Rainfall-Induced Landslides: Influencing, Modelling and Hazard Assessment

1. Introduction to the Special Issue

Landslide hazards pose a great threat to people’s lives and the safety of their property all over the world, especially in mountainous areas. Rainfall is one of the main trigger factors of landslides, and the induced mechanism is complicated and affected by multiple environmental factors such as rainfall intensity, rainfall duration, landform and soil layer structure, and has a high degree of regional specificity and sudden occurrence [1,2]. Rainfall has a significant impact on the mechanical properties of slope soil and underground rock mass, which usually leads to soil softening and pore water pressure increase, and then leads to landslide instability, forming a large number of destructive debris flow or collapse [3,4]. The occurrence process of landslides induced by rainfall includes soil and water loss and erosion on the surface, and also involves water infiltration into the underground, causing deep soil saturation, thus triggering a landslide. The migration and convergence patterns of water flow under different rainfall characteristics have a key influence on the triggering mechanism of a landslide [5]. Extreme rainfall events caused by climate change have become more frequent in recent years, exacerbating the risk of landslides. Therefore, landslide prediction and risk management have received more and more attention from the scientific community [6,7,8]. This Special Issue is devoted to cutting-edge research on the causes, mechanisms, modeling and disaster management methods of rain-induced landslides, with a view to providing new insights and effective mitigation strategies.

In the Special Issue, the research covers the analysis and monitoring of landslide formation, the development and verification of landslide prediction models, the influencing factors and failure prediction of slope stability, the cause and prediction of debris flow and so on. The 14 articles in the Special Issue are roughly divided into three categories. (1) Rainfall-induced landslides: Based on a variety of methods, such as model tests, numerical simulation, artificial intelligence algorithms and theoretical deduction, these papers (nine in total) deeply analyzed how rainfall causes a landslide and its prediction methods, including the development and application of rainfall-induced landslide monitoring systems, the establishment of risk prediction models and the analysis of rainfall-induced landslide mechanisms. (2) Slope stability analysis: Based on field investigations, laboratory tests, numerical simulation and other methods, these papers (three in total) put forward the control factors affecting slope stability and the method of instability prediction. (3) Mutual control factors and prediction methods of debris flow stability: Based on actual debris flow disaster cases, these papers (two in total) used indoor experiments and numerical algorithms to study the factors affecting the stability of debris flow areas and their prediction schemes, revealed the spatial distribution characteristics and causes of the mutual control factors of debris flow, and established a risk prediction method for debris flow disasters based on automatic calculation algorithms.

In some papers, the correlative mechanism and prediction method of rainfall-induced landslides are studied by using physical models. Paswan et al. [Contribution 1] developed a rain-induced landslide monitoring system to solve the problem of limited landslide prediction in northern India during the rainy season which can record the real-time displacement and volumetric water content of the slope. Meanwhile, to further test the applicability of the monitoring system, a physical slope model was made based on actual scenarios, and a physical test of the rain-induced slope was carried out. The results show that the developed system can effectively monitor the gradient and abrupt change process of rainfall-induced landslides. Taking the Woda landslide in the upper reaches of the Jinsha River as the engineering background, Li et al. [Contribution 2] designed a model test to study the development of paleo-landslides with cracks under the action of rainfall infiltration and revealed the activation mechanism of rainfall and cracks on paleo-landslides. The depth of slope infiltration directly affects the depth of landslide failure. Therefore, Xiao et al. [Contribution 3], taking the Xiashu loess slope as the engineering background, conducted field rainfall model tests and obtained the main discrimination index of slope rainfall infiltration depth, which laid the foundation for the establishment of a slope rainfall infiltration prediction model.

In addition, some other papers have used numerical simulation to analyze the failure mode and mechanical response characteristics of a slope under the action of rainfall. Huang et al. [Contribution 4] took an overturning bank slope of Lancang River in China as the engineering background and adopted a numerical simulation method to analyze the seepage field characteristics and mechanical response characteristics (displacement, stress-strain, plastic deformation, etc.) of the bank slope under the action of rainfall from two and three dimensions, respectively, for the hydrodynamic failure modes of the bank slope under different rainfall conditions. The evolution model of overturning slope deformation under the action of rainfall was revealed. Based on the DuMux, which is a simulator of fluid flow in porous media, and the concept of the local factor of safety (LFS), Moradi et al. [Contribution 5] conducted a comparative study on the application effects of three simplified models (without considering the dynamic interaction between groundwater flow and soil mechanics) and the complete two-phase flow fully coupled fluid mechanics model for the evaluation of the stability of variable saturated landslide-prone slopes under two rainfall intensity conditions. The results show that the LFS results obtained by the three simplified models and the fully coupled model are consistent. KC et al. [Contribution 6] took the landslide and debris flows in Kalli village, which is in the Acham District of Nepal, located in the Lesser Himalayas Mountains, as the engineering research background; they carried out numerical simulation based on a multiphase flow model and adopted GRASS GIS 8.3 to analyze the evolution characteristics of debris flow during landslide movement. Yu et al. [Contribution 7] took the fanling landslide in Shandong, China, as the engineering background; they conducted a numerical study on the response characteristics of the seepage field of the landslide under different rainfall conditions and found that short-term fluctuating rainstorms were more likely to cause landslides than long-term stable rainfall.

