Search

(1) Background: The aim of this paper was to study landslide susceptibility mapping based on interpretable machine learning from the perspective of topography differentiation. (2) Methods: This paper selects three counties (Chengkou, Wushan and Wuxi counties) in northeastern Chongqing, delineated as the corrosion layered high and middle mountain region (Zone I), and three counties (Wulong, Pengshui and Shizhu counties) in southeastern Chongqing, delineated as the middle mountainous region of strong karst gorges (Zone II), as the study area. This study used a Bayesian optimization algorithm to optimize the parameters of the LightGBM and XGBoost models and construct evaluation models for each of the two regions. The model with high accuracy was selected according to the accuracy of the evaluation indicators in order to establish the landslide susceptibility mapping. The SHAP algorithm was then used to explore the landslide formation mechanisms of different landforms from both a global and local perspective. (3) Results: The AUC values for the test set in the LightGBM mode for Zones I and II are 0.8525 and 0.8859, respectively, and those for the test set in the XGBoost model are 0.8214 and 0.8375, respectively. This shows that LightGBM has a high prediction accuracy with regard to both landforms. Under the two different landform types, the elevation, land use, incision depth, distance from road and the average annual rainfall were the common dominant factors contributing most to decision making at both sites; the distance from a fault and the distance from the river have different degrees of influence under different landform types. (4) Conclusions: the optimized LightGBM-SHAP model is suitable for the analysis of landslide susceptibility in two types of landscapes, namely the corrosion layered high and middle mountain region, and the middle mountainous region of strong karst gorges, and can be used to explore the internal decision-making mechanism of the model at both the global and local levels, which makes the landslide susceptibility prediction results more realistic and transparent. This is beneficial to the selection of a landslide susceptibility index system and the early prevention and control of landslide hazards, and can provide a reference for the prediction of potential landslide hazard-prone areas and interpretable machine learning research. [ABSTRACT FROM AUTHOR]

Landslide deformation is the most intuitive and effective characterization of the evolution of landslides and reveals the inherent risk of landslides. Considering the inadequacy of existing deformation monitoring data for early warnings regarding landslide hazards, resulting in insufficient disaster response times, this paper proposes a time-domain correlation model. Based on the process of rainfall-induced landslide deformation, the time-domain correlation between regional rainfall and landslide deformation is proposed, which can reflect the temporal characteristics of landslide responses to rainfall, and the calculation method of the impulse response function is designed to quantitatively model and calculate the correlation. Furthermore, rainfall monitoring data are used to optimize the landslide deformation monitoring indicator system for early warnings regarding landslide instability. The feasibility of the method proposed in this paper is verified by analyzing the historical monitoring data of rainfall and landslide deformation at nine typical locations in five landslide hazard areas in Fengjie County, Chongqing city. (1) The correlation models for the XP landslide involve a delayed rainfall response time of 5 for deformation, respectively, as well as the existence of a cycle of 55–56 days, which means that the above area can advance the landslide warning by one lag time based on the cycle; (2) The correlation models for the OT landslide show consistent correlations under a 48–50-day cycle, which means that the deformation in the above areas can be predicted based on rainfall accumulation. (3) The HJWC landslide presents a turbulence correlation, which means that other monitoring data need to be supplemented and analyzed. [ABSTRACT FROM AUTHOR]

A calamitous landslide happened at 22:00 on September 1, 2014 in the Yunyang area of Chongqing City, southwest China, enforcing the evacuation of 508 people and damaging 23 buildings. The landslide volume comprised 1.44 million m3 of material in the source area and 0.4 million m3 of shoveled material. The debris flow runout extended 400 m vertically and 1600 m horizontally. The Xianchi reservoir landslide event has been investigated as follows: (1) samples collected from the main body of landslide were carried out using GCTS ring shear apparatus; (2) the parameters of shear and pore water pressure have been measured; and (3) the post‐failure characteristics of landslide have been analyzed using the numerical simulation method. The excess pore‐water pressure and erosion in the motion path are considered to be the key reasons for the long‐runout motion and the scale‐up of landslides, such as that at Xianchi, were caused by the heavy rainfall. The aim of this paper is to acquired numerical parameters and the basic resistance model, which is beneficial to improve simulation accuracy for hazard assessment for similar to potentially dangerous hillslopes in China and elsewhere. [ABSTRACT FROM AUTHOR]

