Introduction

Loess is a porous, weakly cemented quaternary sediment that is mainly composed of silt-sized particles (Pye 1995; Ding et al. 2019). It is widely distributed in Asia, Europe, North and South America (Li et al. 2020; Banak et al. 2013; Marković et al. 2015), and China. Loess covers a total area of 640,000 km2 in China (Peng et al. 2019). Due to its typical collapsibility, large-pore overhead structure, and sensitivity to water, loess is prone to geological disasters (Zhou et al. 2002). Landslides destroy traffic and buildings, block rivers, bury villages, and cause a great deal of human and property losses. It is widely known that earthquakes and rainfall are two factors that induce loess landslides (Havenith and Bourdeau 2010; Peng et al. 2015; Sorbino and Nicotera 2013).

To explore the causes of landslides, many field investigations and experiments have been conducted to analyze the landslide mechanism. Szalai et al. (2017), Szokoli et al. (2018) used electrical resistivity tomography (ERT) and the pressure probe (PreP) method to detect cracks in a slowly moving loess landslide located along the Danube River in South Hungary. Hong et al. (2021) proposed that sliding liquefaction was the cause of the high-speed and long-distance sliding of some homogeneous loess landslides. Through surveys on a landslide that occurred at the village of Zaoling in Xiangning County in 2019, Shi et al. (2020) found that perched water penetrated the soft layer, which decreased the loess shear strength. Ground shaking caused by earthquakes affects the soil structure of loess and promotes fluid circulation and a rapid increase in pore pressure, which leads to a decrease in shear strength (Zhang and Wang 2007). Meanwhile, rainfall infiltration on loess landslides is significantly affected by preferential flows through fissures and macropores (Chen et al. 2018). In this respect, Bai et al. (2014) used a quantitative Antecedent Soil Water Status (ASWS) to study the effects of daily and antecedent rainfall on the triggering of landslides.

Generally, there is an immediate cause for landslides. However, the causes of landslides are very complex. It is difficult to analyze landslides under complex conditions by using investigations and experiments. Finite element analysis has been developed and has become popular for complex slope stability analyses (Azarafza et al. 2021a, b; Fawaz et al. 2014; Xie et al. 2021). Limit equilibrium methods (Azarafza et al. 2021a, b; Avcı et al. 1999; Solov’Ev 2008; Huang 2007) are still widely used in slope stability analysis to solve many difficult problems of slope stability in geotechnical engineering. Combined with the modern computer program, the theory can analyze the slope stability at a faster speed. At present, FLAC (Danneels et al. 2008; Sarkar et al. 2012) and PFC (Ma et al. 2020; Zhang et al. 2018; Yang et al. 2021) softwares are widely used to analyze stability of slopes. Researchers usually adopt Morgenstern-Price (Kainthola et al. 2013) method or Spencer method to solve the critical slip surface to obtain the safety factor. In this case, the cohesion and friction angle of the soil are the key factors to calculate the stability and safety factor of the slope (Chen et al. 2021). However, they have not raised attention on the structural damage of slope soil induced by earthquake and rainfall, which requires the support of scene investigation and experiments. At the same time, recent numerical simulations rarely consider the local damage of the slope caused by rainwater infiltration.

This paper aims to analyze the Changjiahe loess landslide that occurred on 14 September 2019 and to explore the landslide triggering mechanism. Field investigations, experiments, and numerical simulations are used to develop models of the failure mechanism of slope instability that is induced by the effects of earthquakes and rainfall. Based on drone site surveys, landslide phenomena can be observed intuitively. The characteristics of the variations in loess strength under the influence of water and dynamic loads are obtained from triaxial tests. Numerical simulations can restore the original appearance of landslides, and indicate when and where landsides are initiated on slopes. The effects of rainfall and earthquakes on slopes can also be understood. The mechanism of loess landslides that are induced by earthquakes and rainfall effects can be effectively understood by using these approaches. It is especially emphasized that the displacement caused by earthquake and the change of moisture content of slope caused by rainfall infiltration are considered in the analysis of landslide mechanism.