At the same time, some papers systematically analyze the landslide disasters induced by rainfall through theoretical research, establishing a disaster analysis model and putting forward a landslide failure time prediction method. Tseng et al. [Contribution 8] took the landslide in Pingtung County, Taiwan Province, as the research object. By establishing the evaluation indexes, namely, the rainfall trigger index (I_RT) and an index of slope environmental strength potential (I_SESP), of landslide damage to land use after four heavy rainfall events, they could effectively estimate the damage of rainfall-induced landslides to land use. Tao et al. [Contribution 9] used four kinds of filters to test the velocity time series, compared and analyzed the prediction effect of landslide failure time, and finally proposed a hierarchical prediction method combining a short-term smoothing filter (SSF) and a long-term smoothing filter (LSF). They then verified its practicability. Both debris flow disasters and landslides have the characteristics of sudden occurrence and are closely related to topography, precipitation and geological conditions. However, the main cause of debris flow is the increase in surface runoff caused by sudden rainfall or snowmelt, and the hydraulic action makes the soil and rock mixed materials flow rapidly in the gully, which is very likely to pose a threat to people’s lives and the safety of their property along the entire foot of the slope [9]. Wang et al. [Contribution 10] collected soil samples in the Beichuan mud flow gully region, chosen as the engineering background, and analyzed the spatial distribution pattern and causes of the mutual control factors of stability in the dangerous area of debris flow, which mainly include soil particle size, permeability coefficient, shear strength, porosity, etc., so as to provide a scientific basis for the prediction of debris flow disasters in this area.

Furthermore, some papers have studied the induced behavior prediction of landslides and debris flow under the action of rainfall by using artificial intelligence algorithms. Choo et al. [Contribution 11] adopted the CTRL-T automatic calculation algorithm to obtain the optimal allowable distance between the weather station and the debris flow disaster area suitable for the topography of Korea in order to solve the problem of researchers’ subjectivity in the selection of weather stations in previous studies of debris flow in South Korea, which affected the reliability of the results. A nomogram for sediment disaster risk prediction and early warning was further established and applied to past projects. The results showed that the risk of sediment flow could be predicted 4–5 h in advance. Based on a field test, Xiao et al. [Contribution 3] further optimized a BP neural network by using the particle swarm optimization algorithm; they established a PSO-BP neural network prediction model and compared it with the other two models. The results show that the new model has a higher prediction accuracy in predicting the infiltration depth of the Xiashu loess slope under different rainfall conditions. Yang et al. [Contribution 12] used the decision tree model (GBDT) after gradient elevation of the Google Earth Engine (GEE) cloud platform to conduct a dynamic assessment of landslide risk and a landslide sensitivity analysis of the Three Gorges Reservoir area of China. The research results show that the model maintained a high accuracy in the dynamic assessment of landslide hazards. Subsequently, it can provide theoretical and technical support for real-time landslide hazard assessments and disaster reduction strategies in similar areas around the world.

Finally, in addition to rainfall-induced landslides, some scholars have focused their research on the causes of slope instability and the resolution of uncertainty in slope stability evaluation. Gui et al. [Contribution 13] took a large landslide-prone area in the Central Mountain Range of Taiwan as the research object and adopted multi-temporal satellite and aerial images, field investigations, geophysical tests and other technical means to propose the main trigger factors that induced sudden and local slope instability failure, namely, rainwater intrusion, the rising of river bed elevation and the erosion of large slope foot banks. Li et al. [Contribution 14] regarded the spatial distribution of slope soil shear strength parameters as random and utilized their mean value, variance and correlation scale as characteristics to establish the correlation between relevant parameters and the factor of safety (FS), providing an economical and effective tool for dealing with uncertainty in slope stability analysis.

2. A Summary of the Special Issue

This Special Issue focuses on the multidimensional effects of rainfall-induced landslides, numerical simulation methods and disaster assessment systems. A rainfall-induced landslide is an important type of geological disaster; its triggering mechanism is complicated, and it is affected by many factors such as rainfall intensity, geological structure, terrain slope and so on. The research in this Special Issue covers the physical mechanism of landslide occurrence, the relationship between rainfall intensity and landslide incidence, the construction and optimization of landslide models, and risk assessment models of landslide hazards. Through the combination of physical model tests, artificial intelligence algorithms, numerical simulation, theoretical analysis and other methods, this collection of studies provides a scientific basis for establishing a comprehensive and effective landslide disaster management and emergency response system.

The research results in this Special Issue provide a new perspective and method for understanding and coping with rainfall-induced landslides. Future research can further integrate multidisciplinary monitoring and modeling techniques to achieve accurate landslide prediction and risk control through dynamic acquisition and analysis of real-time data. At the same time, strengthening regional and international cooperation will be a key step in dealing with landslide disasters under climate change.