoring system. This approach outperforms previous studies that only learned temporal features from a single monitoring point and maximally weighs the prediction performance and the priori graph of the monitoring system. The proposed method performs better than SVM, XGBoost, LSTM and DCRNN models in terms of RMSE (1.35 mm), MAE (1.14 mm) and MAPE (0.25) evaluation metrics, which is provided to be effective in future landslide failure early warning.     

Application and verification of fuzzy algebraic operators to landslide susceptibility mapping

The aim of this study was to apply and to verify the use of fuzzy logic to landslide susceptibility mapping in the Gangneung area, Korea, using a geographic information system (GIS). For this aim, in the study, a data-derived model (frequency ratio) and a knowledge-derived model (fuzzy operator) were combined. Landslide locations were identified by changing the detection technique of KOMPSAT-1 images and checked by field studies. For landslide susceptibility mapping, maps of the topography, lineaments, soil, forest, and land cover were extracted from the spatial data sets, and the eight factors influencing landslide occurrence were obtained from the database. Using the factors and the identified landslide, the fuzzy membership values were calculated. Then fuzzy algebraic operators were applied to the fuzzy membership values for landslide susceptibility mapping. Finally, the produced map was verified by comparing with existing landslide locations for calculating prediction accuracy. Among the fuzzy operators, in the case in which the gamma operator (λ = 0.975) showed the best accuracy (84.68%) while the case in which the fuzzy or operator was applied showed the worst accuracy (66.50%).     

GIS-based landslide susceptibility evaluation using fuzzy-AHP multi-criteria decision-making techniques in the Abha Watershed,Saudi Arabia

Landslides are natural geological disasters causing massive destructions and loss of lives, as well as severe damage to natural resources, so it is essential to delineate the area that probably will be affected by landslides. Landslide susceptibility mapping (LSM) is making increasing implications for GIS-based spatial analysis in combination with multi-criteria evaluation (MCE) methods. It is considered to be an effective tool to understand natural disasters related to mass movements and carry out an appropriate risk assessment. This study is based on an integrated approach of GIS and statistical modelling including fuzzy analytical hierarchy process (FAHP), weighted linear combination and MCE models. In the modelling process, eleven causative factors include slope aspect, slope, rainfall, geology, geomorphology, distance from lineament, distance from drainage networks, distance from the road, land use/land cover, soil erodibility and vegetation proportion were identified for landslide susceptibility mapping. These factors were identified based on the (1) literature review, (2) the expert knowledge, (3) field observation, (4) geophysical investigation, and (5) multivariate techniques. Initially, analytical hierarchy process linked with the fuzzy set theory is used in pairwise comparisons of LSM criteria for ranking purposes. Thereafter, fuzzy membership functions were carried out to determine the criteria weights used in the development of a landslide susceptibility map. These selected thematic maps were integrated using a weighted linear combination method to create the final landslide susceptibility map. Finally, a validation of the results was carried out using a sensitivity analysis based on receiver operator curves and an overlay method using the landslide inventory map. The study results show that the weighted overlay analysis method using the FAHP and eigenvector method is a reliable technique to map landslide susceptibility areas. The landslide susceptibility areas were classified into five categories, viz. very low susceptibility, low susceptibility, moderate susceptibility, high susceptibility, and very high susceptibility. The very high and high susceptibility zones account for 15.11% area coverage. The results are useful to get an impression of the sustainability of the watershed in terms of landsliding and therefore may help decision makers in future planning and mitigation of landslide impacts.     

Evaluation of potential landslide damming: Case study of Urni landslide,Kinnaur, Satluj valley,India

This work aims to understand the process of potential landslide damming using slope failure mechanism,dam dimension and dam stability evaluation. The Urni landslide, situated on the right bank of the Satluj River, Himachal Pradesh(India) is taken as the case study. The Urni landslide has evolved into a complex landslide in the last two decade(2000-2016) and has dammed the Satluj River partially since year 2013,damaging ～200 m stretch of the National Highway(NH-05). The crown of the landslide exists at an altitude of ～2180-2190 m above msl, close to the Urni village that has a human population of about 500.The high resolution imagery shows ～50 m long landslide scarp and ～100 m long transverse cracks in the detached mass that implies potential for further slope failure movement. Further analysis shows that the landslide has attained an areal increase of 103,900 ± 1142 m^2 during year 2004-2016. About 86% of this areal increase occurred since year 2013. Abrupt increase in the annual mean rainfall is also observed since the year 2013. The extreme rainfall in the June, 2013; 11 June(～100 mm) and 16 June(～115 mm),are considered to be responsible for the slope failure in the Urni landslide that has partially dammed the river. The finite element modelling(FEM) based slope stability analysis revealed the shear strain in the order of 0.0-0.16 with 0.0-0.6 m total displacement in the detachment zone. Further, kinematic analysis indicated planar and wedge failure condition in the jointed rockmass. The debris flow runout simulation of the detached mass in the landslide showed a velocity of ～25 m/s with a flow height of ～15 m while it(debris flow) reaches the valley floor. Finally, it is also estimated that further slope failure may detach as much as 0.80 ±0.32 million m^3 mass that will completely dam the river to a height of 76±30 m above the river bed.     

