Groundwater is one of the most important factors affecting the stability of landslides. It has always been the focus of landslide safety research [1
]. Catastrophic landslides are widely distributed and cause great damage. Rainfall, especially extreme heavy rainfall, is the main cause [3
]. As a result of differences in the soil layer permeability inside a slope, there is often an uneven distribution of groundwater. When rainfall occurs, the external replenishment of a slope and the infiltration inside a slope increase, resulting in confined water inside [6
]. Confined water is the groundwater that fills the aquifer between two aquifers, and it plays an increasingly important role in landslide research [7
]. Groundwater is mainly divided into diving and confined water, and the effects of the two on the slope are different. Diving is reflected by the buoyancy, i.e., when calculating the normal stress on the sliding surface, the part of the sliding body below the diving surface is considered according to the buoyancy. The pressurized water is reflected by the upward pressure perpendicular to the sliding surface. The mechanical action mode is more complicated, and it is more sensitive to changes in the front edge permeability conditions. As a result, the impact on slope stability is also more significant. Different permeable layer characteristics determine the action mechanism of different confined water [8
]. When Sun Junjie et al. studied the effect of groundwater on landslides, they did not distinguish the difference between diving and confined water in detail. When considering the distribution of groundwater, it was too simplified and only considered as a diving action, and the final safety factor was too large [9
]. In future landslide stability calculations, diving action should no longer be considered purely, but more attention should be paid to the appearance and influence of confined water.
Under rainfall conditions, the appearance, characteristics and effects of confined water inside a slope play a very important role in the slope’s stability, which has attracted considerable attention from scholars from various countries. Zhang Zhuoyuan et al. identified that the existence of confined water is one of the key factors in landslide formation. In addition, they deduced the critical water head in regard to slope instability for a trailing edge fissure. Confined water was introduced into the stability analysis of landslides [7
]. Through model tests, Zhao Quanli et al. found that the deformation and stress adjustment of gently sloping landslides lag behind the action of confined water. The time lag characteristics of gently sloping landslides with confined water are summarized therein. The concept of the range of action of confined water on the slope was put forward [8
]. Jiao. J. J. identified that confined water, which formed through the influence of heavy rainfall, has a great influence on the stability of the slope. They discussed the importance of correctly establishing hydrogeological models under confined water [10
]. Borgatti. L. et al. analyzed the resurrection of an ancient landslide in a mountainous area in Italy from the perspective of hydrodynamics. It was concluded that confined water caused the reappearance of ancient landslides [12
]. Matja. M. et al. studied two landslides in Slovenia and proposed that the main reason for such landslides is high confined water produced by continuous rainfall [13
]. Tu Yuan et al. constructed a confined water model of a translational landslide affected by rainfall. It is believed that the water head of the trailing edge has a large influence on the range of confined water, seepage flow, and seepage height, eventually causing landslide disasters [14
In the mountainous areas of central and southern China, rainfall is intense and lasts for long periods, increasing the frequency of landslide disasters [15
]. The landslides occur in the rainy season. The main form these take is that of a creeping landslide on a gentle slope on the front of the mountain. The scale and potential damage caused by these is greater. The traditional view is that the gentle slope on front of the mountain is equivalent to a loading berm, which is beneficial to the stability of the slope. However, as a result of the influence of rainfall and weather, there are often many cracks on such gentle slopes, which cause the front edge to slide, and can even cause a large-scale integral landslide involving with the upper mountain [19
]. At present, there is a lack of systematic research and understanding concerning the formation mechanism of such landslides. No corresponding systematic governance measures have been formed.
This paper takes the Fanshantou landslide in Zhejiang, China as an example. The sliding mechanism of this gentle slope was studied from the stratum lithology of the landslide site and rainfall conditions. The results show that the Fanshantou landslide has typical stratigraphic features that are favorable to the formation of confined water, i.e., the upper part of the mountain is steep with a gravel soil layer; the middle and lower parts are widely covered with loose Quaternary sediments, which form a gentle slope on front of the mountain. The surface layer of the gentle slope is dominated by clay, and the middle layers are dominated by a gravel soil layer. A special “permeable binary structure” is formed on the strata. The cracks in the gravel soil layer on the upper part of the mountain were developed under heavy rainfall conditions, which is conducive to the infiltration of precipitation but is basically stable. Rainwater infiltration is small on gentle slopes in the middle and lower parts of the mountain. Groundwater mainly infiltrates into the slope body through the gravel soil layer on the upper part of the mountain; thus, the groundwater table level rises, which facilitates the formation of “temporary confined water”. The upward floating force generated by the temporary confined water causes the creep displacement of the upper slope body, and then, landslides occur. This type of confined water only exists when rainfall occurs and is related to the rainfall intensity and initial groundwater table level. When the rainy season is over, this temporary confined water disappears. This type of slope is stable under non-rainfall conditions, but under continuous rainfall conditions, the slope may become deformed and landslides may occur. Herein, the influences of different rainfall conditions and different initial groundwater table levels on accumulation landslides are analyzed using a numerical simulation.
