Search

Landslides pose a significant risk to human life and property, making landslide susceptibility mapping (LSM) a crucial component of landslide risk assessment. However, spatial correlations among mapping units are often neglected in statistical or machine learning models proposed for LSM. This study proposes KNN-GCN, a deep learning model for slope-unit-based LSM based on a graph convolutional network (GCN) and the K-nearest neighbor (KNN) algorithm. The model was experimentally applied to the Lueyang region and validated through the following steps. Firstly, we collected data for 15 landslide causal factors and from landslide inventories and established a slope unit map (SUM) through slope unit division. Next, we performed a multicollinearity analysis of landslide causal factors and divided the training and test sets at a 7:3 ratio. We then constructed a GCN model based on a slope unit graph (SUG) generated from the SUM using the KNN algorithm. The proposed KNN-GCN model was tuned using a grid search with fivefold cross-validation on the training set, and then trained and validated on training and test sets separately. Finally, the performance of the KNN-GCN model was compared with that of six other models which were categorized into two groups: CG#1 was the traditional KNN, support vector regression (SVC), and automated machine learning (AutoML), and CG#2 included KNN-G, SVC-G and AutoML-G with additional spatial information. Our results demonstrate that the proposed model achieves superior performance (area under the curve [AUC] = 0.8351) and generates the most comprehensible susceptibility map with distinct boundaries between different susceptibility levels. Notably, while the proposed KNN-GCN model displays exceptional performance in slope-unit-based LSM, its implementation requires high-level computing resources, and it is not recommended for small datasets. [ABSTRACT FROM AUTHOR]

Timely evaluation of landslide risk based on displacement data is crucial for the effective warning of landslide events. However, the complexity and diversity of displacement curves obtained in the field make the selection of evaluation criteria and thresholds challenging and subjective. Thus, this study proposes a new early warning criterion called the deformation probability index (DPI) to overcome this limitation. By analyzing features of landslide displacement curves obtained from the field, four linetypes were identified, and differences in velocity data between pre-failure and post-failure landslides were summarized from a probabilistic perspective. Subsequently, the DPI is defined to assess the early warning level of a landslide by converting observed velocity values into standardized velocity values using cumulative distribution function standardization. The resulting DPI values are divided into five early warning intervals with three alerting levels to indicate the relative stability of the landslide. Applying the proposed method to 16 landslides from around the world demonstrates the effectiveness of the DPI criterion in evaluating landslide risk, with a high correlation between alerting levels and observed landslide occurrences. The DPI criterion demonstrates notable sensitivity exclusively to displacement velocity and monitoring frequency, exhibiting remarkable generality and accuracy, thus offering a new method for landslide early warning. [ABSTRACT FROM AUTHOR]

Analyzing and quantifying the deformation process of landslides is of paramount importance in facilitating landslide early warning. As such, this study is committed to proposing a universal phenomenological model for deformation stages division and early warning of landslides based on the kinematic features. First, five landslide deformation patterns were classified based on the creep theory, and suggestions for stage division of each deformation pattern are presented. Then, the statistical characteristics of landslide velocity were analyzed, and a probability-based deformation stage division method was proposed. Finally, the Comprehensive Standardized Deformation Index (CSDI) model, which includes the calculation of the CSDI M - M (Min-Max normalization) and CSDI M (Mean normalization) was proposed and verified in 24 landslides worldwide. The results show that, except for the oscillating pattern, the CSDI M - M is feasible in the stages division of all deformation patterns with a strong correspondence with the actual state of the landslides. The CSDI M is a reliable landslide warning criterion and threshold determination method, as it is effective in the early warning of imminent landslides with a low false alarm rate. The CSDI model provides new insight into the division of landslide deformation stages and landslide risk assessment. [ABSTRACT FROM AUTHOR]

