Landslide Warning Research Articles

Machine Learning–Based Systems for Early Warning of Rainfall-Induced Landslide

Machine Learning–Based Systems for Early Warning of Rainfall-Induced Landslide

Ultra-High Sensitivity Real-Time Monitoring of Landslide Surface Deformation via Triboelectric Nanogenerator.

Monitoring surface deformation is crucial for the early warning of landslides, facilitating timely preventive measures. Triboelectric nanogenerator (TENG) demonstrates great potential for self-powered distributed monitoring in remote and power-scarce landslide areas. However, landslides deform typically at a rate of a few millimeters per day (mm d-1), making it challenging for TENG to directly monitor the deformation process. Herein, a method for monitoring surface deformation of landslides by constructing an ultra-low-speed triboelectric displacement sensor (US-TDS) is reported. Utilizing a force storage-release device and an accelerator, the US-TDS can produce obvious sensing signals at a linear input speed of 4.32mm d-1. The coefficient of determination (R2) for the fitting curve of the pulse signals within the speed range of 21.6 to 129.6mm d-1 reaches 0.999. Moreover, US-TDS can detect deformation displacement as small as 0.0382mm. The stability of US-TDS displacement measurements is confirmed at a speed of 108mm d-1, with relative errors under 1%. Ultimately, a real-time monitoring and early warning system for landslide surface deformation is constructed and verified through a combination of indoor simulations and outdoor experiments. This work provides a feasible solution for the scientific monitoring and early warning of the landslide development.

Landslide Hazard and Rainfall Threshold Assessment: Incorporating Shallow and Deep-Seated Failure Mechanisms with Physics-Based Models

Landslides pose a significant threat worldwide, leading to numerous fatalities and severe economic losses. The city of Manizales, located in the Colombian Andes, is particularly vulnerable due to its steep topography and permeable volcanic ash-derived soils. This study aims to assess landslide hazards in Manizales by integrating shallow planar and deep-seated circular failure mechanisms using physics-based models (TRIGRS and Scoops3D). By combining hazard zonation maps with rainfall thresholds calibrated through historical data, we provide a refined approach for early warning systems (EWS) in the region. Our results underscore the significance of the landslide hazard maps, which combine shallow planar and deep-seated circular failure scenarios. By categorizing urban areas into high, medium, and low-risk zones, we offer a practical framework for urban planning. Moreover, we developed physics-based rainfall thresholds for early landslide warning, simplifying their application while aiming to enhance regional predictive accuracy. This comprehensive approach equips local authorities with essential tools to mitigate landslide risks, refine hazard zoning, and strengthen early warning systems, promoting safer urban development in the Andean region and beyond, as the physics-based methods used are well-established and implemented globally.

Construction of landslide warning by combining rainfall threshold and landslide susceptibility in the gully region of the Loess Plateau: A case of Lanzhou City, China

Regional forecasting of the landslide rainfall threshold is complex, and the empirical prediction of the rainfall threshold relies on historical statistical data. To obtain the intensity (I) ∼ duration (D) rainfall threshold, Lanzhou City was selected as the research object, and a field investigation, 3S technology, mathematical statistics and a Python language environment were employed to develop a set of automatic rainfall threshold calculation programs. Factor elimination, the information value method, weight of evidence, the frequency ratio, the density method and receiver operating characteristic (ROC) curve were applied to obtain landslide susceptibility, and the I ∼ D rainfall threshold (temporal probability) and landslide susceptibility (spatial probability) were combined to construct regional landslide rainfall threshold warning levels (red, orange, yellow and blue). The results showed the rainfall duration of historical landslides ranged from 0.99 to 216 h, thereinto, it was accounted for 92.78 % with that of between 6 and 96 h, and the rainfall intensity was between 0.11 and 23.6 mm/h, and the α and β of rainfall threshold curve was gradually increased with an increase in the rainfall threshold probability. Additionally, when the rainfall condition probability was the same, landslides occurrence was more frequent in Chenguan District, where there are large areas of bare land, and slope and annual rainfall are less than 10 °and 300 mm, respectively. Fifteen, nine, and six factors were used to obtain landslide susceptibility, and the accuracy of the ROC ranged from 0.78 to 0.80 under different methods and different factor combinations. The spatial distribution of landslide susceptibility was very similar. There were 35, 12, 25 and 108 landslides associated with red, orange, yellow and blue levels, respectively, under the spatial–temporal probability, but 30, 26, 54 and 70, respectively, under temporal probability. Under the different conditions, there were 89, 69 and 22 landslide points with unchanged, increased and decreased warning levels, respectively. The ability and accuracy of rainfall threshold warning and prediction were effectively improved by integrating the temporal and spatial probabilities of the rainfall threshold.