3. Simulation of the Formation Process of Temporary Confined Water
Analyses of the deformation and failure development of the Fanshantou landslide revealed that the temporary confined water generated by rainfall infiltration was the main factor that induced the landslide. Numerical simulation studies on the formation process of temporary confined water, which is generated by rainfall infiltration, will help us to gain a better understanding of the genesis mechanism of the Fanshantou landslide. First, a numerical simulation of seepage based on the measured daily rainfall data from 2018 was conducted, and a comparison between the calculated and measured values of groundwater was used to validate the numerical simulation method. Then, the same modeling method was used to investigate the conditions under which temporary confined water formed by rainfall infiltration.
3.1. Numerical Model and Its Validation Analysis
Profile 1-1′ in Figure 3
was selected as the study section, and a corresponding A numerical model of slope seepage was constructed, as shown in Figure 7
. The length and height of the model are 470 and 140 m, respectively. In the model, ZK1 and ZK6 are groundwater table level monitoring holes. The numerical simulation process was simulated using Seep/w in GEO-STUDIO 2012 software to study the characteristics of the slope seepage field.
The input hydraulic parameters are given in Table 2
, which include the permeability coefficient, saturated volumetric water content, and the residual volumetric water content. Using the van Genuchten formula to fit the moisture characteristic curve, the residual volumetric water content was assumed to be one tenth of the saturated volumetric water content [20
]. In addition, the model imposed zero flow at the contact between the second layer (gravel soil layer) and the bedrock. It should be noted that the bedrock was regarded as a water-tight layer because the bedrock’s permeability is very low, i.e., set to 0 in simulation.
As input parameters for the simulation, the measured water table level on 31 July 2008 was set as the initial water table level; the saturated volumetric water content and residual volumetric water content were assigned to the same stratum under and above the groundwater table level, respectively. No matter whether there was rainfall or not, the slope surface was the normal passage of water. Under the condition of no rainfall, there was no external infiltration on the slope surface. The idea of simulation was to define the fixed water head at the front and back edge of the landslide area, respectively, and set different periods of rainfall on the surface of the slope (Figure 8
). Rainfall was added to the slope surface in the form of daily rainfall. Rainfall data were actually measured on site. The bedrock boundary of the entire model was impervious, and only the overlying clay layer and gravel soil layer were permeable. As a result of the poor permeability of the clay layer, the external replenishment of the slope can only be considered as the infiltration of gravel soil at the rear of the slope, as shown in Figure 8
. Furthermore, it should be noted that the GEO-STUDIO software simulates the water quantity in the infiltration slope and the surface runoff according to the permeability of the surface rock and soil. Furthermore, the software ignored the impact of the surface runoff and simulated the change in groundwater table level in the slope body for different rainfall intensities and different initial water table levels.
The measured daily rainfall data from 1 August to 31 October 2018 (Figure 6
) were selected to validate the model. The rationality of the model was verified by comparing the calculated groundwater depth with the measured groundwater depth. The measured water table level on 31 July 2018 was taken as the simulated initial water table level and also as the boundary fixed head of the model. The rainfall condition was set as a function of water unit flow and time, and the daily actual rainfall data was input. The depths of water table levels in boreholes ZK1 and ZK6 were simulated for 3 months, and the simulated results were compared with the measured water table level from 1 August to 31 October 2018. The comparison results are given in Figure 9
. The results in Figure 9
reflect the following findings. (a) With respect to ZK1, the calculated values are very close to the measured values at different moments; the majority of differences are less than 1 m, and the mean difference is 0.5 m. (b) With respect to ZK6, the calculated values are also very close to the measured values at different times; the majority of differences are less than 1.2 m, and the mean difference is 0.8 m. (c) The established numerical calculation model and method can reflect the characteristics of groundwater seepage in the Fanshantou landslide and can be further used to simulate the groundwater seepage field of other landslides.