Methodology

The relevant hydrogeological structure data of the study area were collected to conduct field investigation of trough samples and landslides, mainly including structural failure characteristics and topography around the sliding area. UAV aerial survey technology is used to survey and map the landslide to obtain the overall shape and elevation distribution characteristics of the landslide. A high-density resistivity measurement line is laid along the sliding direction of the slope to detect geological information such as the stratum structure and the thickness of the sliding body in the landslide area. It provides information for studying the formation of the landslide.

Laboratory experiments are conducted to explore the influence of water content and dynamic factors on loess strength. The purpose of the test was to determine the strength parameters of unbroken loess and provide data for the subsequent numerical simulations, and only the confining pressure was considered. Thus, three confining pressures were set at 60, 120, and 170 kPa for the same sample. The test includes static triaxial and dynamic triaxial experiments where soil samples are the same. To avoid the uncertain influence of soil moisture content and the structural differences of soil samples from different depths on the test results, loess samples from 6 m below the surface were used in the test. The dynamic test uses constant-amplitude sinusoidal cyclic loading to carry out multistage cyclic loading on the sample. Considering that the maximum frequency of the ground vibrations is between 1 and 5 Hz, the frequencies used in this section are 0.5, 1, and 2 Hz. All experimental conditions were set as consolidation without drainage.

GeoStudio software is used to study the variation characteristics of slope stability considering the factors of rainfall and earthquake. Four modules in the GeoStudio software are used to analyze the landslide. Quake/w is used to load seismic waves. Sigma/w can calculate the displacements of the monitoring points for 180 s under the influence of the earthquake. The water flow characteristics and their changes in time and space can be inferred from the results of Seep/w. The most critical data are obtained from the slip analysis, which are obtained from the Slope/w module. It is important to pay attention to water migration and deformation instability in order to explore the failure mechanism under different working conditions.

Case study

Geological background

Changjiahe town is located in the southernmost part of Tongwei County, Dingxi City, in Gansu Province. The studied landslide occurred on Majiayang Mountain, Changjiahe town. The geographic coordinates of the center point of the landslide are approximately 34°59′46.8″N, 105°11′44.4″E. Tongwei County is located in the Lanzhou-Tianshui seismic zone, a region in which strong earthquakes with PGA values of 0.2 g have been recorded (Fig. 1a). According to the data provided by the China Earthquake Administration, 63 earthquakes occurred in the vicinity of the landslide site from 128 AD to 2021, and there were 22 events with magnitudes ≥ 6.0. A magnitude Ms 7.0 earthquake occurred at the site of the Changjiahe landslide on June 19, 1718 (Sun et al. 2017a, b). Due to the prolonged history of frequent earthquakes in this area, the soil structure here has been loosened, and the region has become highly prone to failure, resulting in slope instability (Fig. 1b).

Fig. 1
figure 1

a Seismic map of Tongwei. b Geographical location of the Changjiahe landslide

Full size image

The region has a temperate continental monsoon climate. The annual rainfall is approximately 400 mm. Precipitation occurs mainly from June to September, with a decrease from December to March. During September 8 and 9, before the landslide occurred, the precipitation in Tongwei County reached 72.6 mm.

Tongwei County is located in the southern part of the Loess Plateau. The terrain is high in the northwest and low in the southeast. Due to the developed water system and concentrated rainstorms, a plateau hilly and gully region with gully development and widespread valley terraces has formed. According to the results of historical geological surveys, the geological structure of Tongwei County can be divided into two types: faults and folds. According to the different origins and morphologies, the area can be subdivided into five geomorphic types, namely, valley basin accumulation plains, landslide accumulation valley plains, deep river valley plains, “V”-shaped valleys cutting loess hills, and loess hills with developed landslides. Due to tectonic uplift, earthquakes, and rainfall, many landslides are distributed in the valley and hilly areas.