Early warning of landslides based on statistical analysis of landslide motion characteristics and AI Earth Cloud InSAR processing system: a case study of the Zhenxiong landslide in Yunnan Province, China

Early warning of landslides based on statistical analysis of landslide motion characteristics and AI Earth Cloud InSAR processing system: a case study of the Zhenxiong landslide in Yunnan Province, China

Displacement residuals reveal landslide regime shifts

AbstractDespite significant progress in the development of advanced technologies for detecting and monitoring unstable slopes, accurately predicting catastrophic landslides remains a challenge. To tackle this challenge, our research integrates advanced prediction models and granular systems theory to provide insights into regime shifts within slow-moving deep-seated landslide dynamics. Our approach is designed to discern exceptional departures from historical landslide dynamics. The approach leverages the “group dynamics,” crucial for identifying precursory failure indicators, according to the generic dynamics of the precursory failure regime in granular systems. We select three different monitored slow-moving landslides as test cases. We employ an error correction cointegration vector autoregression model together with an exogenous regressor to encode historical spatiotemporal landslide dynamics and predict displacement at multiple locations by considering the historical landslide motion and relationship with external triggers. Displacement residuals are obtained by computing the difference between predicted and measured displacement for a given historical calibration time window. Threshold values for the displacement residuals are determined by analyzing the historical distribution of these residuals. Lastly, persistence in time of the threshold exceedance and the number of monitoring points that exceed the threshold at the same time are considered to encode the group dynamics. This approach offers several advantages, including the effective identification of critical regime shifts, adaptability, and transferability, and it introduces regime shift information into local landslide early warning systems. This approach can enhance confidence in the resultant alert, particularly when integrated with conventional alert systems, thereby improving the reliability of landslide warning systems.

Comparative analysis of conventional and machine learning techniques for rainfall threshold evaluation under complex geological conditions

This research focuses on the essential task of defining rainfall thresholds in regions with complex geological features, specifically at a regional scale. It examines a variety of methodologies, from traditional empirical-statistical methods to cutting-edge machine learning (ML) techniques, for establishing these thresholds. The Emilia-Romagna region in Italy, known for its intricate geological structure and prevalence of weak rocks that often lead to large and deep-seated landslides, serves as the study area. The region’s complex interplay between rainfall and landslide incidences poses a significant challenge in accurately determining rainfall thresholds. The effectiveness of ML methods is compared against conventional empirical-statistical approaches, evaluating factors such as prediction accuracy, model complexity, and the interpretability of results for use by regional landslide warning system operators. The findings indicate that machine learning techniques have an edge over traditional methods, yielding higher performance scores and fewer false positives. Nevertheless, these advancements are modest when considering the increased complexity of ML methods and the incorporation of additional rainfall parameters. This underlines the continued need for improvements in data quality and volume. The study stresses the importance of enhancing data collection and analysis techniques, especially in an era where advanced AI tools are increasingly available, to improve the accuracy of predicting rainfall thresholds for effective landslide warning systems.

