Vertical acceleration effect on landsides triggered by the Wenchuan earthquake, China

Abstract

The 2008 Wenchuan earthquake with Ms8.0 triggered extensive throwing-pattern landslides in the area within or near the seismic faults. The resultant landslides from this earthquake brought to the fore the effect of vertical earthquake acceleration on landslide occurrence. The pseudostatic analysis and the dynamic response on landslide stability due to the Wenchuan earthquake are studied with the Chengxi (West Town) catastrophic landslide used as a case study. The results show that the epicenter distance is an important factor which affects the vertical acceleration and thus the stability of landslide. Also, the vertical acceleration was found to have a significant impact on the FOS of landslide if the earthquake magnitude is quite large. Within the seismic fault, the amplitude effect of vertical acceleration is very dominant with the FOS of landslide, for vertical acceleration ranging from positive to negative, having a variation of 25 %. The variation of FOS of landslide for vertical acceleration ranging from positive to negative are 15 and 5 % for landslides near seismic fault and outside seismic fault, respectively. For landslide with a slope angle <45°, the FOS of landslide with both horizontal and vertical accelerations is significantly greater than the one without vertical acceleration. Further, the results computed from both the pseudostatic method and dynamic analysis reveal that the FOS during the earthquake varied significantly whether vertical acceleration is considered or not. The results from this study explain why lots of throwing-pattern catastrophic landslides occurred within 10 km of the seismic fault in the Wenchuan earthquake.

Introduction

The 2008 Wenchuan earthquake with Ms8.0 triggered extensive landslides in the area of the Mt. Longmenshan, which is a narrow and steep valley. The characteristics of the triggered landslides have been studied by many scholars (Huang and Li 2009; Wang et al. 2009; Yin 2010). Huang and Li (2009) summarized that the distribution of the landslides associated with this earthquake was a zonal distribution along the coseismic fault and linear distribution along the rivers. The distribution had a close relationship with the hanging wall effect of the seismic faults, steep slope of topography, and lithology. The throwing and long run-out characteristics of the landslides in this area were very unique. They were quite different from conventional landslide events due to the topography, heavy rainfall and associated moderate earthquake aftershocks. Yin et al. (2009) classified the failure planes of the landslides triggered by the Wenchuan earthquake, based on their overall shapes, into three types, namely concave shaped, convex shaped, and staircase shaped. They also examined the effect of vertical earthquake acceleration on the landslides within or near the seismic faults.

The stability and permanent displacement of a slope subject to combined horizontal and vertical accelerations have been researched, and the results reveal that vertical acceleration is large (Ling et al. 1997). The vertical movement and the anomalously high concentrations of some earthquake landslides in and adjacent to the seismic fault were also documented (Coller and Elnashai 2001; Harp and Jibson 2002).

Few studies have however been conducted on the vertical acceleration effect on landslide stability in Wenchuan earthquake area. This paper performs a pseudostatic analysis and studies the dynamic response on the stability of landslides caused by the Wenchuan earthquake using the strong ground motion data recorded at the Qingping Station. The Chengxi catastrophic landslide (Fig. 1) at ruined Beichuan town is used as a case study.

Fig. 1

Sketch map of density of landslides triggered by Wenchuan earthquake (also indicates the location of the Qingping station for strong ground motion records and the Chengxi landslide at ruined Beichuan town)

Acceleration characteristics of landslides triggered by Wenchuan earthquake

The vertical earthquake force effect on landslide failure and reinforced concrete (RC) structure has been studied by lots of researchers (Anderson and Bertero 1973; Qian 1983; Elnashai and Papazoglou 1997; Zhou et al. 2003; Cui et al. 2010). Figure 2 shows the Taihongcun landslide in Beichuan County. It is a typical earthquake landslide which occurred within the main central seismic fault and has the characteristics of the staircase or terraced failure plane (Yin et al. 2009). This landslide consists of sandstone, shale, and slate of Silurian age. The upper part of the landslide body has a height of 200 m, a longitudinal length of 150 m, an area of 18,000 m² in longitudinal direction, and a width of about 50 m, resulting in a volume of 0.9 million m³. It is believed that the upper sliding body failed first and was thrown upward by the earthquake force, returned and collided forcefully into the underlying bedrock and triggered the lower landslide. The dislodged rock material from the head scarp, forming an impacted erosive mass of chipped soils and rock of about 20,000 m³ volume and showing characteristics of an air blast, buried a cornfield. The dimensions from the upper failure terrace to the lower impact slope are about 300 m in height, a longitudinal length of 480 m, an area of 26,000 m² in longitudinal direction, and a width of 100 m, resulting in a volume of 2.6 million m³ due to entrained erosion. The thickness of the landslide deposit is about 50 m forming a massive landslide dam of about 3.5 million m³ in volume from the source and the entrainment.