Characteristics of landslide

According to detailed field investigations, the Changjiahe landslide in Tongwei County exhibits an armchair shape. The primary strata of the landslide consist of Neogene and Quaternary formations. It has been reported that, based on the history of the landslide site that since 2008, human activities have significantly changed the shapes of gullies and cracks located on the slope (Wang et al. 2020). The number of sinkholes on the slope has also increased.

Tongwei County is an area with multistep agricultural plantations. The elevation at the top of the landslide source area is approximately 1610 m. The length of the landslide is approximately 780 m, and the height difference is approximately 150 m. The maximum sliding distance is approximately 60 m. The landslide has profoundly affected the original topography and landscape, as it has created an abundance of stepped fractures and much ground subsidence. To describe and analyze the deformation and failure characteristics of the landslide in detail, the whole landslide is divided into Sections I, II, and III (Fig. 2a).

Fig. 2
figure 2

Deformation characteristics of landslide. a Investigations of partition map. b The circular-shaped multistage earth wall formed by landslide. c The collapse of an unstable boundary. d X087 Road damage. e Infrastructures severely damaged in Section III

Full size image

In Section I, the main fault wall, with a height of approximately 20 m and slope of approximately 85°–90°, is present at the back edge of the landslide. The overall integrity of the trailing edge platform is poor, whereas the structural characteristics are not clear. The platform area consists mostly of a multistage “earth wall” with a transverse arc and strip, where the soil has good transverse connectivity (Fig. 2b). The orientation of the terrace near the back edge is roughly the same as the direction of the landslide, while that of the lower terrace is the opposite and mostly overturns towards the foot of the slope. When the lower part of the soil moves downward owing to a loss of support, the soil moves in the direction of motion and overturns. The soil in the rear region slides backward due to traction. Part of the southern boundary is affected by the pull-shear action of the landslide. The field investigations determined that some of the steep walls had collapsed.

The deformation and failure degree of Section II are more complex and intense than those in Section I. Several wide and deep longitudinal cracks appeared in this center part of the landslide due to the displacement and expansion of the overlying soil mass (Fig. 2c). Due to the production activities of a nearby brick factory, a steep loess “precipice” approximately 20 m high had developed in the middle of Section II. Although it did not collapse during the landslide, many deep and wide cracks formed due to the downward movement and extrusion of the upper soil layers.

The lower part of Section II contains the most severely damaged infrastructure, and County Road X087 (Fig. 2d) is mainly characterized by “brittle faulting” and “loose disintegration.” The main reason for the two different failure modes is that the soil water content under the subgrade is different. In the southern part of Section II, there is no infiltration from surface runoff, and there are no open channel replenishment systems, while the amount and depth of rainfall infiltration are limited. A slope runoff path is present in the northern part, and a small, closed water storage pit with an area of approximately tens of square meters was located under the road. The water contents of the soil in the northern and southern parts are very different, which influence the strength of the soil.

The lower part is the landslide accumulation area (Section III). Its failure characteristics are pronounced, which include blocking of river channels, shearing of roads, development of broad and deep longitudinal cracks, erosion of bridges, and formation of unstable scarps (Fig. 2e). A bridge is located directly below the main sliding direction, and thus, when the bottom of the landslide shifts considerably, the foundation of the bridge moves, and the bridge collapses. The material carried by the slide rapidly moved into the river, covered the main body of the bridge, and buried the river that flows from north to south. Subsequently, the capacity of the river to accumulate sediments along its channel decreased. Furthermore, many cracks and scarps were formed along the riverbanks.

High-density resistivity surveys were conducted in these areas to provide information regarding the stratigraphic structure. Due to the difficulty of laying out the survey line and the exploration requirements, the survey line is located in the landslide area (Fig. 3a). The survey line is extended along the southern boundary of the landslide in a longitudinal section with a total of 239 electrodes. As shown in Fig. 3b, several high-resistance halo groups are distributed in a range of 20 m below the upper surface. It can be inferred that the surface layer is composed of Quaternary loose loess that was generated by landslides. At a certain depth below the surface layer, there is a loose slip surface. The deep layer comprises the bedrock, which is confirmed to be red bed mudstone. Several low-resistivity patches distributed longitudinally within the middle of the section are presumed to represent entrapment zones that were formed by rainwater infiltration. Midway along the profile, the deep layer exhibits high resistance, which is possibly related to some bedrock protrusions, and they might provide resistance to the landslide process.