Monitoring and disaster prevention of high and steep sandstone slopes along highways under construction

High and steep sandstone slopes along highway line are at high risk of disasters such as landslides, cracking of support structures, and so on. The monitoring, early warning, and emergency response of such slope disaster face enormous challenges, especially during the rainy season. In this paper, intelligent monitoring, early warning and forecasting system were carried out for the high steep sandstone slope with a transmission line tower at the slope crest along the highway under construction in Guangxi, China. The automatic monitoring data, emergency rescue program and rescue effect were analyzed, and emergency rescue measures for high steep slope protection were taken. The research results show that timely access to disaster warning information can effectively support the analysis of disaster causes and the evaluation of disposal programs. Deep-hole monitoring of deformation characteristics can determine the stable state of slopes, and the tangent angle warning criterion can be used for early warning and prediction of high steep slope landslides. By analyzing the location of the sliding surface and taking timely emergency disposal measures such as layered counterpressure method and micropipes, the landslide activities can be effectively controlled to prevent further acceleration of slope collapse. This study can provide an important reference for the monitoring, early warning, forecasting and emergency rescue of sandstone slopes along highways under construction.

Identification of sliding surface and classification of landslide warning based on the integration of surface and deep displacement under normal distribution theory

Advanced identification of the potential sliding surface of a slope and accurate early warning are crucial prerequisites for effective management of landslides and timely and prevention of catastrophic accidents. This study analyzes the statistical characteristics of landslide displacement evolution. Based on the normal distribution theory, random variables of displacement velocity and acceleration with random errors are introduced into the analysis of surface displacement information, and random variables of relative displacement with random errors are introduced into the analysis of deep displacement information. When the random variables do not follow the normal distribution, the warning time can be obtained. Therefore, an advanced landslide classification warning method is established. The analysis results showed that analysis results from the April 30 landslide project at an open pit mine indicate that the earliest warning time for landslide initiation is 2020/2/19, while the earliest warnings for acceleration occur on 2020/4/15 and the fast acceleration on 2020/4/25. These three-level warning times align with reality, and the inferred slip surface position corresponds to the actual weak layer range. The primary power source driving landslide originates from behind the sliding body which subsequently pushes rock mass along weak layers near the south wing, north wing, and front in succession. Research findings can enhance landslide warning accuracy, facilitate advance identification of sliding surface, provide scientific basis for open-pit slope engineering design, as well as mitigate casualties and property losses.

Time Series Prediction of Reservoir Bank Slope Deformation Based on Informer and InSAR: A Case Study of Dawanzi Landslide in the Baihetan Reservoir Area, China

Reservoir impoundment significantly impacts the hydrogeological conditions of reservoir bank slopes, and bank slope deformation or destruction occurs frequently under cyclic impoundment conditions. Ground deformation prediction is crucial to the early warning system for slow-moving landslides. Deep learning methods have developed rapidly in recent years, but only a few studies are on combining deep learning and landslide warning. This paper proposes a slow-moving landslide displacement prediction method based on the Informer deep learning model. Firstly, the Sentinel-1 (S1) data are processed to obtain the cumulative displacement time-series image of the bank slope by the Small-BAseline Subset Interferometric Synthetic Aperture Radar (SBAS-InSAR) method. Then, combining data on rainfall, humidity, and horizontal and vertical distances of pixel points from the water table line, this study created a dataset with landslide displacement as the target feature. After that, this paper improves the Informer model to make it applicable to our dataset. This study chose the Dawanzi landslide in the Baihetan reservoir area, China, for validation. After training with 50-time series deformation data points, the model can predict the displacement results of 12-time series deformation data points using 12-time series multi-feature data, and compared with the monitoring values, its Mean Square Error (MSE) was 11.614. The results show that the multivariate dataset is better than the deformation univariate data in predicting the displacement in the large deformation zone of bank slopes, and our model has better complexity and prediction performance than other deep learning models. The prediction results show that among zones I–IV, where the Dawanzi Tunnel is located, significant deformation with the maximum deformation rate detected exceeding –100mm/year occurs in Zones I and III. In these two zones, the initiation of deformation relates to the drop in water level after water storage, with the deformation rate of Zone III exhibiting a stronger correlation with the change in water level. It is expected that deformation in Zone III will either remain slow or stop, while deformation in Zone I will continue at the same or a decreased rate. Our proposed method for slow-moving landslide displacement forecasting offers fast, intuitive, and economically feasible advantages. It can provide a feasible research idea for future deep learning and landslide warning research.