Fig. 2

Section map of a typical throwing-pattern landslide triggered by Wenchuan earthquake—the staircase-shaped rockslide, the Taihong village, Beichuan (Yin 2010). a Photo of Taihong landslide. b Section map of Taihong landslide

The above effect of the vertical seismic force during the Wenchuan earthquake in the meizoseismal area has been discussed. Wang et al. (2010) conducted research work on the relationships between ground motion parameters obtained from 33 strong ground motion stations in the Wenchuan earthquake area, and that obtained within 500 km from the main seismic fault (Fig. 3). They observed that, when the peak ground acceleration (PGA) reaches 2 m/s², the landslide hazards were very intense. Also, after a threshold of the ground motion intensity, they found the landslide density fluctuates as the ground motion intensity increases. Generally, the acceleration values could be divided into three categories based on the distance of the earthquake landslide to the seismic fault (Yin 2010). These categories are:

Fig. 3

Curves of the vertical and horizontal accelerations vs. epicenter distance to seismic fault from the Wenchuan Earthquake (after Yu et al. 2009)

• (a) Landslides within seismic fault These are landslides which occur at a distance 0–5 km from the seismic fault. For this category, the ratio of the vertical to horizontal components of the acceleration is >1, i.e., Va/Ha ≥1.0, according to the strong ground motion records. Most of the major landslides, especially, long run-out rockslide and avalanche occurred in this area. The vertical acceleration is dominant to trigger the staircase-shaped landslides.

• (b) Landslides near seismic fault This category is for landsides occurring at a distance of about 5–10 km from the seismic fault. The ratio of the vertical to the horizontal components of the acceleration varies from 0.5 to 1.0, i.e., Va/Ha = 0.5 ~ 1.0, according to the strong ground motion records. Landslides are widely triggered, but, few staircase-shaped and long run-out rockslide and debris occurred in this area.

• (c) Landslides outside seismic fault These are landslides occurring at distance of 10 ~ 500 km from the seismic fault. The ratio of the vertical to horizontal components of the acceleration is <0.5, i.e., Va/Ha < 0.5, according to the strong ground motion records. Landslides are rarely triggered. None of staircase-shaped and long run-out rockslide has been reported for this category.

Coller and Elnashai (2001) studied the variation of the ratio of the vertical to horizontal components of the acceleration from different magnitudes of earthquakes, with their epicenter distance (Fig. 4). Figure 4 shows that generally, the greater the earthquake magnitude, the higher the ratio of the vertical to horizontal components of earthquake acceleration. The horizontal earthquake acceleration is greater than the vertical earthquake acceleration when the earthquake magnitude is <Ms7.5. However, when the Ms = 7.5, and the epicenter distance is <10 km, the vertical acceleration is greater than the horizontal acceleration.

Fig. 4

Comparison of ratio between the vertical and horizontal accelerations from the Wenchuan earthquake with the Coller’s previous curves (revised by authors from Coller and Elnashai 2001)

Examining the result from the Wenchuan earthquake (also see Fig. 3) reveals that the vertical acceleration is greater than the horizontal one, and that the ratio of the vertical component (Va) to the horizontal component (Ha) of the acceleration varies from 1.0 to 1.2. This explains why most of the major staircase-shaped and long run-out landslides occur within 10 km (more prevalent when within 5 km) from the seismic fault. It is worthy to note that in conventional assessment method for landslide or rock avalanche stability, only the horizontal earthquake force is considered and the vertical force is neglected. This paper shows that the vertical seismic forces in meizoseismal areas trigger landslides and should therefore be considered in the stability analysis of landslide and avalanche.

