Field Monitoring-Based and Theoretical Analysis of Baota Mountain Landslide Stability

Landslide is one of the most widely distributed surface morphological landscapes, and it can cause a series of major economic and human losses. Field monitoring and limit equilibrium methods were applied to investigate Baota Mountain landslide stability, and soil volumetric water content, different scales of rainfall data, and landslide displacements were monitored using various equipment. The theoretical factor of safety was also calculated for the landslide. Finally, the theoretical results were validated by monitoring data in the field. The results demonstrate that soil volumetric water content experienced the greatest change with time at a depth of 0.2 m and then 1 m; however, the change in soil volumetric water content was relatively small with time at a soil depth ranging from 2.0 m to 4.0 m. Soil volumetric water content also did not change with time at a soil depth of 5.0 m and below. In addition, the retardation effect was found in different depths of volumetric water content for continuous rainfall. The safety factors were 2.713 and 1.133 for landslide No. 1 and landslide No. 2, respectively. These results indicate that landslide No. 1 is relatively stable, but there is a probability of the occurrence of movement in landslide No. 2. The monitoring displacement data indicate that landslide No.1 was in a relatively stable state between 2008 and 2013, and this result was in accordance with the value of theoretical calculation. This study provided relevant parameters for numerical simulation of landslides in loess areas.


Introduction
Landslides are one of the most widely distributed surface morphological landscapes and can cause a series of major economic and human losses [1][2][3]. Landslides also constitute the most important disaster type that can influence surface geomorphology processes [4,5]. erefore, research that focuses on landslide stability is both significant and worthwhile [6,7]. e landslide stability problem has been developed for evaluating the safety of hillslopes [8]. Some researchers studied and discussed current methods, which included physical models, field monitoring, and numerical analysis for slope stability analysis in detail [9][10][11]. When physical models were used to simulate slope stability, the soil structure usually changed during the construction of artificial slopes [12]. Some researchers also utilized field surveys and the freeze-thaw cycling theory to elucidate the mechanisms of natural and artificial slope failures in the Loess Plateau, China [13,14]; however, these approaches cannot directly reflect the deformation processes of the landslide prototype. Subsequently, a series of pieces of equipment have been widely applied to monitor landslide displacement, which constitute direct and efficient ways to determine landslide stability and calculate its threshold, especially for single landslides [15,16]. Field monitoring of landslide stability from 2013 to 2016 in the ree Gorges Reservoir, China, was performed with different types of equipment, and the results demonstrated that landslide deformation reverted to seasonal stepwise cumulative displacements influenced by cycles of reservoir drawdown and rainfall [11]. It is worth noting that the monitoring data could be obtained each hour and in even shorter units of time. Tecca et al. (2003) applied various apparatuses to measure pore water pressure and volumetric water content for a landslide in Acquabona, Italy, and found that an instantaneous and vertically upward pore water pressure and volumetric water content were increased in a shallow slope [17]. Uhlemann et al. (2016) assessed ground-based monitoring techniques for landslide investigations in Hollin Hill landslides, UK, and demonstrated that the global positioning system (GPS) network of markers can offer a higher temporal and spatial resolution for landslide stability analysis [18]. Vassallo et al. (2020) implemented a system of inclinometers and GPS to determine landslide displacement in southern Apennines, Italy [19]. e above-mentioned achievements indicated that field monitoring and measurements are highly effective ways to obtain the stability of single landslides. However, monitoring of volumetric water content regarding landslide stability was performed relatively less frequently; in fact, volumetric water content is a key factor to analyze landslide stability [20]. With technological advancements, a large number of numerical methods were applied to calculate the stability of landslides. e basic approach is to determine a safety factor or stability threshold to prevent sliding failure of a given slope, but these methods vary markedly in simplification and accuracy [21,22]. Meanwhile, conventional methods of slope analysis based on the concept of limit equilibrium have been broadly implemented, due primarily to their simplicity and applicability [8]. e main difference was in assumptions made about the shape of the slide surface (circular, plane, and logarithmic) for different limit equilibrium methods, and equilibrium equations were applied [9]. Subsequently, the slip surface would be divided into a series of slices with the process of computation, and the implemented limit equilibrium methods included simplified Bishop's, Janbu's, Spencer's, and Morgenstern-Price methods [22]. ese methods are commonly employed to calculate the safety factor of slope stability in geotechnical engineering applications [23]. Zolkepli et al. (2019) used the modified Fellenius' and Bishop's methods to calculate the safety factor of slopes located at Pahang Matriculation College, Malaysia, and the results showed that the theoretical calculated values were highly similar to the monitoring data [24]. Limit equilibrium methods were also carried out to simulate the whole process of a landslide, particularly in the large deformation of soils during the landslide [22]. Above all, the values of landslide stability can be obtained conveniently by limit equilibrium methods. e aim of this study is to determine the stability of Baota Mountain landslides for Yan'an City, Shaanxi Province, China. Field monitoring and limit equilibrium methods were used to study landslide stability, soil volumetric water content, and landslide displacements. In addition, the theoretical factor of safety was calculated for landslides. Finally, the different results were compared, and the theoretical results were validated by monitoring data. is study provided relevant parameters for numerical simulation of landslides in loess areas.