Predicting the Deformation of a Slope Using a Random Coefficient Panel Data Model

Engineering constructions in coastal areas not only affect existing landslides, but also induce new landslides. Variation of the water level makes the coastal area a geological hazard-prone. Prediction of the slope displacement based on monitoring data plays an important role in early warning of potential landslide and slope failure, and supports the risk management of hazards. Given the complex characteristic of the slope deformation, we proposed a prediction model using random coefficient model under the frame of panel data analysis, so as to take the correlation among monitoring points into consideration. In addition, we classified the monitoring data using Gaussian mixture model, to take the temporal-spatial characteristics into consideration. Monitoring data of Guobu slope was used to validate the model. Results indicated that the proposed model have a better performance in prediction accuracy. We also compared the proposed model with the BP neural network model and temporal – temperature model, and found that the prediction accuracy of the proposed model is better than those of the two control models.

Deformation and collapse of rock slopes considering weak plane orientation

Some slopes experience multiple slides without collapse, while other slopes collapse once they are unstable. The early warning of slope collapse is a difficult but important subject. Considering the influence of weak planes on rock slope deformation and collapse is helpful for interpreting the behavior of deep-seated landslides and designing an early warning system. To investigate the deformation behavior of rock slopes with consideration on weak planes, artificial cemented sand plates were produced and stacked to form physical slope models with different weak plane orientations, where inclined loading was applied to induce the deformation and collapse of the slope. In addition, the deformation of real slopes was examined based on topographic features. The average strain at collapse is referred to as the critical strain, whose value changes for various slopes. Sorted by critical strain in descending order, the slope models include an anaclinal slope with 60° weak planes, an anaclinal slope with 30° weak planes, a cataclinal slope where the 30° weak planes coincide with the slope face, and a cataclinal slope with daylighting 20° weak planes. Similar to the experimental results, anaclinal slopes also present greater average strain values than cataclinal slopes for real slopes. A smaller critical strain implies a higher possibility for slope collapse when unstable. Local deformation does not always lead to collapse, but as the average velocity and the average strain rate of the sliding body increase, or the velocity ratio (VR) between the upper and lower parts of the sliding body approaches 1, a sliding surface inside the slope is likely developing and coalescing. Hence, such deformation features may contribute to a landslide warning system.

Precipitation-induced landslide risk escalation in China’s urbanization with high-resolution soil moisture and multi-source precipitation product

Landslides pose a formidable natural hazard. Accurate risk assessment of landslides triggered by precipitation heavily relies on hydrometeorological factors, specifically precipitation and soil moisture. However, the insufficient ground-based observations and the coarse spatio-temporal resolution hinder the performance of landslide prediction. It is not clear what hydrometeorological thresholds and triggering mechanisms are more likely to trigger landslides in China, particularly in the context of rapid urbanization. To address these questions, this study investigated 1504 shallow landslide events in Chinese urban and non-urban areas from 2007 to 2019. It utilized daily 1 km soil moisture at various depths (20–100 cm) and multi-source precipitation datasets, including gauge-based gridded dataset, in conjunction with three multi-source fusion precipitation products (Multi-Source Weighted-Ensemble Precipitation − MSWEP, the Climate Hazards Group InfraRed Precipitation with Station dataset − CHIRPS, and the Integrated Multi-satellitE Retrievals for Global Precipitation Measurement − GPM-IMERG), along with dynamic urban impervious area datasets. It aims to determine the optimal multi-source precipitation predictor, the critical soil moisture depth that triggers landslides, and to establish the hydrometeorological thresholds for landslides. Additionally, the impact of urbanization on landslide occurrences was estimated by comparing antecedent precipitation accumulation, soil moisture, and impervious surface ratio dynamics between urban and non-urban areas. The results indicated that a combination of 2-day cumulative CHIRPS precipitation and 100 cm soil moisture provided the most accurate predictions for landslides in urban regions of China (accuracy = 88.5 %), outperforming interpolated ground-based observations and other fusion products. Specifically, landslides become more prone when antecedent cumulative rainfall surpasses 97.42 mm in 2 days and soil moisture exceeds 39.6 % m3/m3 saturation in China. Urban areas experienced high antecedent precipitation levels, lower precipitation (64.40 mm) threshold and soil moisture threshold (38.9 %), and shorter durations at landslide sites compared to non-urban areas (71.90 mm, 41.4 %, and 7 days, respectively). The process of urbanization is observed to decrease soil moisture levels while concurrently elevating rainfall amounts. This phenomenon, combined with anthropogenic activities, including distance from roads and urban impervious surface expansion, contributes to 44.6 % of the causes of landslides by reducing slope stability and increasing the presence of loose material. These findings have implications for landslide warnings in urban areas with limited measurements and contribute to understanding urbanization’s impact on landslide risks in developing nations.