Landslides frequently disrupt roadway networks in mountainous regions worldwide. Because of the relatively long roadway extension and low roadway density in mountainous regions, the occurrence of a landslide hazard along a local road segment will cause traffic paralysis on the individual roadway and will further impact regional roadway network accessibility. This paper establishes an integrative risk assessment framework based on risk theory and complex network theory to combine the results of landslide susceptibility mapping along roadways and impact assessment on the roadway network. Through an analysis of the relationship between various geo-environmental conditioning factors and historical landslides along roadways, the support vector machine (SVM) model is used to assess landslide susceptibility across the regional roadway network. Both topological connectivity of the entire roadway network and transport accessibility between local residents are considered in the impact assessment on roadway networks. To illustrate the effectiveness of the proposed risk assessment framework, a case study of the roadway network in Fengjie County, China, which is prone to landslide occurrence, is conducted. The resulting landslide risk heatmap of Fengjie County's roadway network is generated using ArcGIS software. The most critical road segments are identified as being highly susceptible to landslides, and if they are disrupted, the entire roadway network will suffer significant performance loss. The results can support adaptive strategies for landslide mitigation, preparedness, and emergency response services, as well as improve roadway plans to reduce exposure and associated consequences by adding new links to the existing roadway network. [ABSTRACT FROM AUTHOR]

A landslide susceptibility model based on a metaheuristic optimization algorithm (germinal center optimization (GCO)) and support vector classification (SVC) is proposed and applied to landslide susceptibility mapping in the Three Gorges Reservoir area in this paper. The proposed GCO-SVC model was constructed via the following steps: First, data on 11 influencing factors and 292 landslide polygons were collected to establish the spatial database. Then, after the influencing factors were subjected to multicollinearity analysis, the data were randomly divided into training and testing sets at a ratio of 7:3. Next, the SVC model with 5-fold cross-validation was optimized by hyperparameter space search using GCO to obtain the optimal hyperparameters, and then the best model was constructed based on the optimal hyperparameters and training set. Finally, the best model acquired by GCO-SVC was applied for landslide susceptibility mapping (LSM), and its performance was compared with that of 6 popular models. The proposed GCO-SVC model achieved better performance (0.9425) than the genetic algorithm support vector classification (GA-SVC; 0.9371), grid search optimized support vector classification (GRID-SVC; 0.9198), random forest (RF; 0.9085), artificial neural network (ANN; 0.9075), K-nearest neighbor (KNN; 0.8976), and decision tree (DT; 0.8914) models in terms of the area under the receiver operating characteristic curve (AUC), and the trends of the other metrics were consistent with that of the AUC. Therefore, the proposed GCO-SVC model has some advantages in LSM and may be worth promoting for wide use. [ABSTRACT FROM AUTHOR]

The evaluation of landslide runout behavior plays a vital role in the risk assessment of landslides. In this study, the runout behavior of a landslide that occurred on December 10, 2019, in the Three Gorges Reservoir region, China, known as the Kamenziwan landslide, is studied with the discrete element method. A field survey is first conducted to study the geological characteristics and failure mechanism of this landslide. The survey results indicate that this landslide is possibly caused by the long-term fluctuation of the reservoir water level. Then, a three-dimensional (3-D) numerical model of this landslide is constructed on the basis of the terrain data obtained before and after the landslide occurrence. The critical input parameters to the numerical model, in terms of the contact micro-parameters among particles, are calibrated with the machine learning method and uniaxial compression tests. To depict the effectiveness of the built numerical model, the simulation results are compared with the video snapshots taken by a local resident. Finally, the runout behavior and the energy transformation of the landslide are analyzed based on the validated numerical model. [ABSTRACT FROM AUTHOR]