Fig. 3
figure 3

Results of high-density resistivity surveys

Full size image

Properties of loess

Basic mechanical parameters of loess

The test loess was taken from the back edge of the Changjiahe landslide in Tongwei County. Several undisturbed loess samples were taken at different depths (e.g., 3, 6, and 9 m) from the original ground. Conventional laboratory geotechnical tests were carried out. The natural density of the undisturbed loess ranges from 1.88 to 2.20 g/cm3. The natural water contents in the 3-m, 6-m, and 9-m undisturbed loess samples are 13.70–14.56%, 16.34–18.94%, and 20.63–23.13%, respectively.

An FG-III photoelectric liquid-plastic limit combined tester was used to measure the liquid-plastic limit of the soil samples. The plastic and liquid limit values of the test loess samples are shown in Table 1.

Table 1 Test soil sample parameters
Full size table

By using a Mastersizer 2000 laser diffraction particle size analyzer, the curves of the cumulative particle size distribution of the test soil were obtained, as is shown in Fig. 4. It is obvious that the loess particle size distributions at different depths are relatively continuous, with clay particles accounting for 6.94 to 11.88% and fine silt accounting for 17.13 to 23.38%. Medium silt accounts for 24.52–25.24%, coarse silt 28.92–24.17%, and sand accounts for 10.58–17.27%. The particle sizes of the samples are mainly those of silt.

Fig. 4
figure 4

Cumulative curve of grain size gradation of loess

Full size image

Triaxial experiments

As shown in Fig. 5, with increasing load, the level of plastic deformation increases, the strain increments are inconsistent under the same stress increment, and the curve develops a nonlinear shape. When the confining pressure reaches a specific value, the stress increases to the peak value, and the deviatoric stress decreases gradually with increasing strain and exhibits a strain-softening phenomenon. The peak stress is between 10 and 18% of the strain, and the peak stress of the curves is not apparent and shows a weak strain-softening phenomenon.

Fig. 5
figure 5

Eccentric stress–strain curve of static triaxial test: a 3 m, b 6 m, and c 9 m

Full size image

Figure 6(a)–( c) show the failure characteristics of the specimen after shear. Under a condition of low confining pressure (60 kPa), there is only one prominent shear zone (single type) in the sample, and the angle between the shear zone and horizontal direction is 50–60°. With an increase in depth (water content), a pressure heave phenomenon is obvious in the deep loess sample. When the depth (moisture content) is constant and the test confining pressure is 120 kPa or greater, two intersecting shear bands (“X” intersecting type) appear in the sample, and both exhibit swelling phenomena. The above experimental results and observations show that the confining pressure and water content of consolidation affect the deformation characteristics of the soil.

Fig. 6
figure 6

a Mohr’s circle envelope. b Soil sample damage of 3 m. c Soil sample damage of 6 m. d Soil sample damage of 9 m

Full size image

The Mohr circle envelope line is shown in Fig. 6d. The deep soil has a high water content and dense texture. After reaching the consolidation standard, the sample continues to retain a high water content, and the pore pressure is generated in the undrained shear mode. The role of the pore water pressure becomes more evident, as the effective stress is adopted as a strength parameter. With increasing depth, the cohesion decreases gradually, while the internal friction angle increases first and then decreases. Meanwhile, with increased water content, the effective strength of the soil decreases.

As shown in Fig. 7, the stress–strain curve for loess under dynamic action is nonlinear. When the consolidation ratios are 2.0 and 2.5, the strain range at the end of the test is between 0.09% and 0.22%. However, when the consolidation ratio is 1.5, the strain range at the end of the test is between 0.15% and 1.0%, and the stress range is between 60 and 90 kPa.