Pseudostatic analysis on factor of safety of landslide with different slope angles

A pseudostatic analysis reveals the effects of earthquake shaking with accelerations that create inertial forces. In limit equilibrium method of the Geo-Studio software, these forces act in the horizontal and vertical directions at the centroid of each slice. The horizontal inertial forces are applied as a horizontal force on each slice and the vertical inertial forces are added to the slice weight. Vertical accelerations can be positive or negative. A positive vertical acceleration indicates inertia forces acting downward in the direction of gravity while negative vertical acceleration indicates inertia forces acting upward against gravity.

To analyze the effect of vertical acceleration on the slope stability, simple landslide models with three different slope angles of 20°, 45° and 70° are considered (Fig. 5). The internal friction angle is setup to 20°, 30° and 30° for slope angles of 20°, 45° and 70°, respectively, that typifies the lower, middle and steep slope in this area. First, for the comparing reason, the initial factor of safety (FOS) of the landslide with three different slope angles is better to be 1.0 under the cohesion is setup to 0 and acceleration = 0 using Bishop’s method (Tables 1, 2 and 3). Then, the variation of cohesion is considered from 0 to 200 Pa. To analyze the effect on vertical acceleration to landslide, the average horizontal acceleration is setup to 0.3 g, and the variation of vertical acceleration is from −0.45 to 0.45, −0.3 to 0.3 and −0.1 to 0.1 g, respectively, within, near and outside seismic fault, according to the strong ground motion records from the Wenchuan earthquake (Yin et al. 2009). Three different situations based on the distance of landslide from the seismic fault are discussed below.

Fig. 5

Analysis models for landslide stability from three different slope angles (20°, 45° and 70°)

Table 1 FOS of landslide vs. cohesions from three slope angles under Va/Ha = 3/2

Table 2 FOS of landslide vs. cohesions from three slope angles under Va/Ha = 1

Table 3 FOS of landslide vs. cohesions from three slope angles under Va/Ha = 1/3

Landslide within seismic fault

These are landslides triggered within the seismic fault, i.e., landslides falling within 0–5 km from the seismic fault. In this category the vertical vibrations are found to be more dominant than the horizontal vibrations with vertical acceleration (Va) of 0.45 g and horizontal acceleration (Ha) of 0.3 g (Va/Ha > 1).

In Fig. 6, the red line indicates the relationship between factor of safety (FOS) of landslide and cohesion when there is only horizontal acceleration (EW = 0.3 g), without any vertical acceleration (UD = 0) (also see Table 1). Under this condition, the FOS of landslide is observed to increase (from 0.765 to 1.182) with increasing cohesion (from 50 to 200 kPa) for a slope angle of 45°. The trend of the relationship between FOS and cohesion for the various slope angles (20°, 45° and 70°) is found to be similar. The relationship for the slope angle of 45° is therefore selected for detail analysis and discussion. When the condition is such that there are both horizontal (EW = 0.3 g) and vertical accelerations (UD = 0.45), the FOS of landslide is observed to in crease from 0.832 to 1.150 as cohesion increases from 50 to 200 kPa (black line in Fig. 6) for the slope angle of 45°.

Fig. 6

FOS of landslide vs. cohesions from three different slope angles with/without vertical acceleration (Va/Ha > 1, the epicenter distance: <5 km) (also see Table 1)

It can also be observed in Fig. 6 that, when the cohesion is less than about 150 kPa, the FOS of landslide with both horizontal and vertical accelerations (black line) is higher than that without vertical acceleration (red line). However, the FOS with both horizontal and vertical accelerations is lower than that without vertical acceleration when the cohesion is >150 kPa. The relationship between FOS with horizontal and negative vertical acceleration (UD = −0.45 g) is indicated by the blue dashed line in Fig. 6. The FOS of landslide is observed to increase from 0.640 to 1.242 as cohesion increases from 50 to 200 kPa for the slope with angle of 45°.

When the cohesion is less than about 150 kPa, the FOS of landslide with both horizontal and negative accelerations (blue dashed line) is less than the one without vertical acceleration (red line) and vice versa when cohesion is >150 kPa,

Landslide near seismic fault

The landslide is considered to be triggered near seismic fault when the distance from the landslide to the seismic fault is from 5 to 10 km. In this category, both vertical and horizontal vibrations are prevalent and the same (Va = 0.3 g and Ha = 0.3 g) with Va/Ha being equal to 1.

As presented in Fig. 7 (also see Table 2), FOS of landslide without vertical acceleration (red line) increases from 0.765 to 1.182 as cohesion increases from 50 to 200 kPa for the angle of slope of 45°.