Landslide susceptibility assessment (LSA) based on machine learning methods has been widely used in landslide geological hazard management and research. However, the problem of sample imbalance in landslide susceptibility assessment, where landslide samples tend to be much smaller than non-landslide samples, is often overlooked. This problem is often one of the important factors affecting the performance of landslide susceptibility models. In this paper, we take the Wanzhou district of Chongqing city as an example, where the total number of data sets is more than 580,000 and the ratio of positive to negative samples is 1:19. We oversample or undersample the unbalanced landslide samples to make them balanced, and then compare the performance of machine learning models with different sampling strategies. Three classic machine learning algorithms, logistic regression, random forest and LightGBM, are used for LSA modeling. The results show that the model trained directly using the unbalanced sample dataset performs the worst, showing an extremely low recall rate, indicating that its predictive ability for landslide samples is extremely low and cannot be applied in practice. Compared with the original dataset, the sample set optimized through certain methods has demonstrated improved predictive performance across various classifiers, manifested in the improvement of AUC value and recall rate. The best model was the random forest model using over-sampling (O_RF) (AUC = 0.932). [ABSTRACT FROM AUTHOR]

On October 2017, due to continuous rainfall, there was a massive river blockage induced by a landslide in Guang'an Village, Chongqing, China. Long-term monitoring analysis has shown that there remain four strong deformation areas on the slope, which seriously threaten the life and property of nearby residents. In this paper, a granular flow model and an elasto-visco-plasticity model were applied to reproduce and predict the landslide event that hit Guang'an Village. The results showed that the landslide gradually moved along the sliding surface, pushing loose deposits and blocking the Xixi River. The numerical reproduction results of the 2017 event are consistent with the actual slope deformation and failure process and deposit morphology. The simulated maximum depth-averaged velocity of this landslide was approximately 1.89 m/s, and the height of the landslide dam was approximately 10 m. After the landslide occurred in 2017, several large deformation areas appeared in the vicinity of the sliding area, and the right rear side of the sliding mass in area III has the largest deformation volume, accompanied by the most developed surface crack and the most intense deformation. There is a risk that the Xixi river will be blocked again. Therefore, with the same parameter and numerical model, a sliding–pushing–blocking dynamic prediction analysis of the strong deformation area III was conducted. The pushing motion of the mass in this area will reactivate the landslide mass observed in 2017. The maximum depth-averaged velocity of deformation area III was 0.5 m/s, and the maximum depth-averaged velocity of landslide deposition was 0.45 m/s. The length of the blocking dam formed by the mass of deformation area III along river was approximately 780 m, 30 m longer than that in 2017. The predicted height of the landslide dam was 14.5 m, approximately 4.5 m higher than that in 2017. The length of the landslide dam reservoir was predicted to be 2.55 km along the Xixi River, which may submerge the Waping Village. This study supports the landslide hazard prevention, reveals the whole movement process of sliding-pushing-blocking, and provides a new research idea and method for the landslide movement prediction. Hence, this study can serve as a reference for the hazard prevention and mitigation of such chain disasters. [ABSTRACT FROM AUTHOR]

China is a country with frequent geological disasters. The geological disasters in different regions are different, such as earthquakes, landslides, collapses, karst collapses, and ground subsidence. With the development of science and technology and the improvement of engineering technology, the combination of image technology and digital technology is constantly developing. People can improve the accuracy of prediction by studying the characteristics of rock formation under different geological conditions while collecting geological data and sorting various information. And reliability, can more accurately identify poor geological conditions, can increase construction safety and reduce economic losses. Based on the comprehensive CT geological survey data, this paper uses nanomaterials to construct software DEM to model the porous medium of the Yanziyan dangerous rock mass in Jinfo Mountain, Nanchuan District, Chongqing and to estimate the area and volume of the collapse potential area of the high-steep dangerous rock mass parameter. [ABSTRACT FROM AUTHOR]