Fig. 7
figure 7

Dynamic stress–strain curve

Full size image

The stress and strain are relatively high under low confining pressures and low consolidation ratios (e.g., σ3 = 60 kPa, Kc = 1.5; Fig. 7a), even if the stress is applied at different frequencies. With increases in confining pressure and consolidation ratio (e.g., σ3 = 120 kPa, Kc = 2.0 and 2.5; Fig. 7c and d), the dynamic stress begins to increase. In addition, it is noted that under low confining pressure and high frequency, less dynamic stress is required to produce the same amount of strain. It can be inferred that the deformation produced by dynamic loading under high confining pressures is less than that under low confining pressures.

Numerical simulation

To understand the mechanism of the Changjiahe landslide, a series of simulations to study the effect of rainfall have been carried out. In addition, the combined impact of earthquakes and rain is also simulated. The differences between the two scenarios are compared to study the effect of the earthquake on the landslide.

Modeling and analytical methods

GeoStudio software was used for numerical simulation analysis by using the data obtained from the experiments. GeoStudio is widely used in landslide analyses (Avşar et al. 2014; Shen et al. 2018; Shole and Belayneh 2019; Wang et al. 2021). In this paper, the seepage, displacement, and extent of the slip surface are calculated by using different software modules. The results obtained under the two scenarios that are caused by rainfall and earthquakes are compared and analyzed. The water content, porosity, elastic modulus, and shear angle of the loess, which were obtained from the soil tests and triaxial experiments, were used in the simulations. The location of the contact between loess and bedrock is consistent with the results of the electrical survey, but it was simplified to some extent because of the limitations of GeoStudio. The modeling results provide insights into the evolution mechanism of the temporal and spatial dynamics of the deformation, instability and slip of loess slopes, and the long-term stability of loess landslides.

The features of the model are shown in Fig. 8. The width of the slope field is approximately 780 m, and the height is approximately 150 m. The slope consists of two parts: the upper layer consists of loess, and the lower layer consists of red mudstone. The two layers are separated by a material interface. A valley river is located at the end of the slope, and there is a slightly sloping water line in the slope. Seven representative nodes on the slope are selected, which are evenly distributed from the shoulder to foot of the slope. The material parameters in the model are shown in Table 2.

Fig. 8
figure 8

Schematic diagram of the numerical model and rainfall effect

Full size image
Table 2 Material parameters of the model
Full size table

Based on studying the deformation characteristics of loess under static conditions, the constitutive model that is used in the rainfall simulation stability calculations in this section is the elastic–plastic constitutive model. The Mohr–Coulomb yield criterion is used as the yield function of the elastic–plastic model, where GeoStudio defaults to a 1-m-long grid cell for the analysis. Limit equilibrium theory is used to analyze the pore water pressure distribution in the slopes to analyze the slope stability and slip surfaces.

To reveal the mechanism of the Changjiahe landslide, a series of simulations of rainfall effects is carried out. In addition, the condition of the combined effect of earthquakes and rainfall is simulated to analyze the mechanism of the Changjiahe landslide. Then, the differences between the two conditions are compared to study the effects of earthquakes on landslides.

Results of the simulation

Rainfall effect

This section shows the slope seepage and deformation levels over 2 days under the rainfall effect.

As stated in the background, the annual average rainfall of the Loess Plateau is 400 mm, and rainfall mainly occurs in July, August, and September, which represents 90% of the annual rainfall. Therefore, the rainfall intensity was set to 30 mm/day, which lasted for 2 days. Rainfall will be applied to the slope surface in the form of boundary conditions (Fig. 8). The slope stability analysis is conducted after 2 days of rainfall.

Figure 9a, b show the seepage vector diagram of the infiltration rainwater seepage rate in the slope on days 1 and 2. Figure 9a shows that the seepage flow is high in small areas of the foot and middle part of the slope surface on day 1. The seepage flow is approximately 0.02–0.08 m3/s. At the bottom of the slope near the river, there is a very small area with high seepage, whose rate is approximately 0.12 m3/s. The seepage rates in many other parts of the slope are negligible. One day later, the seepage rates in the middle and lower parts of the slope increase (Fig. 9b). It is impossible to ignore the seepage in these places, which has a range from 0.02 to 0.05 m3/s. Meanwhile, the seepage field spreads to the depth of the slope. On the one hand, the topography promotes water accumulation in the concave area. This enhances the osmotic pressure, and on the other hand, it is possible that the resistance of the infiltration channel at the base of the slope decreases, which increases the seepage rate.