Baota Mountain
Landslide. Baota Mountain is located in the Baota District, Yan'an City, Shaanxi Province, in the northern of part of Xi'an City (Figure 1(a)). e study area is approximately 320 m long and 30-50 m wide and constitutes typical loess beam topography, which is surrounded by deep valleys and gulleys. In addition, Baota is located at the top of Baota Mountain (Figure 1(b)) and was built during the Tang Dynasty. Baota is 44 m high, with nine floors, and is a wellknown symbol of the historic Yan'an City.
ere are two landslides (landslide No. 1 and landslide No. 2) on Baota Mountain. Landslide No.1 is distributed in the western part of the loess slope and is 85 m long from east to west and 95 m wide from north to south, and its main sliding direction is NW 65°. e maximum thickness of the landslide is approximately 20 m, and its total volume is 9.4 × 10 4 m 3 .
e elevation of the trailing edge of the landslide is 1,040 m, and the escarpment of the trailing edge ranges from 1 m to 3 m, within 2.5 m-4 m and 1 m-2 m in the north and south, respectively. erefore, a concave slope with three sides of steep slope was formed, which exhibits typical landform characteristics of landslides. e shear outlet of the landslide is 974 m, and the height of the slope is approximately 66 m. e surface morphology of the landslide alternates with steep slopes and gentle slopes. Vegetation is comparatively better in the sliding area, especially in the middle of the landslide (Figure 1(c)).
Landslide No. 2 is distributed in the eastern part of Baota Mountain. It is a small landslide that was developed in loess areas. e elevations of the trailing and the leading edge are approximately 1,040 m and 1,015 m, respectively, and the main sliding direction is NE 80°. e landslide is approximately 17 m long from east to west and approximately 25 m wide from north to south, and its total volume is approximately 975 m 3 . ere are a series of steep walls in the three sides of the landslide (north, west, and south), within 1 m and 3 m high. An artificial retaining wall 2 m-3 m high is in the front of the landslide (Figure 1(d)). e soil mass of the landslide is fractured into pieces, and many fissures exist. Moreover, the vegetation is well developed on the landslide, and it is in a generally stable state. However, the slopes are becoming increasingly steep due to the construction of roads, which has led to a rising risk of local slides.

Geologic and Climatic Settings.
e lithology of Baota Mountain is Jurassic sandstones underlying Quaternary loess, which are interspersed with lenticular sand-gravel layers of Early Pleistocene in the middle (Figure 2(b)). e maximum thickness of the Upper Pleistocene loess is 12.5 m, among which the components of particles of the Upper Pleistocene loess are largely silt, and with a loose structure and large pores [25,26]. e maximum thickness of the Pleistocene loess is 63.6 m, with relatively high clay content, and the average thickness of the loess is between 20 m and 30 m. e thickness of Jurassic sandstone is relatively thick in the Yan'an region. e top surface of the basement rock of Baota Mountain is a pale-eroded terrain that slopes slightly northeast. ere is no groundwater within 5 m below the bedrock surface.
e study area is located in the Northwest part of China, exhibits a semihumid and semiarid continental monsoon climate, and has a mean annual rainfall of 562 mm. Rainfall mainly occurs in the period from June to September, accounting for approximately 70% of the total annual rainfall, and thundershowers occur frequently. Daily and hourly maximum rainfalls were observed in 1981 (∼139.9 mm) and 1979 (∼62 mm), respectively [27]. In addition, flash and prolonged heavy rainfall events often take place in the region, thereby triggering a series of geological disasters, especially landslides and debris flows.

Methods
e limit equilibrium method was selected to calculate landside stability [28], and field monitoring was applied to obtain landslide displacement. Subsequently, the results of theoretical calculation were validated by the results of field monitoring.

