Search

With the development and application of machine learning, significant advances have been made in landslide susceptibility mapping. However, due to challenges in actual field landslide investigations, current landslide susceptibility mapping is usually characterized by insufficient landslide samples (positive samples) and low reliability of non-landslide samples (negative samples). Considering Lianghe County in Yunnan Province, China, as an example, this paper aims to research the effectiveness of three oversampling models in generating positive samples for landslides: Conditional Tabular Generative Adversarial Networks (CTGAN), Generative Adversarial Networks (GAN), and the traditional Synthetic Minority Oversampling Technique (SMOTE) algorithms. Additionally, three machine learning methods, including 1D Convolutional Neural Network-Long Short-Term Memory Neural Network (CNN-LSTM), Random Forest (RF), and Gradient Boosting Decision Tree (GBDT) classifiers, are used for landslide susceptibility assessment. We also devise a non-landslide data (negative samples) screening method utilizing a self-trained support vector machine within a semi-supervised framework. The results show that by training on the dataset after negative sample screening, the AUC values for the 1D-CNN-LSTM, RF, and GBDT models have shown significant improvement, increasing from (0.778, 0.869, 0.849) to (0.837, 0.936, 0.877). Compared with the original training set, the prediction accuracy of the three machine learning models is improved after training on the augmented data by CTGAN, GAN, and SMOTE models. The RF model, augmented with 200 positive samples generated by CTGAN, achieves the highest prediction accuracy in the study (AUC = 0.962). The 1D CNN-LSTM model achieves its highest prediction accuracy (AUC = 0.953) when augmented with 200 positive samples from GAN. Similarly, the GBDT model reaches its highest prediction accuracy (AUC = 0.928) when augmented with 200 positive samples created by SMOTE. In addition, the spatial distribution of data indicates that the data generated by the generative adversarial model exhibits higher diversity, which can be used for landslide susceptibility assessment. [ABSTRACT FROM AUTHOR]

There are many landslide deposits in the reservoir and bank areas of hydropower projects in southwest China. These deposits are easy to deform and fail along the existing sliding zone under hydrodynamic conditions, such as rainfall and reservoir water fluctuation. The slip soil in Dahua landslide deposits which are located at the Dahuaqiao Hydropower Plant in the Lancang river basin of southwest China, was studied as an example. The experimental study was carried out to explore the soil-water characteristic curve (SWCC) of slip soil. The permeability characteristics and the evolution in the unsaturated seepage process were analyzed, and the saturated permeability characteristics under different confining pressures and osmotic pressures were studied. The results indicated that the VG model could be used to describe the SWCC and the variation of relative permeability coefficient of Dahua slip zone soil. Through the saturated permeability test, it is found that under the same other conditions, the larger the confining pressure is, the weaker the saturated permeability of the slip soil is; while the larger the osmotic pressure is, the stronger the saturated permeability is. Furthermore, the saturated permeability characteristics of the slip zone soil manifested a strong non-Darcy flow behavior, and the relationship between permeability velocity and hydraulic gradient satisfied the Forchheimer binomial model. When the confining pressure was low, the linear term coefficient of velocity had a more significant effect on the permeability, whereas the quadratic term coefficient of velocity gradually dominated in the seepage process with the increase of confining pressure.

Landslide-generated impulse waves (LGIWs) often cause enormous damage. This study proposes a two-way coupled discrete element method (DEM) and smoothed particle hydrodynamics (SPH) method to study the whole hazard chain of LGIWs in mountain reservoirs. In this method, a two-way coupling code is developed to connect the DEM program Particle Flow Code 3D (PFC3D) and the SPH code DualSPHysics for simulating landslide sliding and fluid motion. A series of discretization operations are conducted to transform the DEM balls into discrete balls composed of SPH particles, and the dynamic boundary condition method is used to calculate the interaction forces between fluid and solid phases. This work is validated through a comparative analysis with published physical model experiments. Using the proposed method, the LGIWs induced by the RS deposit in the RM reservoir is simulated. The processes of landslide motion, impulse waves generation, propagation, and running up on the dam are studied. The results show that the proposed method possesses the capability to simulate and predict the hazards of LGIWs. This article represents the first implementation of PFC3D within a coupled DEM-SPH framework to study LGIWs, and provides a valuable way for assessing the associated risks. [ABSTRACT FROM AUTHOR]