Landslide is one of the most destructive geohazards around the world. The destruction of a landslide can be estimated from its deformation and runout behaviors, which might be simulated with numerical software such as particle flow code (PFC). In the PFC simulation of the runout behavior of a landslide, the results are dependent upon the particle‐particle contact micro‐parameters (e.g., contact modulus and friction coefficient). The calibration of the micro‐parameters of the PFC models is a challenge and various uncertainties are involved. The uncertainty in the selection of the modelling parameters leads to uncertainty in the simulated runout behavior of the landslide. As a result, the damage potential of the landslide cannot be accurately evaluated. To this end, this paper presents a probabilistic analysis of the runout behavior of the Jiweishan landslide, in which the uncertainty in the selection of the particle‐particle contact micro‐parameters is explicitly considered. Here, the input micro‐parameters are modeled as discrete random variables, the possible realizations of which are obtained through an orthogonal analysis of the PFC simulation‐based uniaxial compression tests. The derived possible realizations of the micro‐parameters are then adopted as the inputs to the built model of the landslide. With the results obtained from these PFC simulations, the runout behaviors of the Jiweishan landslide, such as the runout distance and the deposit thickness, are studied probabilistically. The results provide an improved estimate of the landslide runout behavior and further aid in making an informed risk assessment of the landslide. [ABSTRACT FROM AUTHOR]

Featured Application: The proposed decomposition-ensemble learning model can be efficiently used to enhance the prediction accuracy of landslide displacement prediction and can also be extended to other difficult forecasting tasks in the geosciences with extremely complex nonlinear data characteristics. As vital comments on landslide early warning systems, accurate and reliable displacement prediction is essential and of significant importance for landslide mitigation. However, obtaining the desired prediction accuracy remains highly difficult and challenging due to the complex nonlinear characteristics of landslide monitoring data. Based on the principle of "decomposition and ensemble", a three-step decomposition-ensemble learning model integrating ensemble empirical mode decomposition (EEMD) and a recurrent neural network (RNN) was proposed for landslide displacement prediction. EEMD and kurtosis criteria were first applied for data decomposition and construction of trend and periodic components. Second, a polynomial regression model and RNN with maximal information coefficient (MIC)-based input variable selection were implemented for individual prediction of trend and periodic components independently. Finally, the predictions of trend and periodic components were aggregated into a final ensemble prediction. The experimental results from the Muyubao landslide demonstrate that the proposed EEMD-RNN decomposition-ensemble learning model is capable of increasing prediction accuracy and outperforms the traditional decomposition-ensemble learning models (including EEMD-support vector machine, and EEMD-extreme learning machine). Moreover, compared with standard RNN, the gated recurrent unit (GRU)-and long short-term memory (LSTM)-based models perform better in predicting accuracy. The EEMD-RNN decomposition-ensemble learning model is promising for landslide displacement prediction. [ABSTRACT FROM AUTHOR]

The mechanical properties of stabilizing piles are directly affected by the spatial structure of the bedrock of landslides. To effectively reinforce landslides with a sandwich-type bedrock, the interaction mechanism between manually excavated stabilizing piles and the different types of bedrock must be investigated. The research presents experimental models of manually excavated stabilizing piles for two kinds of sandwich-type bedrock structures, namely hard rock–dominant interlayered bedrock (HDIB) and weak rock–dominant interlayered bedrock (WDIB). Six physical models with different bedrock structure types and dip angles θ were prepared to explore the influence of these two parameters on the pile-bedrock interaction mechanism. The mechanical properties of the pile were investigated based on the monitoring results of the earth pressure distributions against the piles and the bending moment distributions. The embedded section of the pile experiences a rotation subjected to lateral soil movement. The center point of the rotation in the HDIB is at a greater elevation than in the WDIB. The earth pressure in the HDIB appears to increase with dip angle θ but is similar to each other in the WDIB. The pile embedded in the HDIB exhibits flexible behavior, whereas the pile embedded in the WDIB exhibits rigid behavior. With the increase of dip angle θ in the WDIB material, the bending moment steadily rises followed by fluctuation. The maximum bending moment of the embedded section increases with θ. These results provide a basis for improving the design of manually excavated stabilizing piles in sandwich-type bedrock. [ABSTRACT FROM AUTHOR]