Landslide susceptibility mapping (LSM) is critical for risk assessment and mitigation. Generalization ability and prediction uncertainty are the current challenges for LSM but have been rarely investigated. The generalization ability refers to the ability of trained models to assess the landslide susceptibility of new areas and make accurate predictions. The prediction uncertainty mainly comes from the possibility of wrongly selecting the unstable landslide samples as stable ones from incomplete landslide inventory. This paper proposes a hybrid model by integrating the convolutional neural network (CNN) with physical model transient rainfall infiltration and grid-based regional slope-stability analysis (TRIGRS) to address the challenges above by combining the advantages of the two approaches. CNN is the main structure of the hybrid model and serves as a binary classifier to capture the spatial and inter-channel correlation among landslide conditioning factors and landslide inventory. TRIGRS characterizes the differences among grids caused by lithology by converting originally spatially discrete and banded lithology information into spatially continuous safety factors (Fs) within a fixed range and pre-selects training samples to ensure the correctness of the selected non-landslide grids. Two towns (Zhuyuan and Qinglian) in Fengjie, Chongqing, China, are used as the study area. A landslide inventory and landslide conditioning factor maps with 30 m resolution consist of the database. The performance of CNN and the proposed hybrid model is compared using the receiver operating characteristic curve and relative landslide density index (R-index). The superiority of the hybrid model and the effect of pre-selection of training samples are investigated. The results reveal that the generalization ability is enhanced and the prediction uncertainty is reduced by the proposed hybrid model. [ABSTRACT FROM AUTHOR]

Long-runout landslides occur frequently in the sandstone and mudstone mountainous areas in southwestern China under heavy rainfall conditions. This has been a key issue in the field of disaster prevention and reduction. Considering the Niuerwan landslide in Wulong, Chongqing, on 13 July 2020, as an example, we employed technical methodologies, including unmanned aerial vehicle (UAV) images, field investigation, geological condition analysis (including geomorphology and topography, stratigraphic structure and formation lithology, etc.), interferometric synthetic aperture radar (InSAR) monitoring and Particle Flow Code 3D (PFC3D) simulations to study failure mechanism and a long-runout motion model of flow-like landslides induced by the heavy rainfall. The results showed that (1) the large differences between the upper and lower strata are the root cause of the instability and long-runout fluidization movement; (2) heavy rainfall is the key driving factor of slope instability and deep-seated landslides, leading to long-distance movement of the upper saturated residual soil; (3) the long-runout fluidization model of bedding landslides is mainly divided into the overall sliding in the lower layer, the mixing of coarse and fine particles in the middle layer, and saturation fluidization in the upper layer; and (4) the long-runout fluidization process of bedding landslides is composed of three stages: overall instability, mixed acceleration, and fluidization accumulation. In view of these findings, in the risk evaluation and prediction of long-runout fluidization landslides in sandstone and mudstone mountainous areas, this particular disaster model can be used to provide quantitative references for disaster prevention and mitigation. [ABSTRACT FROM AUTHOR]