Modeling of rainfall-triggered shallow landslide

By integrating hydrological modeling with the infinite slope stability analysis, a rainfall-triggered shallow landslide model was developed by Iverson (Water Resour Res 36:1897-1910, 2000). In Iverson’s model, the infiltration capacity is assumed to be equivalent to the saturated hydraulic conductivity for finding pressure heads analytically. However, for general infiltration process, the infiltration capacity should vary with time during the period of rain, and the infiltration rate is significantly related to the variable infiltration capacity. To avoid the unrealistically high pressure heads, Iverson employed the beta-line correction by specifying that the simulated pressure heads cannot exceed those given by the beta line. In this study, the suitability of constant infiltration capacity together with the beta-line correction for hydrological modeling and landslide modeling of hillslope subjected to a rainfall is examined. By amending the boundary condition at ground surface of hillslope in Iverson’s model, the modified Iverson’s model with considering general infiltration process is developed to conduct this examination. The results show that the unrealistically high pressure heads from Iverson’s model occur due to the overestimation of infiltration rate induced from the assumption that the infiltration capacity is identical to the saturated hydraulic conductivity. Considering with the general infiltration process, the modified Iverson’s model gives acceptable results. In addition, even though the beta-line correction is applied, the Iverson’s model still produces greater simulated pressure heads and overestimates soil failure potential as compared with the modified Iverson’s model. Therefore, for assessing rainfall-triggered shallow landslide, the use of constant infiltration capacity together with the beta-line correction needs to be replaced by the consideration of general infiltration process.     

Three-dimensional mapping of a landslide using a multi-geophysical approach: the Quesnel Forks landslide

A landslide located on the Quesnel River in British Columbia, Canada is used as a case study to demonstrate the utility of a multi-geophysical approach to subsurface mapping of unstable slopes. Ground penetrating radar (GPR), direct current (DC) resistivity and seismic reflection and refraction surveys were conducted over the landslide and adjacent terrain. Geophysical data were interpreted based on stratigraphic and geomorphologic observations, including the use of digital terrain models (DTMs), and then integrated into a 3-dimensional model. GPR surveys yielded high-resolution data that were correlated with stratigraphic units to a maximum depth of 25 m. DC electrical resistivity offered limited data on specific units but was effective for resolving stratigraphic relationships between units to a maximum depth of 40 m. Seismic surveys were primarily used to obtain unit boundaries up to a depth of >80 m. Surfaces of rupture and separation were successfully identified by GPR and DC electrical resistivity techniques.     

PROMETHEE II and fuzzy AHP: an enhanced GIS-based landslide susceptibility mapping

This paper describes the application of a well-known multi-criteria decision-making technique, called preference ranking organization method for enrichment evaluation (PROMETHEE II), in combination with fuzzy analytical hierarchy process (FAHP), as a weighting technique to explore landslide susceptibility mapping (LSM). To this end, eight landslide-related geodata layers of the Minoo Dasht located in the Gorgan province of Iran, involving slope, aspect, distance to river, drainage density, distance to fault, mean annual rainfall, distance to road and lithology have been integrated using the PROMETHEE II enhanced by FAHP technique. Afterward, the receiver operating characteristics (ROC) curves for the proposed LSM were drawn using an inventory of landslides containing 83 recent and historic landslide points, and the area under curve = 0.752 value was calculated accordingly. Additionally, to further verify the practicality of such susceptibility map, it was also evaluated against the landslide inventory using simple overlay. The outcome was that about 11 % of the occurred landslide points fall into the very high susceptibility class of the LSM, but approximately 52 % of them indeed fall into the high and very high susceptibility zones together. Also, it resulted that no recorded landslide occurred in the zone of very low susceptibility. According to the results of the ROC curves analysis and simple overlay evaluation, the produced map has exhibited good performance.