A rapid and long-traveling landslide was triggered by a monsoon rainfall of a total of 446.5 mm for 3 days in the Aranayake area, Sri Lanka. The landslide destroyed 75 houses and killed 127 persons along the course of this landslide. The mechanism of this landslide was investigated by a joint team of the International Consortium on Landslides (ICL) and the National Building Research Organization (NBRO) of the Government of Sri Lanka in 2017 and 2018; the investigation was part of the WCoE and IPL projects. Physical properties of soil samples taken from the source area of the Aranayake landslide were examined through a series of undrained ring-shear tests, and the measured properties were used in a numerical method, LS-RAPID, to simulate initiations and motions of landslides in this area. The authors then tried to first reproduce a landslide, which was considered, from available pieces of evidence, to have most likely happened in the past, then, to assess the risk of a potential future landslide that is expected to occur close to the 2016 Aranayake landslide. Two possible sliding surfaces were assumed to develop through the soil mass, one along the boundary between the weathered surface soil and the jointed bedrock and another along a potential aquiclude inside the soil. This attempt to estimate movements of both past and future landslides based on the numerical reproduction of the 2016 Aranayake landslide hazard will help develop landslide hazard assessment technologies. [ABSTRACT FROM AUTHOR]

Data from in situ monitoring of displacement is of fundamental importance, not only in hazard assessments, but also for understanding the deformation mechanisms of landslides. However, acquiring reliable and continuous displacement data from complex landslides is often a problem because of the high cost of instruments and maintenance. Among more than 250 dangerous landslides monitored in the Chinese Three Gorges reservoir area, the most systematic and complete monitoring data have been obtained from the Shuping landslide. A ten-year record of GPS displacements measured in eight different locations, plus rainfall and reservoir level data, are utilised to identify the movement characteristics and causal factors of the Shuping landslide. The cumulative displacement curves show a series of abrupt increases that occur every year and are separated by periods of steady, slower deformations. The maximum cumulative displacement that occurred during the ten-year monitoring period was 4.7 m at site SP-2. The maximum displacement rate was 35 mm/day, which was observed on June 13, 2012 at site ZG86. Causal factors of the Shuping landslide are identified through qualitative comparisons of the monitoring data and cross-examined by grey correlation degree analysis. The results show that reservoir water level drawdown is the main factor accelerating the movement of the Shuping landslide, with heavy rainfall playing a subordinate role. The proposed use of field data and analytical methods has the potential for providing procedures and solutions for the interpretation of landslide monitoring data. [ABSTRACT FROM AUTHOR]

The Huangtupo riverside slope, which extends an area of 1.35 × 10 m and has a total volume of 6.934 × 10 m, is one of the most noted large-scale landslides with great concern about the safety of immigrants in the Three Gorges reservoir area of China. During the normal operation, the reservoir water level fluctuates periodically. Water level fluctuation will soften the strength of the geologic materials on the banks, induce underground water fluctuation, and in turn affect the slope stability. On the basis of the thorough engineering geological survey, the analysis of engineering geological conditions and deformation mechanism is conducted primarily. Then, the Huangtupo landslide is divided into four minor landslides, all of which have typical three-dimensional characteristics. To these landslides, a three-dimensional rigorous limit equilibrium method is proposed, which can accommodate failure surfaces of any shape, and satisfies all the equilibrium conditions. In terms of the strength parameters of geologic materials, the stability evolution of the four minor landslides at the Huangtupo landslide under periodical water level fluctuation is investigated in detail. From the results, the 3D safety factors turned out to be somewhat bigger than the 2D results even though they displayed the same trends. The differences between 2D and 3D results are about 2.92-6.11 %. The maximum reduction in safety factors with the rapid down of water level is 0.11. It is suggested that the safety margin of the two minor landslides near the riverside is insufficient and might become instable under periodical water level fluctuation. [ABSTRACT FROM AUTHOR]