Landslides, the second largest geological hazard after earthquakes, result in significant loss of life and property. Extracting landslide information quickly and accurately is the basis of landslide disaster prevention. Fengjie County, Chongqing, China, is a typical landslide-prone area in the Three Gorges Reservoir Area. In this study, we newly integrate Shapley Additive Explanation (SHAP) and Optuna (OPT) hyperparameter tuning into four basic machine learning algorithms: Gradient Boosting Decision Tree (GBDT), Extreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), and Additive Boosting (AdaBoost). We construct four new models (SHAP-OPT-GBDT, SHAP-OPT-XGBoost, SHAP-OPT-LightGBM, and SHAP-OPT-AdaBoost) and apply the four new models to landslide extraction for the first time. Firstly, high-resolution remote sensing images were preprocessed, landslide and non-landslide samples were constructed, and an initial feature set with 48 features was built. Secondly, SHAP was used to select features with significant contributions, and the important features were selected. Finally, Optuna, the Bayesian optimization technique, was utilized to automatically select the basic models' best hyperparameters. The experimental results show that the accuracy (ACC) of these four SHAP-OPT models was above 92% and the training time was less than 1.3 s using mediocre computational hardware. Furthermore, SHAP-OPT-XGBoost achieved the highest accuracy (96.26%). Landslide distribution information in Fengjie County from 2013 to 2020 can be extracted by SHAP-OPT-XGBoost accurately and quickly. [ABSTRACT FROM AUTHOR]

Physically-based landslide susceptibility models have been widely used to predict rainfall-triggered landslides. Calibration of these models is mainly conducted by tuning various inputs, but spatially varying soil properties are not considered. In this study, an efficient probabilistic model calibration method is proposed, which integrates discrete cosine transform (DCT) representation of spatial variability with Bayesian parameter estimation to characterise the spatial variation of soil properties in a physically-based landslide susceptibility model. The efficacy of DCT inversion in representing fields with binary information about landslide occurrence and non-occurrence is investigated. To illustrate the capability and feasibility of the proposed method, a hypothetical example and a case study of a 2014 regional rainfall-triggered landslide event in Qinglian Town, Chongqing, China are presented. The results demonstrate that the location, size and shape of landslide have little influence on the DCT inversion. Considering spatial variation in model calibration can significantly improve model prediction performance for regional rainfall-induced landslides. The topography of the landslide initiation area may affect the influence of the model calibration if the field observations do not agree with the fundamental physics underlying the stability equation used in the model. [ABSTRACT FROM AUTHOR]

Landslide prediction is critical for the early warning of a landslide occurrence. Existing stepwise landslide displacement prediction methods are mostly data-driven approaches. However, these models are vulnerable to overfitting, and the low-dimensional numerical features with high numerical volatility prevent them from precisely quantifying the rapid increase in daily displacement in the acceleration phase. Therefore, we propose a semantic information-driven stepwise landslide displacement prediction model comprising an identifier in the displacement phase and a predictor in the acceleration phase. First, the raw landslide monitoring data are converted into text-based semantic information and the semantic features are fused. Subsequently, based on the daily displacement and velocity, we propose a sliding window phase division algorithm to divide the stepwise landslide phase into stationary and acceleration phases. Finally, the landslide displacement phase is identified, and the displacement during the acceleration phase is predicted. The experimental results of the model on the Xinpu and Qingshi landslides in Chongqing, China, show that the proposed model exploits the derived semantic information to identify the landslide acceleration phase qualitatively, and predict the daily displacement of the acceleration phase quantitatively. The proposed model provides a valuable reference for the early warning of stepwise landslides. [ABSTRACT FROM AUTHOR]

Landslides frequently occur in forest areas with a steep hillside, especially when severely disturbed by human activities. After sustained heavy rainfall, a landslide occurred near the Tianwan tunnel entrance of the Chongqing-Huaihua railway in China. Fortunately, the landslide debris was successfully intercepted by a flexible barrier originally installed to stop rockfalls, which is, to date, the first publicly reported case of landslide debris having been successfully intercepted by a flexible barrier without any damage, in mainland of China. A field investigation was first conducted, and then a back analysis of the landslide mobility and the interaction between the landslide and the flexible barrier was carried out. The back analysis showed that the impact energy was three-times larger than the rated energy capacity of the flexible barrier. It also showed that the elongation of the brake rings and the deflection of the flexible barrier from the numerical simulation was comparable to that from the field measurements. The fact that these brake rings were not elongated to their limit indicated that the capacity of the flexible barrier still had a surplus. Finally, to investigate the maximum energy capacity of a flexible rockfall barrier in resisting landslide debris, parametric analyses of a flexible barrier impacted by landslide debris with different impact energies and velocities were carried out using a coupled ALE-FEM modeling technique. The results showed that the flexible barrier dissipated less than 40% of the total energy of the landslide debris. With an increase of impact energy, the energy dissipation ratio of the flexible barrier decreased linearly. The maximum energy capacity of a flexible rockfall barrier in resisting landslide debris is four-times that of resisting a rockfall. [ABSTRACT FROM AUTHOR]