Fig. 7

FOS of landslides vs. cohesions from three slope angles with/without vertical acceleration (Va/Ha = 1, the epicenter distance: 5 ~ 10 km) (also see Table 2)

The FOS of landslide with both horizontal (EW = 0.3 g) and vertical accelerations (UD = 0.3 g) on the other hand increases from 0.814 to 1.159 as the cohesion increases from 50 to 200 kPa for the slope angle of 45° (black line in Fig. 7).

Also, when the vertical acceleration is negative, i.e., UD = −0.3 g, the FOS of landslide increases from 0.694 to 1.217 as the cohesion increases from 50 to 200 kPa for the slope angle of 45° (blue dashed line).

Landslide outside seismic fault

Landslides occurring outside seismic fault are those that are at a distance of 10–500 km from the seismic fault. The horizontal vibration (Ha = 0.3 g) in this case is more dominant than the vertical vibration (Va = 0.1 g) and therefore has an acceleration ratio (Va/Ha) of <1. Figure 8 presents the relationship of FOS and cohesion for this category of landslides. When there is no vertical acceleration (EW = 0.3 g and UD = 0; Table 3), FOS of landslide increases from 0.765 to 1.182 as the cohesion increases from 50 to 200 kPa for a slope angle of 45° (red line). However, the vertical and horizontal accelerations are the same (EW = 0.3 and UD = 0.3 g), the FOS increases from 0.783 to 1.173 as cohesion increases from 50 to 200 kPa (black line). Finally, when the vertical acceleration is negative (UD = −0.1 g), the FOS of landslide increases from 0.747 to 1.192 as cohesion increases from 50 to 200 kPa for the slope angle of 45° (blue dashed line).

Fig. 8

FOS of landslide vs. cohesions from three slope angles with/without vertical acceleration (Va/Ha < 1, the epicenter distance: 10 ~ 100 km) (also see Table 3)

It is apparent from the above discussion that the epicenter distance (distance of seismic fault from landslide) is an important factor which influences the vertical acceleration and thus the stability of landslide. Within the seismic fault, the amplitude effect of vertical acceleration is very dominant with the FOS of landslide for vertical acceleration ranging from positive to negative having a variation of 25 %. The variation of FOS of landslide for vertical acceleration raging from positive to negative are 15 and 5 % for landslides near seismic fault and outside seismic fault, respectively.

The influence of vertical seismic coefficients on the FOS of landslide is very significant when the earthquake magnitude is quite high. It must be noted that the effect of vertical inertial forces, which alters the slice weight, on the stability of landslides is significant only when frictional strength components without cohesion are used for estimating shear strength. When the vertical acceleration is low, it alters the normal force of the slice which increases the base shear resistance. The added mobilized shear arising from the added weight is therefore offset by the increase in shear strength. However, if the cohesive strength component is considered, the added mobilized shear arising from the added weight cannot be offset by the increase in shear strength.

Simulation on the dynamic response of landslide with different slope angles

Above pseudostatic analysis is more precise than dynamic analysis to provide a value under certain parameters and situations. But, time-depended vibration processes should be discussed, the dynamic analysis is necessary to consider the effect of variation between positive and negative vertical accelerations and the vibration duration since the pseudostatic analysis could not account for these effects. The dynamic simulation is conducted on the variation of the FOS of landslides for the three slope angles (20°, 45° and 70°), respectively, that typifies the lower, middle and steep slope in this area, from the Wenchuan earthquake (Ms = 8.0). The acceleration records at the Qingping Station within the seismic fault of the Longmenshan are used as input (Fig. 9) (Yu et al. 2009). The duration of vibration is about 70 s. The peak accelerations are −0.824, 0.803, and 0.623 g in the EW, SN and UD (vertical) directions, respectively. The compound peak acceleration from the three directions is 1.3 g.

Fig. 9

Acceleration records with ∆t = 0.001 s at the Qingping station, Mianzhu, Sichuan (from National Strong Ground Motion Center 2008)

In the stability analysis with QUAKE/W from Geo-Studio (2007 Version), dividing the total mobilized dynamic shear by the potential sliding mass provides an average acceleration value. The average acceleration obtained for each time integration step during the shaking is representative of the resultant of both horizontal and vertical applied accelerations (Fig. 10).