A variational method, using a lateral force as a functional and incorporating equilibrium in-the-small, is proposed for limit equilibrium analysis, in which a conventional safety functional in the form of a quotient and the need to use the method of slices or Lagrange multipliers can be avoided. The lateral force is a real force acting on a retaining wall, but is a fictitious force in slope stability analysis. Furthermore, the fictitious force is related to the factor of safety. The force is a push, pull, or null when the slope is unstable, stable, or critical. By setting this lateral force to zero, a critical stability state can be obtained. The proposed variational method is capable of reproducing the classical solutions and yielding new useful analytical results. The proposed method can be a viable alternative technique, because of its effectiveness and ease of use is comparable to the conventional variational approaches. The limit-equilibrium analysis technique is widely adopted in the practice of slope stability and earth pressure on retaining walls. Among the existing techniques in limit-equilibrium analysis, the variational calculus method from a mathematical perspective is more accurate than traditional techniques. However, this method was greatly limited due to the cumbrous solution on the quotient form functional. The proposed method uses a lateral force as functional, incorporating equilibrium in-the-small. It can simplify the solution significantly and the assumption of slip surface or relation of forces can be avoided. Results obtained from the proposed method can be used in the design of clay slopes, seismic stability of cohesionless slopes, lateral earth pressure on retaining walls under seismic conditions, and lateral earth pressure on rough retaining walls. The charts can be used to conveniently assess the stability of slopes or earth pressure on retaining walls under certain conditions in the preliminary design. The suggested method can also be further developed into a broader range of practical applications. [ABSTRACT FROM AUTHOR]

The primary objective of this article is to present a case study of valley deformation in high arch dam and its relationship with water level fluctuations. Data of displacement versus reservoir level monitored over a five-year period are used to determine possible relationship between displacement and reservoir level. It is found that valley deformations are closely related to the filling-drawdown cycles of reservoir. Comparison of velocity of displacement of the valley with reservoir water level reveals that the reservoir water drawdown has a much stronger influence on the valley deformation. Based on these results, it is anticipated that the present research will provide additional insight into the mechanism of valley deformation and provide guidelines for preventive strategies and treatments. [ABSTRACT FROM AUTHOR]

To enrich understanding of how fluid flows through the natural filled jointed rock, the permeability experiments of the natural filled jointed dacite rock under repeated loading and unloading conditions are carried out. Experimental results show that the response of permeability of filled jointed dacite rock to confining pressure under repeated loading and unloading conditions is considered as porous media. Permeability and gas pressure are positively correlated under repeated loading and unloading conditions. Relationship between permeability and gas pressure shows that the higher confining pressure, the smaller permeability increasing rate. The calculated equivalent crack apertures are very small at the order of magnitude about 10−8 m. The power law is found to well describe the relationship between equivalent crack aperture and effective stress of natural filled jointed dacite rock. With the parameters determined based on experiment results, an empirical expression is proposed. [ABSTRACT FROM AUTHOR]

During the reservoir filling of many high arch dams, it was observed that the valley slopes on both sides deformed towards the river. The measured value of valley narrowing lateral deformation (VNLD) reached 89.54 mm by October 2018. This rare phenomenon contradicts the traditional experience of dam engineering and lacks a convincing explanation. An arch dam project in China is selected as a case study to analyze the mechanism of the VNLD. Based on a comprehensive analysis of the geological background and the VNLD measurements, a numerical study is performed to identify the main factors that may contribute to the VNLD phenomenon. It is concluded that the time-dependent valley deformation is due to the rheological behaviour of rock mass with the change of hydrogeological conditions. The artesian water head in the deep-confined aquifer is increased and the mechanical parameters in fault and shear zones are weakened after reservoir filling, which is associated with the specific hydrogeological structures. The rise of artesian water head is the main induced factor that contributes to the uplift deformation, which is observed at the dam site, and can cause the VNLD in the horizontal direction. The simulated results considering these factors agreed well with the observed data. [ABSTRACT FROM AUTHOR]