Machine learning-based methods are commonly used for landslide susceptibility mapping. Most of the recent publications focused on quantitative analysis, i.e., improving data processing methods, comparing and perfecting the data-driven model itself, but rarely taking the qualitative aspects of the local landslide occurrences into consideration and the further analysis of the key features was always lacking. This study aims to combine qualitative and quantitative analysis and examine its effect on mapping accuracy; based on the feature importance ranks and the related literature, the key features for identifying landslide/non-landslide points of different sub-zones were further analyzed. Before modeling, the study area Yunyang County, Chongqing City, China, was manually divided into four sub-zones based on the information from geological hazards exploration in Chongqing, including the mechanism of landslide formation and sliding failure and geomorphic unit characteristics. Upon the qualitative analysis basis, five grid searches tuned random forest models (one for the whole region and four for the sub-zones independently) were established by 1654 data points and 20 conditioning features. Compared with the conventional data-driven method, the integrated quantitative evaluation based on the qualitative analysis results showed higher reliability, which not only improved the mapping accuracy but also increased the AUC values of all four sub-models, which were 8.8%, 2.3%, 1.9% and 9.1% higher than that of the parent model. Moreover, the quantitative evaluation based on the qualitative analysis revealed the key factors affecting local landslide formation. Therefore, qualitative analysis is recommended in future landslide susceptibility modeling with the additional combination of data-driven methods. [ABSTRACT FROM AUTHOR]

The extraction of landslide deformation using terrestrial laser scanning (TLS) has many important applications. The landslide deformation can be extracted based on a digital terrain model (DTM). However, such methods usually suffer from the ill-posed problem of a multiplicative error model as illustrated in previous studies. Moreover, the edge drift of commonly used spherical targets for point cloud registration (PCR) is ignored in the existing method, which will result in the unstable precision of the PCR. In response to these problems, we propose a method for extracting landslide deformations from TLS data. To archive the PCR of different period point clouds, a new triangular pyramid target is designed to eliminate the edge drift. If a fixed target is inconvenient, we also propose a PCR method based on total station orientation. Then, the use of the Tikhonov regularization method to derive the weighted least squares regularization solution is presented. Finally, the landslide deformation is extracted by DTM deference. The experiments are conducted on two datasets with more than 1.5 billion points. The first dataset takes Lashagou NO. 3 landslide in Gansu Province, China, as the research object; the point cloud data were collected on 26 February 2021 and 3 May 2021. The registration accuracy was 0.003 m based on the permanent triangular pyramid target and 0.005 m based on the total station orientation. The landslide deforms within 3 cm due to the ablation of the frozen soil. The second dataset is TLS data from the Lihua landslide in Chongqing, China, collected on 20 April 2021 and 1 May 2021. The overall deformation of the Lihua landslide is small, with a maximum value of 0.011 m. The result shows that the proposed method achieves a better performance than previous sphere-based registration and that the weighted least square regularization iterative solution can effectively reduce the ill-condition of the model. [ABSTRACT FROM AUTHOR]