Complexity of analysis of landslide hazard is due to uncertainty. In this study, a novel approach multi-gene genetic programming based on separable functional network (MGGPSFN) is presented for predicting landslide displacement. Moreover, Pearson's cross-correlation coefficients and mutual information are adopted to look for the potential input variables for a forecast model in the paper. The performance of new model is verified through one case study in Baishuihe landslide in the Three Gorges Reservoir in China. In addition, we compared it with two methods, back-propagation neural network and radial basis function, and MGGPSFN got the best results in the same measurements. [ABSTRACT FROM AUTHOR]

Landslide hazard is a complex nonlinear dynamical system with uncertainty. The evolution of landslide is influenced by many factors such as tectonic, rainfall and reservoir level fluctuation. Using a time series model, total accumulative displacement of landslide can be divided into the trend component displacement and the periodic component displacement according to the response relation between dynamic changes in landslide displacement and inducing factors. In this paper, a novel neural network technique called ensemble of extreme learning machine (E-ELM) is proposed to investigate the interactions of different inducing factors affecting the evolution of landslide. Grey relational analysis is used to sieve out the more influential inducing factors as the inputs in E-ELM. Trend component displacement and periodic component displacement are forecasted, respectively; then, total predictive displacement is obtained by adding the calculated predictive displacement value of each sub. Performances of our model are evaluated by using real data from Baishuihe landslide in the Three Gorges Reservoir of China, and it provides a good representation of the measured slide displacement behavior. [ABSTRACT FROM AUTHOR]

In the Three Gorges Reservoir area (TGRA), landslides typically exhibit slow-moving characteristic deformations in response to seasonal changes in precipitation and reservoir level. This study proposed an optimal combination of methods in which the frequency components of decomposed inducing factors are considered for predicting landslide displacement. In preprocessing, the monitored surface displacements are decomposed into trend displacement and periodic displacement components via ensemble empirical mode decomposition (EEMD); the time series of the inducing factors (rainfall and reservoir level) were reconstructed and decomposed into high- and low-frequency time series via complete ensemble empirical mode decomposition (CEEMD) and t -test methods, among which two dominant inducing factors are identified by analyzing the longest common subsequence (LCSS) distance of the normalized data. In the prediction of the periodic displacement component, three optimal methods that use the support vector regression (SVR) model have been considered for comparison: the artificial bee colony (ABC) algorithm, the genetic algorithm (GA) and particle swarm optimization (PSO). The results demonstrate that the combination of CEEMD-LCSS reconstruction and ABC-SVR yields a robust model with remarkable prediction performance, which can increase the prediction accuracy of landslide displacements by 54% at most in the evaluation of RMSE. [ABSTRACT FROM AUTHOR]

Landslide displacement monitoring plays a fundamental role in the study of landslide evolution mechanisms, forecasting, risk assessment, prevention, and control. To fill the deficiencies of traditional instrumentation for measuring landslide displacement distributed along lateral direction, a landslide displacement measurement method based on deformation-coupled pipeline trajectory measurement is proposed, and a pipeline trajectory inertial measurement instrument is developed. The developed instrument, primarily comprised of a single shaft gyro, two axis accelerometers, and an external roller encoder, is designed as an axial half strapdown-radial half platform structure combined with a mechanical gravity platform. This structure avoids the singularity of pitch angle and roll angle and can expediently calculate a pipeline trajectory with an Eulerian transformation when obtaining several basic physical variables, e.g., the axial linear velocity, pitch angle, roll angle, and azimuth angle. Additionally, the pipeline trajectory, measured at different times, possesses the ability to reflect the displacement evolution feature of landslides. The results of prototype simulation tests imply a single measurement accuracy of a 12 cm/100 m span and a singly periodic multiple (more than five times) measurement accuracy of a 3 cm/100 m span, which meets medium-precision displacement measurement requirements for a landslide. Additionally, the finished instrument has been successfully applied to the deformation monitoring of the Majiagou I# landslide, which further verifies its feasibility and offers a reference for similar landslides. [ABSTRACT FROM AUTHOR]