3.2. Influence of Rainfall Intensity in the Rise in Groundwater Table Level
Different rainfall intensities produce different groundwater seepage field change processes. To facilitate the analysis of the seepage field characteristics of the slopes, the following different rainfall intensities were selected: 0, 10, 20, and 30 mm/d; 5 days was chosen as the continuous rainfall duration. The seepage field variation law under different rainfall intensities was studied under the same initial groundwater table level conditions.
The final groundwater table levels and the initial groundwater table level are given in Figure 9
for 5 d of rainfall with different intensities (0, 10, 20, and 30 mm/d). The results of Figure 10
show the following: (a) Before rainfall, the groundwater table level in the slope remains below the bottom of the clay layer, i.e., before rainfall, there is no confined water in the slope, and the groundwater is phreatic water. (b) During rainfall, the groundwater table level rises gradually, but the rate of increase and the range of the groundwater table level are different under different rainfall intensity conditions. (c) Under the 10 mm/d rainfall intensity condition, the groundwater table level of the slope never reaches the bottom of the clay layer, and confined water never forms. Therefore, under the natural conditions of low and moderate rainfall, groundwater always occurs in the form of phreatic water, and no confined water forms. (d) Under the 20 mm/d rainfall intensity condition, the groundwater table levels are higher than those of a rainfall intensity of 10 mm/d; after 72 h of rainfall, the groundwater table level near ZK6 in the front of the slope rises above the bottom of the clay layer, which means that confined water forms locally; with the continuation of rainfall, the scope of the confined zone gradually expands. (e) Under the 30 mm/d rainfall intensity condition, the most obvious characteristics are that a larger area of confined water, a higher pressure head of confined water, and the seepage area of groundwater appear; a field investigation also found that after heavy rainfall, a large number of spring seepages appear in the corresponding area; after 24 h of rainfall, the groundwater table level at ZK6 in front of the slope rises above the bottom of the clay layer, which means local confined water is formed; after 72 h of rainfall, the groundwater table level reaches the surface. (f) With the increase in rainfall intensity, it is inevitable that confined water will develop and cause frontal seepage.
3.3. Influence of Initial Water Table Level Condition on the Rise in the Groundwater Table Level
The initial groundwater table level condition reflects the accumulation of precipitation in the early stage. The influence of the initial groundwater table level conditions on the rise in the groundwater table level can be studied to reveal the extent of the influence of early rainfall accumulation on landslide stability. The initial groundwater level is the diving level. This is the normal water level inside the slope without rainfall. Under normal circumstances, the rise and fall of groundwater is not large, mostly within 2 m. For this reason, three models were analyzed. The measured initial water table level, the measured initial water table level +2 m, and the measured initial water table level −2 m were used as the initial water table levels of the three models (Figure 11
). In addition, 20 mm/d was selected as the rainfall intensity, and 5 days were used for the duration of the rainfall in the three models.
The modeling results are given in Figure 12
. With respect to the accumulation caused by the permeability binary structure, the influence of rainfall on slope stability is closely related to the initial groundwater table level of the slope. When the initial water table level is low, it takes a long time for the groundwater table level to rise, so confined water does not easily form. When the initial water table level is high, the superposition of rainfall infiltration quickly produces greater volumes of confined water. Generally, the higher the initial water table level, the greater the confined water pressure produced by the rainfall, and the greater the threat to slope stability.
3.4. Influence of Intermittent Rainfall
After discussing different rainfall intensities and different initial water levels, the intermittent rainfall situation was selected for a more realistic simulation. Choose the more distinctive real rainfall data from September 7th to September 13th to simulate the change of groundwater level.
Rainfall is shown in Table 3
The modeling results are given in Figure 13
. It can be seen from the figure that continuous rainfall occurred on September 7–8, and the water level at the rear edge of the slope rose rapidly. The water level in the rear reached its highest value on September 9, when the water level in the front hardly rose. With the transportation and flow of groundwater in the slope, the front water level reached its highest value on September 11, and temporary confined water appeared. At this time, the rear water level had fallen. When it reached September 13, the water levels at the front and rear had fallen, and the slope tended to be stable. This simulation shows that rainfall has a greater impact on the groundwater level of the slope, and the response time of the front and rear edges is different.