eoretical Calculation of Landslide Stability.
e limit equilibrium method is a crucial method to calculate the stability of landslides [29][30][31]. Although some theoretical shortcomings exist in this method, it has been improved after a series of researches [32,33]. For example, the modified Bishop's method is slightly different from the ordinary method of slices in normal interaction forces in that the adjacent slices are assumed to be collinear and the resultant interslice shear force is zero [29]. e constraint introduced by normal forces in slices makes the problem statically indeterminate. As a consequence, iterative methods should be utilized to solve the factor of safety. e method has been shown to produce a safety factor within few percents of the "correct" values. F s for moment equilibrium in Bishop's method was given by Bishop (1955) [28]: where F s is the safety factor; c i ′ is the cohesive force; φ i ′ is the internal friction angle; α i is the slope angle of the bottom of each slice; b i is the width of each slice; W i is the weight of each slice; μ i is the water pressure at the base of each slice; ζ i is the slice index; e i is the horizontal interslice force; and R is the radius of the arc face. e parameters were obtained from drilling data and experimental results, the specific parameters of which are presented in Table 1.    , and the maximum number of convergent iterations was 5,000. e boot range of the sliding surface was set at the top loess layer of Q 3 and Q 4 , and the range of roll-out was set at the foot loess layer of Q 4 for landslide No. 1. With respect to landslide No. 2, the sliding surface was distributed in the top loess layer of Q 3 , and the range of roll-out was set at the foot of the loess layer of Q 2 . Groundwater was not considered in the process of slope stability calculation.

Field
Monitoring. An integrated monitoring system was built and installed during the test pile construction. e monitoring system comprised displacement monitoring  Advances in Civil Engineering 5 stations, and soil volumetric water content, inclinometers (ZK), soil volumetric water content sensors (TDR) in the boreholes, and rain gauges were fixed to the test piles ( Figure 2). e monitoring system consisted of displacement monitoring stations and a soil volumetric water content monitoring system, which includes inclinometers (ZK) and soil volumetric water content sensors (TDR) in the boreholes, and rain gauges were fixed to the test piles.
In 2007, three displacements monitoring sensors (ZK) were installed in landslide No. 1 (Figure 2(a)), the specific information for which is shown in Table 2 and Figure 2(b). e resolution of the equipment was 0.015 mm/h, and the monitoring period was from September 2008 to May 2013. Eight soil volumetric water content sensors (TDR) were fixed in landslide No. 2 (Figure 2(a)), the detailed information of which is presented in Table 2 and Figure 2 Figures 3(b) and 3(c). e results of changes conformed to the natural phenomena, which indicated that the data were relatively reliable. e monitoring data of TDR 2-1 gradually increased during heavy rainfall, and the increase of VWC was earlier when heavy rainfall became greater. e monitoring data of TDR4-1 are shown in Figure 3(a). e VWC was at its maximum value among all monitoring points, because TDR4-1 was located at the top of landslide No. 2. e VWC of TDR 2-1 is shown in Figure 3(b). e soil VWC increased after 5 h of rainfall; however, the soil VWC of TDR 2-1 increased from 2 h after the end of the rainfall (Figure 3(c)), indicating that the changes of VWC were delayed in response to rainfall.

Daily Soil Volumetric Water Content.
e daily change in soil VWC in different monitoring points of landslide No. 2 is presented in Figure 4. is monitoring was representative in loess regions, because the monitoring point of VWC was distributed in a natural slope [34]. e soil VWC was obviously affected by daily rainfall when the soil depth ranged from 0 m to 1 m ( Figure 4); however, the VWC was stable when the soil depth was more than 2 m. In addition, the VWC of surface soil (0.2 m) was slightly increased when the daily rainfall was greater than 10 mm, but the VWC was unchanged in the soil depth of 1 m. e VWC obviously increased above the soil depth of 1 m when daily rainfall reached up to 50 mm, but the VWC was stable at the soil depth of more than 2 m.
Continuous rainfall was selected as the research object. Using monitoring data from 28 August to 26 September, the retardation effect concerning different soil depths of VWC for rainfall was analyzed. e correlation coefficient between VWC and rainfall reached a maximum in the soil depth of 0.2 m after rainfall occurred for approximately 16 d (Figure 5). With respect to the soil depth of 1 m, the correlation coefficient achieved the maximum when rainfall occurred for approximately 24 d.

Monthly Soil Volumetric Water Content.
e relationship between monthly rainfall and VWC was constructed based on the results of the hourly and daily monitoring data. e VWC increased with the increasing of rainfall, and the retardation effect still occurred. In addition, dissimilarities of the VWC peak in different years were obvious (2008-2013). For example, the maximum peak of VWC occurred in August and September in 2009 and 2010, respectively. Dramatic differences of the soil VWC were also evident for different monitoring points and soil depths ( Figure 6). e results demonstrated that the changing of soil VWC was more obvious for soil depth, which was relatively lighter with increasing rainfall. However, the changing in soil VWC was relatively stable for relatively deeper soil depth. Meanwhile, the VWC was affected by rainfall at the soil depth of 0.2 m, and a positive correlation relationship was obvious. e correlation coefficient, however, gradually decreased with increasing soil depth.