Accurately predicting the surface displacement of the landslide is important and necessary. However, most of the existing research has ignored the frequency component of inducing factors and how it affects the landslide deformation. Therefore, a hybrid displacement prediction model based on time series theory and various intelligent algorithms was proposed in this paper to study the effect of frequency components. Firstly, the monitoring displacement of landslide from the Three Gorges Reservoir area (TGRA) was decomposed into the trend and periodic components by complete ensemble empirical mode decomposition (CEEMD). The trend component can be predicted by the least square method. Then, time series of inducing factors like rainfall and reservoir level was reconstructed into high frequency components and low frequency components with CEEMD and t-test, respectively. The dominant factors were selected by the method of dynamic time warping (DTW) from the frequency components and other common factors (e.g., current monthly rainfall). Finally, the ant colony optimization-based support vector machine regression (ACO-SVR) is utilized for prediction purposes in the TGRA. The results demonstrate that after considering the frequency components of landslide-induced factors, the accuracy of the displacement prediction model based on ACO-SVR is better than that of other models based on SVR and GA-SVR. [ABSTRACT FROM AUTHOR]

Landslides cause damages to land and infrastructure and pose serious threat to human survival. To prepare a landslide susceptibility map of the region in Longnan, City Gansu Province, with a 330-kV transmission line, 10 parameters were selected by correlation analysis and sensitivity analysis from initial 18 and five different methods were used, including analytical hierarchy process (AHP), information value (IV), fractal theory (FT), back propagation neural network (BPNN), support vector machine (SVM). The susceptibility maps were validated through receiver operating characteristic (ROC) and cumulative landslides percentage curves based on 77 existing landslide events. The results indicate that BPNN and SVM model are most accurate, time-saving and easily implemented. All of the five methods accurately predict the spatial distribution of landslides and can be well applied to landslide susceptibility mapping. What needs to be emphasized is that the machine learning methods have the advantages of high efficiency, accurate prediction, time-saving, convenient implementation, which are relatively new and better evaluation models of susceptibility. [ABSTRACT FROM AUTHOR]

The mechanical behavior of in situ sliding zone soil is unique and is significant for the state identification and prediction of landslides. This study focuses on interpreting the anisotropic deformation observed on the sliding zone soil of the Huangtupo landslide during an in situ triaxial creep test; specifically, the deformation of the A-Surface (towards 35° dip direction) of the soil cube, which points approximately to the landslide sliding orientation, is significantly large. Such anisotropy could be attributed to damaged structure, assumed as pre-existing directional cracks, of the in situ soil. Therefore, a Discrete Element Method (DEM) model containing prefabricated cracks is established to investigate the deforming behavior of compressed soil during in situ creep test on the mesoscopic scale. Results indicated that the specimen's anisotropic deformation is intimately related to the presence of prefabricated cracks. Subjected to the prefabricated cracks, microcracks generated during the loading process of the specimen initiated and developed in a directional way, subsequently forming directed cracks. Then the directed cracks tend to form a complete slip zone. However, due to the insufficient vertical load, the directed cracks were unable to coalesce, and the soil mass did not fail during the in situ creep test. Thereafter, additional load was applied in the simulation model. The directed cracks then merged and a full slip zone formed, and the soil was completely destroyed. As a consequence, the anisotropic characteristics of compressed in situ sliding zone soil can be well explained in view of the cracks development. This study may help perceive the landslide deformation characteristics and facilitate the evaluation and prediction methods of landslide stability. • Anisotropic deformation of the in situ sliding zone during compression • Formation of cracked slip zone in a compressed soil sample • Brand new recognition on the in situ sliding zone soil with prefabricated cracks [ABSTRACT FROM AUTHOR]