Fig. 9
figure 9

Results of rainfall effect. a Cloud diagram of seepage rate of rainfall for 1 day. b Cloud diagram of seepage rate of rainfall for 2 days. c Cloud diagram of slip surface

Full size image

Figure 9c shows the variation trend of the slip surface. This figure shows that a large potential slip surface is present from the shoulder to the bottom of the slope. The depth of the potential slip surface at the shoulder is approximately 7 m. The depth of the slip surface at the middle of the slope is approximately 15 m, and at the bottom of the slope, the depth is most extreme, approximately 25 m. This indicates that there is a significant possibility of loess landslide occurrence after 2 days of rainfall. With increasing time, the slip surface will develop deeper, and the possibility of sliding will also increase. After rainfall for 2 days, the safety factor obviously reduces from 1.699 to 1.320. There is a certain risk of landslide currently.

The combined effect of earthquake and rainfall

This section shows the slope displacement, seepage, and deformation over 2 days under the combined effect of earthquakes and rainfall.

Seismic waves are entered into the simulation before the rain effect. In this simulation, the seismic wave is defined as the Tian-Shui wave, whose frequency and amplitude are shown in Fig. 10. The rainfall effect is then entered into the simulation, which is the same as that described in the previous section.

Fig. 10
figure 10

Characteristics of the Tianshui waves. a Seismic wave acceleration. b Frequency spectrum

Full size image

The change curve of the relative displacements of the monitoring points on the slope surface under the rainfall effect and combined effect is calculated. Under the rainfall effect, the displacements of the monitoring points are approximately 0.1–0.2 m. This figure shows that, compared with the rainfall effect alone, the displacement at the top and middle parts of the slope increases significantly under the combined effect of earthquake and rainfall. The displacements at monitoring points 1–5 increase. In particular, the increases at monitoring nodes 3–4 are extremely significant, with displacement increases of approximately 0.4 m.

The results also suggest that compared with the rainfall effect, the combined effect influences the seepage field more extensively, as illustrated in Fig. 11. Thus, after 1 day of rainfall, seepage occurs from the shoulder to the foot of the slope and develops further into the slope (Fig. 11a). A seepage field with a seepage flow of 0.05 m3/s is present at the shoulder of the slope. In the middle of the slope, from the surface to depth, the seepage flow is 0.01–0.03 m3/s. Two days after rainfall, the seepage spreads to nearly the entire slope (Fig. 11b). The area with the highest seepage rate has a seepage flow of 0.09 m3/s. Such spreading would mean that nearly the entire slope would be unstable under the combined effect.

Fig. 11
figure 11

Results of combination of rainfall and earthquake. a Cloud diagram of seepage rate of rainfall for 1 day. b Cloud diagram of seepage rate of rainfall for 2 days. c Cloud diagram of slip surface

Full size image

The displacements of the slip surface under the combined effect are shown in Fig. 11c. Compared to the scenario where only rainfall influences stability, the slip surface is much larger under the combined effect. Under the influence of the combined effect, the potential slip surface extends to the top of the slope, which causes the entire slope to be increasingly unstable. Moreover, deeper slip planes develop in the middle and lower parts of the slope. The potential slip surface at the shoulder increases to 10 m, and in the middle of the slope, it increases to approximately 15 m. Alarmingly, the slip surface at the bottom of the slope is 40 m. This means that nearly the entire loess layer at the bottom is at risk of slipping. Meanwhile, the safety factor of the slope reduces to 0.864, reaching an extremely dangerous situation. Thus, it can be concluded that the damage to slopes caused by an earthquake followed by rainfall can be significantly more extensive.