Landslide Deformation Prediction and Automatic Warning by Coupling Machine Learning and Physical Models

AbstractAutomatic deformation forecast and warning of a catastrophic landslide can effectively avoid significant casualties and economic losses. However, currently it has not come to a comprehensive forecast framework covering all the deformation stages of a landslide. Moreover, landslide deformation prediction possesses high error and false alarm rates. This work suggests a novel integrated framework of landslide deformation forecast and warning by coupling machine learning and physical models. The framework can relatively accurately predict all the deformation stages from creeping deformation to critical sliding and features 4 advantages. (a) The forecast indices are established by combining deformation and disaster‐triggering characteristics to improve the prediction accuracy. (b) It leverages the advantage of C5.0 decision tree algorithm in knowledge interpretability to automatically extract deformation forecast criteria. (c) It capitalizes on the precision superiority of a graph convolutional network in time‐series data learning to predict the four deformation stages from creeping deformation to rapidly accelerated deformation. (d) It utilizes the physical and mechanical bases of Morgenstern‐Price method to forecast the critical sliding stage. Zhujiatang Landslide is a large‐scale deep‐seated soil landslide with significant deformation. It is selected as a case study because it has endangered 1,131 persons and may cause a direct financial loss of 100 million RMB. The validating and predicting Accuracy values attain 97.39% and 95.72%, respectively, and the Kappa values reach 0.91 and 0.93, respectively. The landslide will run out when it suffers from a rainstorm with cumulative rainfall of 79.57 mm and an earthquake with a horizontal coefficient of 0.04.

Landslide Warning Area Delineation through Seismic Signals and Landslide Characteristics: Insights from the Silabaku Landslide in Southern Taiwan

Abstract Publishing timely warnings of areas potentially affected by landslides and subsequent events remains challenging in hazard assessment. Recently, seismic monitoring has emerged as a promising technique for landslide detection and location determination. However, the effectiveness of this approach is limited by location errors. To overcome this problem, we propose a hybrid method (a three-step process) through seismic analysis and consideration of prior geomorphological characteristics of areas and aspects in a landslide inventory to constrain the potential landslide locations. The first analysis entails a grid-based single-force inversion of 0.02–0.05 Hz seismic signals by spatial grid spacings of 0.2° and 0.05° to retrieve the single-force direction, the estimated landslide area, and a wide search range for a given source. The second analysis involves two location methods for high-frequency seismic signals (>1 Hz) with a finer spatial grid spacing of 0.01°, gradually reducing the landslide search region. By integrating seismology-determined information with a landslide inventory, potential landslide locations and warning areas could be identified effectively. Then we apply the proposed the three-step analysis method to the 2021 Silabaku landslide in southern Taiwan. The results showed that the seismic techniques we used progressively condensed the initial search range from 1831 to 770 km2, converging to a small area of 99 km2. The estimated landslide area was 0.80±0.04 km2, and the force direction was 154.09° ± 6.76°. Subsequently, according to the earlier information, we identified three possible landslide source locations from the 2020 landslide inventory. These findings have implications for warning for two road sections, one of which was subsequently damaged by debris flow after the Silabaku landslide. This study highlights the effectiveness of the hybrid three-step algorithm in constraining areas potentially affected by landslides, facilitating timely assessments for warnings in southern Taiwan.