Volumetric Water Content in Different Soil Depths.
e changing trends of VWC were essentially the same under different times and different soil depths on the monitoring point (Figure 7). e VWC fluctuated at different soil depths on the TDR2 monitoring points, and there were three wave peaks in the soil depths of 0-8 m, which were distributed in the 3.0 m, 5.0 m, and 7.0 m soil depths, respectively. Furthermore, the changing situation of soil VWC could be summarized based on the monitoring results. e largest change of soil VWC with time occurred at the depth of 0.2 m and then 1 m. However, this change of soil VWC was relatively small with time at the soil depth ranging from 2.0 m to 4.0 m, and the VWC did not change with time at the soil depth of 5.0 m and below. ese results indicated that the soil VWC was obviously influenced by rainfall, especially in the surface soil, and these results were in accordance with extant literature. [35][36][37] In addition, a series of cracks were found on the Baota Mountain ( Figure 8). ese cracks would provide an advantageous channel for rainfall infiltration, resulting in an acceleration of the rate of rainfall infiltration and ultimately forming underground runoff [38]. is would also provide a conducive condition for the occurrence of landslide disasters. Indeed, a large number of investigations have reported that such channel can impart a marked effect on slope stability, thus leading to the occurrence of landslide disasters under the induction of rainfall [39,40]. Consequently, the probability of a landslide in landslide No. 2 was relatively large. 6 Advances in Civil Engineering  Figure 9). Landslide No. 1 is relatively safe based on the division standard of Fs [41]; however, landslide No. 2 is comparably instable. is result is consistent with the results of Section 4.1, in which the probability of a landslide in landslide No. 2 with continuous rainfall or heavy rainfall is relatively large. Salmasi et al. (2019) investigated the relationship between soil material parameters (i.e., cohesive force, internal friction angle, soil specific weight, and constant slope height) and safety factor. e results showed that cohesive force and internal friction angle impart an obvious effect on the safety factor, and this safety factor gradually decreased with the slow diminishing of cohesive force and  Advances in Civil Engineering internal friction angle [42]. In this study, the values of cohesive force and internal friction angle in landslide No. 1 were apparently higher than those of landslide No. 2, and thus the Fs was different.

Monitoring Displacement Results
(1). Different Monitoring Points Displacement. Overall, the landslide displacement was almost 0 m in different monitoring points based on the daily and monthly field monitoring data (Figures 10 and 11) in landslide No. 1; and a small-level displacement occurred in part of the monitoring points, with displacement values of less than 0.05 mm.
Furthermore, landslide displacement was generally unaffected by daily rainfall (Figures 10 and 11). ese results indicated that landslide No. 1 is in a relatively stable state, which was congruent with the theoretical calculation results.
(2). Different Depth Displacement. In general, the changing trends of displacement were essentially the same at different times and different depths of the monitoring points (Figure 12). e displacement fluctuated with different soil depth in the ZK1 monitoring point, and two wave peaks appeared in the soil depth of 0 m-35 m. However, the value of displacement was minimal, that is, less than 0.1 mm, between 2009 and 2012. In addition, a correlation relationship was not found between the displacements of Daily rainfall (mm) 08     different depths and rainfall. ese findings also indicated that landslide No. 1 was comparatively stable, and the results of the theoretical calculation were again verified to be accurate.
In general, the Baota Mountain landslide was stable, based on the results of monitoring data between 2008 and 2013. However, the closest distance was approximately 19 m between Baota and landslide No. 2. Moreover, a series of joints and fissures developed on the surface of the loess soils (Figure 8), and surface spalling always occurred. Shallow landslides will occur under the influence of continuous or heavy rainfall and once a landslide takes place, Baota would be threatened. is result was similar to the findings of other investigations [26]. erefore, it is urgent to take certain protective measures to prevent Baota Mountain landslides.

Conclusion
Loess landslides are widely distributed across the loess regions of China, and landslide disasters are common due to heavy or continuous rainfall in this region. We applied field monitoring and limit equilibrium methods to estimate landslide stability.  impact on VWC with the soil depth from 0 m to 1 m. However, the VWC was stable when the soil depth was more than 2 m. In addition, the VWC of surface soil (0.2 m) slightly increased when the daily rainfall was greater than 10 mm; for the soil depth of 1 m, however, the VWC was unchanged. VWC obviously increased above the soil depth of 1 m with daily rainfall of 50 mm, while VWC was stable when the soil depth was more than 2 m. Moreover, the retardation effect regarding VWC in different soil depths for rainfall was obvious.     Figures S4 and S5).
e Fs values were 2.713 and 1.133 for landslide No. 1 and landslide No. 2, respectively. is indicated that landslide No. 1 was relatively safe, while the probability of a landslide in landslide No. 2 was greater. e results of monitoring displacement indicated that landslide No. 1 was relatively stable between 2008 and 2013, which was consistent with the value of theoretical calculation.
Data Availability e original data are available upon request.

Conflicts of Interest
e authors declare that they have no conflicts of interest.