Discussion

The results of the numerical simulations are in good agreement with the information obtained from the electrical survey. Several low-resistance halo groups are distributed longitudinally on the surface of the middle part of the surface, which are presumed to represent the water entrapment that formed due to the collection of infiltration rainwater. At approximately 30 m below the surface, there is a low-resistance region in the loess, which is inferred to be due to the infiltration rainwater, which may be the primary cause of the landslide. Meanwhile, there are many irregularly distributed high resistivity groups. It can be inferred that there are many fracture zones in the loess and bedrock layers due to the frequent earthquakes in high regions.

When compared to the rainfall effect alone, the loess slope is extremely dangerous when rainfall occurs after an earthquake. The landslide deformation is larger and more extensive than when only rainfall occurs, so the soil mass downward to the river edge also has a considerable landslide depth. From the perspective of the seepage field, after an earthquake, the slope is more conducive to the infiltration and migration of rainwater, and the seepage field is more uniform. When examining the entire slope section, the foot of the slope has the highest rate of rain flow and thus the most damage. In the natural state, the plastic zone mainly appears in the slip zone, slope toe, and rear edge of the landslide. The main failure type consists of shear plastic failure. The maximum relative displacement peak still appears at the shoulder of the slope and gradually decreases towards the foot of the slope.

Before the rains accumulate, the rainwater that infiltrated into the slope directly flows to the lower part of the slope along fissures and slides under the action of gravity. With the continuous rainfall time, the rainfall volume in the infiltration slope body increases and gradually forms a uniform seepage field inside the slope. Water directly enters the deep part of the slope, dissolves the soluble salt components of the loess particles, and forms a water film on the sliding surface, which reduces the friction and anti-sliding ability of the slope. After the earthquake, serious structural damage and loose fractures in the soil mass occur in the slope currently, which provide a good infiltration channel for the rainfall. The seepage effect that was strengthened by the earthquake removes large amounts of soluble salt materials and reduces the cementation of loess particles. The rainfall damage to the slope stability is more significant. For those sections with gentle slopes, the formation of landslides after damage is caused by earthquakes requires more inclined and high-water content strata and geomorphic conditions.

GeoStudio could deal with slope-stability problems, especially when there is a complex combination of factors such as earthquakes and rainfall. This method can easily solve similar slope instability problems. Finite element softwares are easy to use and results were obtained very quickly, which are widely used by many researchers and engineers. However, this method has its own limitations in some ways. The input parameters in the software are determined as these parameters are subject to changes and uncertainties. GeoStudio is a two-dimensional numerical simulation software, which cannot simulate three-dimensional landslide simulation profiles, which means that the characteristics of landslide can only be analyzed from the direction of slope section. With the continuous development of detection facilities and computer technology, the application of finite element analysis in practical engineering has been gradually improved such that the analysis of landslide mechanism will be more comprehensive and in-depth.

Conclusion

In order to analyze a loess landslide occurred in Changjiahe town, triaxial experiments are adopted to analyze the strength of loess under different water content and dynamic loads. GeoStudio is used to analyze the mechanism of Changjiahe landslide where earthquake and rainfall effects were considered. It is concluded that the strength of loess decreases when the moisture content increases. Rainfall infiltration increased the pore water pressure in the slope when the seepage developed to the depth of the slope. Due to cementation loss between loess particles, the effective strength reduced with the moisture content increases. The safety factor of the slope further decreased with the increase of time of rainfall. Loess underwent significant strain softening under high-stress loading conditions. It was also found that the structure of loess was prone to deformation under high dynamic loads as well. Earthquake shaking loosened the loess structure and provided favorable conditions for rainwater seepage within the loess. The strength of loess soils was significantly reduced, which made them more susceptible to slippage and failure. With the combination of earthquake and rainfall, the safety factor of the slope was greatly decreased. Model studies suggested that landslides were mainly concentrated in the middle and lower parts of the slopes, which were consistent with the findings in this field. This paper provides a new method for analysis of landslide mechanism under earthquake and rainfall. This method can be further updated with the development of investigation and finite element analysis techniques.