Fig. 10

Average acceleration vs. times during shaking from three slope angles with/without vertical acceleration (∆t = 0.005 s) (also see Fig. 5)

Figure 11 illustrates the variation of the FOS of landslide with time during vibration. The simulation results from the three slope angles for cohesions of 0, 50 and 100 kPa show that the FOS of landslide is quite different with/without vertical acceleration. For the landslides with 20° and 45° slope angles, the FOS of landslide with vertical acceleration is significantly increased by amplification, in contrast to those without vertical acceleration. However, for the landslide with 70° slope angle, the amplification effect is not significant.

Fig. 11

FOS vs. times during shaking with/without vertical acceleration from three slope angles (∆t = 0.005 s) (also Fig. 5)

Comparing these results to that from the pseudostatic method, the dynamic analysis indicates that the forces from the Wenchuan Earthquake cause the stresses in the landslide to oscillate. Along a potential slip surface, the mobilized sliding force decreases and increases in response to the inertial forces. There may be moments during the shaking that the mobilized sliding force exceeds the available sliding resistance, which causes a temporary loss of stability. During these moments when the FOS of landslide is less than unity, the landslide may experience some displacement. An accumulation of these movement spurts may manifest itself as failure.

A typical case study from the Chengxi landslide, Beichuan town

The Chengxi landslide, at the western of ruined Beichuan Town, destroyed hundreds of houses, including over 50 six-story buildings and caused 1,600 deaths. This was the most catastrophic landslide triggered by Wenchuan earthquake (Fig. 12).

Fig. 12

The Chenier catastrophic landslide triggered by the Wenchuan earthquake and caused 1,600 deaths, at ruined Beichuan town

The Chengxi landslide material comprised sandstone, shale and schist in the lower section of the slide, and the accumulation of old landslide material on the flat section (Fig. 13). The accumulated material of the Chengxi landslide is about 400 m long, 400 m wide and about 30 m thick, resulting in a total landslide volume of about 480 × 10⁴ m³. The Chengxi landslide is only 300 m away from the main central seismic fault of the Wenchuan earthquake. The source of the landslide (earth material from the upper portion of the slope with elevation >800 m) having a volume of about 2 million m³ was thrown upwards, impacted and further eroded the front part of the rock mass with a volume of about 2.8 million m³. The convex-shaped sliding plane with a slope of about 45° reflected the impact characteristics due to both horizontal and vertical forces.

Fig. 13

Section map of the Chengxi landslide, at ruined Beichuan town, Sichuan

To analyze the vertical acceleration effect on the Chengxi landslide, the pseudostatic method from the strength reduction of FLAC^3D is applied. And the strength reduction technique from FLAC3D is powerful to search the unclear sliding belt (plane) of the Chengxi landslide before failure automatically. Generally, the slope failure is identified as having the failure shear strain developed from the toe to the top of the slope by reducing the value of c and Φ gradually until the slope failure occurs, and the corresponding reduction factor is defined as the factor of safety (Matsui and San 1992; Dawson et al. 1999). The user’s guide of FLAC^3D software specified that the “strength reduction technique” is typically applied in factor of safety calculations by progressively reducing the shear strength of the material to bring the slope to a state of limiting equilibrium (Itasca Consulting Group Inc. 2005). The FOS of landslide is defined according to the equations:

Reduction friction angle φ _γ :

$$ \varphi_{\gamma } = \arctan (\tan \varphi /f_{\text{s}} ) $$

(a)

Reduction cohesion c _γ :

$$ c_{\gamma } = c/f_{\text{s}} $$

(b)

where, f _s is the factor of safety.

Combining with on-site survey and backward analysis, the basic physical and mechanical parameters of the Chengxi landslide estimated are as follows:

• the landslide material: internal friction angle φ = 30.0°; cohesion c = 100 kPa; density = 20.0 kN/m³; shear modulus S = 50 MPa.

• bedrock: internal friction angle φ = 35.0°; cohesion c = 500 kPa; density = 24.0 kN/m³; shear modulus S = 100 MPa; bulk modulus B = 200 MPa.