3.5. Analysis of Slope Factors of Safety under Different Conditions
In order to evaluate the stability of the landslide under the different simulation conditions discussed above, the limit equilibrium method was used. The module SLOPE/W in the GEO-STUDIO software was used, and the Morgenstern–Price method was selected. The groundwater table levels were directly imported, which are the simulated results above from Figure 9
, Figure 10
, Figure 11
and Figure 12
. The input parameters for stability analysis are given in Table 4
In the process of calculation, for the gravel soil layer with good permeability in the lower part, saturated and natural parameters were assigned to the same formation below and above the groundwater table, respectively. For the clay layer with poor permeability in the upper part, as a result of the action of confined water, natural parameters were selected for the stratum between the water-resistant roof and the water surface. The bottom surface of the gravel soil layer was selected for the sliding surface. It should be noted that the groundwater table levels under the different simulation conditions discussed above were obtained using the Seep/W module in GEO-STUDIO.
The factors of safety obtained under different rainfall conditions are plotted in Figure 14
. Because the initial water table level conditions were the same, the initial factors of safety of the slope are the same. Figure 14
clearly shows that with the occurrence of rainfall, the slope’s factor of safety decreases to different degrees. When the rainfall intensity is 10 mm/d, the factor of safety decreases a little but is always larger than 1. When the rainfall intensity increases to 20 mm/d, the temporary confined water is formed gradually. The factor of safety gradually begins to drop below 1, and the slope becomes dangerous. When the rainfall intensity increases to 30 mm/d, the confined water begins to form more quickly and the factor of safety continues to drop below 1. Here, the slope is a safety issue. Therefore, rainfall intensity has a great influence on the stability of this type of slope.
The factors of safety obtained under different initial groundwater table level conditions are plotted in Figure 15
. It clearly shows that: (a) For different initial groundwater table levels, the factor of safety decreases with the increase in rainfall duration. (b) The higher the initial water table level, the lower the factor of safety, and the easier it is for the slope to lose stability. (c) The factor of safety can be kept in a relatively safe range when the groundwater remains in the form of diving water in the permeable layer. When the groundwater crosses the bottom of the upper clay layer and produces temporary confined water, the factor of safety drops further, and the stability is greatly reduced.
The factors of safety obtained under intermittent rainfall conditions are plotted in Figure 16
. Simulate the factors of safety of the slope under different times and different rainfall conditions. As the rainfall progressed, the factor of safety began to decrease due to the rising water level at the rear of the slope. As time progressed, the front groundwater level began to rise, temporary confined water began to appear, and the rate of decrease in the factor of safety accelerated. When the rain stopped for a period of time, the water levels at the front and back dropped, and the factor of safety continued to increase, returning to a safe state. This also proves that temporary confined water only appears under rainfall conditions, threatening the safety of slopes.
On the basis of a field survey and monitoring data, this study explores the phenomenon of temporary confined water produced by gentle slopes on front of mountains due to their special dual stratum structure. The permeability of the surface clay of gentle slopes on the front of mountains is higher than the lower gravel soil layer. Under rainfall conditions, a large amount of rainwater infiltrates the supply area at the rear edge of the slope, and a drainage channel is formed inside the gentle front slope. As a result of the narrow discharge area, the gentle slope becomes an area that holds temporary confined water. Confined water forms easily and can remain for a relatively long time. The temporary confined water produced by the dual structure inside the gentle slope of the mountain is considered to be the key factor in inducing landslides. This kind of temporary confined water only appears when the rainfall intensity is sufficient, thus causing instability. GEO-STUDIO software was used to simulate the changes in the groundwater table level under different rainfall intensities and different initial water table levels. The simulation results clearly show that temporary confined water may appear under a certain combination of rainfall intensity and initial water table level. On the basis of the simulated groundwater table level, the slope stability was analyzed. It is obvious that the emergence of temporarily confined water has seriously threatened the safety of the slope. This kind of temporary confined water produces a larger floating force at the junction of impervious layers, causing cracks in and deformation of the slope. This is the fundamental reason why landslide disasters occur on gentle slopes on the front of mountains.
The landslide body is made up of different soil bodies with large differences in permeability, with spatial variability. The permeability depends on the composition of the fine content. This study does not consider these variabilities and uses the mean values as input parameters (Table 2
). This may be the source of errors in the simulation results. Incorporating spatial variabilities of strata into the seep simulation may be considered in future research. In addition, future research can select more parameters for more numerical simulations. Get more specific conclusions with more pictures.