Derivation of Landslide Rainfall Thresholds by Geostatistical Methods in Southwest China

Derivation of Landslide Rainfall Thresholds by Geostatistical Methods in Southwest China

Time-dependent probabilistic evaluation of rainstorm-induced shallow landslides considering influence of impermeable bedrock

ABSTRACT A time-dependent probabilistic evaluation method is proposed for assessing the stability of soil slopes during rainstorm events. The method considers multiple failure scenarios, including sliding along the wetting front and shallow soil–rock interface. The application of the method is demonstrated by a typical rainstorm-induced shallow landslide in Zhejiang Province, China. The results indicate that the landslides occurred when the failure probability exceeded a certain level, providing references for real-time landslide warnings. Moreover, the factors influencing the probabilistic results have been investigated. A two-dimensional limit state surface is used to explain the change of slope failure scenarios during the rainfall process, confirming the controlling role of the impermeable bedrock surface. The probabilistic evaluation results using the actual rainfall data can better reflect slope performance via comparing the computed failure probability using idealised rainfall patterns and actual rainfall data. Additionally, the parameter sensitivity analysis shows that the matric suction, the effective cohesion and the effective internal friction angle significantly influence the slope performance during the early stage of rainfall. However, during the later stage of rainfall, the saturated permeability coefficient becomes the most sensitive parameter. The applicability of the proposed method was further verified via three other typical rainstorm-induced shallow landslide cases.

Novel evaluation method based on critical arch height as instability criterion for sustaining arch locked-segment-type slopes

In sustaining arch locked-segment-type slopes, natural soil arches play a key anti-sliding role in the slope's evolution. In this study, a self-developed model test device was used to simulate the whole process of deformation evolution of sustaining arch locked-segment-type slopes, and the formation of natural sustaining arch and its locking control effect on slope stability were studied. The test results show that the continuous formation and progressive destruction of the sustaining arch were observed. The sustaining arch formed in the second time has the best locking effect, and the anti-sliding force reaches its stress peak point. However, the slope is not in a critically unstable state, instead, the stress is continuously adjusted to form a larger range of soil arch to resist the slope thrust. Consequently, the slope destabilizes until the ultimate shear strength of arch foots is exceeded, at which point the critical arch height of the arch is reached. The critical arch height mechanical model for slope stability analysis was developed based on the soil arching effect and limit equilibrium theory. The applicability of the model was demonstrated by the physical test and Xintan slope data, which can provide some guidance for early warning of landslides.

Creating a big data source of landslide deformation stages: New thoughts on identifying displacement warning thresholds

The determination of displacement warning thresholds is a difficult and crucial task for landslide early warning, as it requires a deep understanding of displacement monitoring curve patterns and significantly impacts the effectiveness of natural hazard mitigation efforts. In recent years, the emergence of advanced technologies such as the global navigation satellite system (GNSS) and interferometric synthetic aperture radar has enabled large-scale monitoring of landslide surface displacement, while also introducing new challenges in establishing displacement warning thresholds. While displacement monitoring curves exhibit various patterns in large-region monitoring environments, the conventional Saito method assumes a landslide deformation trend of three creep stages and cannot achieve ideal early warning results. Consequently, this study is conducted based on an analysis of a GNSS surface displacement curve database built by the China Institute of Geological Environmental Monitoring. An identification method for identifying landslide deformation stages is developed based on the idea of morphology, leading to the creation of a big data source of landslide deformation stages. The spatial distribution of these deformation stages within the data source serves as the foundation for developing a technique for identifying deformation thresholds and a corresponding multilevel early warning method. The early warning performance is validated using long-duration displacement monitoring curves, and the spatiotemporal distribution characteristics of the deformation stages are discussed. This study can provides valuable insights for displacement monitoring, early warning, and hazard prevention of large-scale landslides.

Model test study on natural frequency evolution and early warning of reservoir landslides under water level fluctuations

Model test study on natural frequency evolution and early warning of reservoir landslides under water level fluctuations