On 21 October 2017, days of heavy rainfall triggered a landslide in Guang'an Village, Wuxi County, Chongqing, China. According to the field investigation after the incident, there is still a massive accumulation body, which could possibly reactivate the landslide. In this study, to explore the long-term evolution of the deformation after the initial Guang'an Village Landslide, a time-series InSAR technique (TS-InSAR) was applied to the 128 ascending Sentinel-1A datasets spanning from October 2017 to March 2022. A new approach is proposed to enhance the conventional TS-InSAR method by integrating LiDAR data into the TS-InSAR process chain. The spatial–temporal evolution of post-event deformation over the Guang'an Village Landslide is analyzed based on the time-series results. It is found that the post-event deformation can be divided into three main stages: the post-failure stage, the post-failure and reactivation stage, and the reactivation stage. It is also suggested that, although the study area is currently under the reactivation stage, there are two active deformation zones that may become the origin of a secondary landslide triggered by heavy rainfall in the future. Moreover, the nearby Yaodunzi landslide might also play an important role in the generation and reactivation of a secondary Guang'an Village Landslide. Therefore, continuous monitoring for post-event deformation of the Guang'an Village Landslide is important for early warning of a secondary landslide in the near future. [ABSTRACT FROM AUTHOR]

Landslides are one of the most severe and common geological hazards in the world. The purpose of this research is to establish a coupled landslide warning model based on random forest susceptibility zoning and precipitation. The 1520 landslide events in Fengjie County, Chongqing, China, before 2016 are taken as research cases. We adapt the random forest model to build a landslide susceptibility model. The antecedent effective precipitation model, based on the fractal relationship, is used to calculate the antecedent effective precipitation in the 10 days before the landslide event. Based on different susceptibility zones, the effective precipitation corresponding to different cumulative frequencies is counted as the threshold, and the threshold is adjusted according to the fitted curve. Finally, according to the daily precipitation, the rain warning levels in susceptibility zones are further adjusted, and the final prewarning model of the susceptibility zoning and precipitation coupling is obtained. The results show that the random forest model has good prediction ability for landslide susceptibility zoning, and the precipitation warning model that couples landslide susceptibility, antecedent effective precipitation, and the daily precipitation threshold has high early warning ability. At the same time, it was found that the precipitation warning model coupled with antecedent effective precipitation and the daily precipitation threshold has more accurate precipitation warning ability than the precipitation warning model coupled with the antecedent effective precipitation only; the coupling of the two can complement each other to better characterize the occurrence of landslides triggered by rainfall. The proposed coupled landslide early warning model based on random forest susceptibility and rainfall inducing factors can provide scientific guidance for landslide early warning and prediction, and improve the manageability of landslide risk. [ABSTRACT FROM AUTHOR]

At 2:00 a.m. on 1 July 2020, after several days of continuous heavy rainfall, the Xinfangzi landslide occurred in Zhengping Village, Ganshui Town, Qijiang District, Chongqing City, China. The area of the landslide was about 3.85 × 104 m2 and the volume was about 71.22 × 104 m3. The Xinfangzi landslide can be defined as a push-type landslide. Specifically, the main sliding area pushed the front squeezing area, causing it to slide laterally. The entire slip path of the landslide was a broken line, while the right edge and the front shear outlet position slipped loosely in a circular arc. In this study, finite element numerical calculations were used to compare and analyze the multisection plastic deformation of the landslide under natural and rainfall conditions based on field investigations. The formation and failure characteristics of the Xinfangzi landslide were further revealed. The results show that the cross and longitudinal sections of the landslide were in a critical state of instability under natural and rainfall conditions. A compound section was established along the slip path based on the cross and longitudinal sections. Due to the extrusion of the trailing edge of the landslide, the compound section's leading edge was in a state of instability under natural conditions. Under rainfall conditions, the increase in the unit weight of the sliding mass was superimposed on the compound section, which amplified the thrust of the trailing edge and further accelerated the overall sliding process of the landslide. Based on the macroscopic deformation of the landslide, it was found that the sliding of the trailing edge of the landslide is the key factor promoting the failure of the front edge, and local slump is possible under rainfall conditions. [ABSTRACT FROM AUTHOR]