Numerical simulations, using finite element method, discrete element code and finite difference method, have become powerful tools for simulating the slope failure (Griffith and Lane, 1999; Qi et al. 2003). Matsui and San (1992) defined the slope failure as the failure shear strain developed from the toe to the top of the slope by a shear strength reduction technique, with the shear strain increment very similar to the failure slip surface using the Bishop’s method. Recently, Sun et al. (2012) conducted research work on the effect of vertical earthquake motion on slope stability using UDEC simulation for two case scenarios: (1) only the horizontal earthquake forces, and (2) both horizontal and vertical earthquake forces. The simulated result shows distinct contour map of the plastic zone in the two cases. It can be observed that when the vertical seismic force is considered, more tensile failures occur at the head scarp, and the vertical peak acceleration leads to distinct reduction of the slope stability factor. Li et al. (2012) also simulated a typical landslide at the Donghekou, Qingchuan County, due to the Wenchuan earthquake, to present the kinematic behavior and a rapid long run-out characteristic of this landslide using a 2D discrete element model.

Since the landslide is within the major seismic fault (near NNE), and the fault movement is thrusting, EW and UD direction accelerations are only considered. In this paper, the shear strain increment (SSI) is estimated and used in analyzing the slope failure and slide surface characteristics. Figure 14a shows the deformation characteristics of the Chengxi landslide in acceleration conditions during shaking from the Wenchuan Earthquake. The shear strain increment (SSI) is observed to be very large along the shear exit zone in the front part of the landslide with values ranging from 140 to 160. The SSI values however ranged from 40 to 120, in the middle and back part of the landslide and the Quaternary deposits.

Fig. 14

FLAC^3D simulated contour cloud maps of the Chengxi landslide triggered by the Wenchuan earthquake (time steps = 50,000)

Figure 14b shows the displacement characteristics of the slide. It is shown in the figure that, the amount of material displaced from the landslide surface is significantly greater than that from the deep section of the slide. The amount of displacement increased from 6.0 to 8.0 m in the section of the slide, to 12.0–14.0 m at slide surface. These results show that the mode of the landslide failure during the Wenchuan earthquake was a throwing failure.

Using the shear strength reduction technique, the FOS of landslide is calculated before and during the Wenchuan earthquake (Table 4). Before the Wenchuan Earthquake, the FOS is estimated to be about 1.1 indicating that the landslide is stable. During the Wenchuan Earthquake, the absolute value of average acceleration varies from 0.35 to 0.4 g (Fig. 15). When only the horizontal acceleration (Ha = 0.3 g) is considered, the FOS is computed to be 0.68. If both the horizontal component (Ha = 0.3 g) and vertical component (Va = −0.45 g) of the acceleration are considered as inputs, i.e., Va/Ha = 1.5, the FOS computed is 0.61. The negative value of the vertical acceleration measured indicates throwing effect of the landslide material.

Table 4 FOS of the Chengxi landslide before/during Wenchuan earthquake with various vertical accelerations

Fig. 15

Variation in average acceleration of Chengxi landslide vs. times during shaking from the Wenchuan earthquake (∆t = 0.005 s)

Figure 16 shows the variation in FOS of the Chengxi landslide during shaking from the Wenchuan Earthquake with time. The FOS varied from 0.7 to 2.3 when both horizontal and vertical accelerations (EW + UD) are considered while FOS varied slightly from 0.8 to 1.4 when only the horizontal component of acceleration is considered.

Fig. 16

Variation in FOS of the Chengxi landslide vs. times during shaking from the Wenchuan earthquake (∆t = 0.005 s)

The results computed from both the pseudostatic method and dynamic analysis reveal that the FOS during the earthquake varied significantly whether vertical acceleration is considered or not. This explains why lots of throwing-pattern catastrophic landslides occurred within 10 km of the seismic fault in the Wenchuan earthquake.

Conclusion

The pseudostatic analysis and the dynamic response on landslide stability due to the Wenchuan earthquake are studied in this paper with the Chengxi (West Town) catastrophic landslide used as a case study. The results show that the epicenter distance is an important factor which influences the vertical acceleration and thus the stability of landslide. Also the vertical acceleration was found to have a significant impact on the FOS of landslide if the earthquake magnitude is quite large. The slope angle of the sliding surface is also observed to affect the amplification of vertical acceleration and thus the FOS of landslide. Finally, the results explain the reason why lots of throwing-pattern catastrophic landslides occurred due to vertical force processes within 10 km of the seismic fault in the Wenchuan earthquake (Yin et al. 2009). That is much different with conventional toppling failure due to gravity.