Due to the landslide-prone geological conditions, many landslides have occurred in the Three Gorges Reservoir Region (TGRR), most of which are slow-moving ancient landslides characterized by preexisting shear surfaces. In this study, with the purpose of making full use of reactivated residual strength to prevent and treat the slow-moving landslides, such as designing anti-slide piles, the Huangtupo landslide as a typical reactivated ancient landslide in the TGRR was selected to study the overconsolidation (OCR) effect on the reactivated residual strength of deep-seated sliding zone soil by conducting a series of laboratory ring shear tests. It was found that after a short rest of one day, the reactivated residual strength increased with the increase of OCR under a given normal effective stress, but it was lost after a small shear displacement. The overconsolidation effect on reactivated residual strength at a lower stress was more prominent than that at a higher stress. Very coarse sand with a size of 1–2 mm inside the remoulded deep-seated sliding zone soil sample exerts direct control on the reactivated residual strength, and it significantly influences the overconsolidation effect on reactivated residual strength. The results could provide a reliable scientific basis for improving the stability analysis and reinforcement measures of the slow-moving landslides in the TGRR, as well as can be applied to similar slow-moving landslides in other areas. • The reactivated residual strength increased with the increase of OCR after a short rest of one day. • The overconsolidation effect at a lower stress was more prominent than that at a higher stress. • Very coarse sand inside the soil sample significantly influences the overconsolidation effect. [ABSTRACT FROM AUTHOR]

Unmanned Aerial Vehicle (UAV) technique has been widely utilized in geohazards assessment. In this work, the effectiveness of the UAV photogrammetry in the remote sensing and assessment of the landslide behavior is demonstrated through a case study of a landslide that occurred in Guizhou, China on 10 June 2018. The post-landslide assessments were conducted through a field investigation by a team of experts; and, two UAV photogrammetry-based surveys were carried out. On the basis of a detailed inspection of the high-resolution aerial photographs collected from the UAV photogrammetry and historical satellite imagery, subsurface stratigraphic configuration revealed from borehole explorations and the local rainfall data collected, the failure mechanism of this landslide is investigated. The occurrence of this landslide is probably attributed to the combined influence of the long-term rainfall and the engineering activities at this site. An automatic landslide cracks recognition model, which is based on the deep learning-based image recognition technique of RetinaNet, is further developed to map the cracks at the landslide site. The effectiveness of this automatic crack recognition model is validated through quantitative comparisons between the landslide cracks recognized and the field survey results. Based upon the landslide cracks identified on two different dates after this landslide event (by the field survey and automatic crack recognition model) and the subsurface displacement revealed from a drilled hole, the evolution behavior of this landslide is analyzed. The results show that the stability of this landslide was not achieved during the first slide in June 2018 and the limit of the landslide was increased much from June to November 2018. In such a situation, the elements at risk in the zones that are potentially impacted by this landslide are identified (with the aid of the UAV images collected), and a preliminary consequence assessment is conducted. • The failure mechanism of a landslide in Guizhou, China is investigated. • An automatic landslide cracks recognition model is developed using RetinaNet. • The landslide behavior is studied based on the cracks recognized on two dates. • A UAV images-based preliminary assessment of the landslide consequence is made. [ABSTRACT FROM AUTHOR]

Interval prediction is an efficient approach to quantifying the uncertainties associated with landslide evolution. In this paper, a novel method, termed lower upper bound estimation (LUBE), of constructing prediction intervals (PIs) based on neural networks (NNs) is applied and extended to landslide displacement prediction. A random vector functional link network (RVFLN) is adopted as the NN used in the improved LUBE. A hybrid evolutionary algorithm, termed PSOGSA, that combines particle swarm optimization (PSO) and gravitational search algorithm (GSA) is utilized to train LUBE. The loss function of LUBE is redesigned by considering the quality of PI centre, which allows for a more comprehensive evaluation of PIs. The population initialization in the training process of LUBE is implemented by transferring the weights of a series of pre-trained RVFLNs. The performance of the improved LUBE method is validated by considering a comprehensive set of cases using seven benchmark datasets. In addition, a hybrid method that integrates ensemble empirical mode decomposition (EEMD) with the improved LUBE is proposed for the special case of landslide displacement prediction. Six real-world reservoir-induced landslides are considered to validate the capability and merit of the proposed hybrid method. [ABSTRACT FROM AUTHOR]