# Investigation of the initiation mechanism of an earthquake- induced landslide during rainfall: a case study of the Tandikat landslide, West Sumatra, Indonesia

- Fikri Faris
^{1}Email author and - Wang Fawu
^{1}

**1**:4

**DOI: **10.1186/s40677-014-0004-3

© Faris and Fawu; licensee Springer 2014

**Received: **21 August 2014

**Accepted: **27 August 2014

**Published: **3 October 2014

## Abstract

### Background

A large earthquake struck Padang Province, West Sumatra, Indonesia, at 17:16 on September 30, 2009. The earthquake had a moment magnitude of Mw 7.6, and triggered landslides in Tandikat, Padang Pariaman Regency. The landslides occurred during rainfall, and originated on mountains mantled with loose pumice, taking many lives. The unfortunate combination of intensive rainfall and strong earthquake probably decreased slope stability. This study seeks to examine the initiation mechanism of earthquake- induced landslides during rainfall, and to develop a new approach to predict pore pressure increase by assuming reciprocal relationships between strain, stiffness, and pore pressure.

### Results

In order to assess slope stability, the concept of stiffness degradation was used to predict pore pressure increase due to earthquake. This was achieved by developing empirical formulation based on cyclic triaxial test results. A new procedure based on the “rigid block on quasi plastic layer” assumption was developed to assess slope stability of earthquake-induced landslides during heavy rainfall. Results from cyclic triaxial test experiments showed that effective confining pressure and initial shear stress had considerable influence on increase in pore pressure. Slope stability analysis using actual earthquake acceleration suggest that landslide occurred due to pore pressure build up and the factor of safety decreased rapidly before earthquake acceleration reached its peak.

### Conclusions

The results emphasize the high risk of catastrophic earthquake-triggered landslides in tropical regions with high rainfall. It also suggest that landslide with similar mechanism of pore pressure increase are likely to occur on saturated sliding zones during smaller earthquakes.

### Keywords

Earthquake induced landslide Rainfall Pore pressure increase Pumice sand Cyclic triaxial test Indonesia## 1Background

_{w}7.6 (M7.6 2009.9.30), and caused over 1,000 deaths (EERI [2009]). The M7.6 2009.9.30 Padang earthquake triggered many landslides, and these accounted for more than 60% of the total death toll. The most extensive landslides occurred in Tandikat, Padang Pariaman Regency. These buried hundreds of people, and flattened some villages (Figure 1). A loose pumice ash layer on the mountains is thought to have been saturated by extensive and heavy rainfall before the earthquake triggered the landslides. In this particular area, the probability of concurrent earthquake and rainfall events is high, since it has a tropical rainforest climate, and is also situated on a seismically active plate margin. Consequently, it is essential to study the initiation mechanisms of landslides of saturated pumice sand, while considering the effects of the unfortunate combination of independent events such as heavy rainfalls and earthquakes.

Studies of initiation and post-failure mechanisms are conducted using different laboratory test methods. Studies of initiation mechanisms commonly use laboratory shearing tests considering “limited displacement” conditions (i.e. triaxial, hollow cylinder torsional shear, direct shear, and simple shear tests). Studies of post-failure mechanisms typically consider “large displacement” tests using ring shear tests.

The post-failure mechanism of volcanic deposits has been examined in several studies. A study of a long run-out landslide in pyroclastic strata by Wang and Sassa ([2000]) used undrained ring shear apparatus to confirm grain crushing mechanism during shearing. Wang et al. ([2010]) also evaluated the post-failure mechanism of long run-out pumice material from the Tandikat landslide, using the same apparatus.

Several studies have examined the initiation mechanism of landslides in common volcanic soils, and the role of pore pressure build-up. Hyodd et al. ([1998]) examined the liquefaction characteristics of several crushable volcanic deposits, and Suzuki and Yamamoto ([2004]) emphasized the liquefaction characteristics of the Shirashu pyroclastic deposit in Japan, using cyclic triaxial tests on disturbed and undisturbed samples. However, specific research on the dynamic properties of pumice sand and its relation to pore pressure generation is rare.

Researches based on pore pressure models mainly use regular sand for laboratory tests, with very limited effort focused on the dynamic behaviour of volcanic sand, especially pumice sand. Seed et al. ([1976]) and Lee and Albaisa ([1974]) used clean sand to study liquefaction, and to develop a pore pressure model based on the number of cyclic loads. Yamazaki et al. ([1985]), Sugano and Yanagisawa ([1992]) and Jafarian et al. ([2012]) used Toyoura silica sand to derived pore pressure models based on the strain energy concept, using a variety of laboratory tests. Work on the behaviour of pumice during dynamic load has been reported by Marks et al. ([1998]) and Orense and Pender ([2013]). These authors studied the liquefaction characteristics and resistance of crushable pumice soils from North Island, New Zealand, based on undrained cyclic triaxial tests and field test data. They confirmed that pore pressure built up during shearing. Nevertheless, an empirical model for pore pressure generation in such material has not yet been developed.

This study examines the initial mechanism of earthquake-induced landslides during rainfall, by development of a pore pressure model using local pumice sand, and the use of a cyclic triaxial apparatus. The pore pressure model proposed is based on assumption of a reciprocal relationship between strain, stiffness, and pore pressure. This model was then incorporated with groundwater simulation and slope stability analysis to encompass the problem of earthquake-induced landslides during rainfall. To fulfil this purpose, field investigations and laboratory tests using a stress-controlled cyclic triaxial apparatus were conducted to examine the physical behaviour of pumice sand from the Tandikat landslide district.

## 2Study site

### 2.1 Geological setting

The landslides are located in a mountainous area around two volcanoes (Mt. Tandikat and Mt. Singgalang), and are extensively distributed on steep slopes inclined at ~30 to 50 degrees. The slopes are mainly mantled by unconsolidated volcanic deposits that were derived from the nearby mountains. This topographical condition is considered to be an important contributory factor for landslides in Tandikat.

### 2.2 Seismicity and meteorology

Several high magnitude earthquakes have been recorded in the subduction zone along the west coast of Sumatra in the last few decades. The M_{w} 9.0 Aceh earthquake of December 26, 2004 caused the catastrophic Indian Ocean tsunami. The Nias earthquake of March 28, 2005, and the South Sumatra earthquake of September 12, 2007, had magnitudes of M_{w} 8.7 and M_{w} 8.4, respectively. The most recent large earthquake was the September 30, 2009 Padang earthquake, which had a moment magnitude of 7.6. The epicenter was located offshore, WNW of Padang City, and the hypocenter was located at a depth of 80 km, within the oceanic slab of the Indo-Australian plate. This earthquake has been interpreted as an indication of a higher possibility of an imminent mega-earthquake in this region (Aydan [2009]).

## 3Methods

### 3.1 Field investigation

*k*

_{s}) and Green-Ampts suction at the wetting front, (

*ψ*) (Muñoz-Carpena et al. [2002]). This method was developed by Philip ([1993]) to estimate saturated hydraulic conductivity using Green-Ampt analysis approximation. The method reasonably satisfies the Green-Ampts model, which is used later in this study in assessing soil saturation due to rainfall infiltration. Falling head permeameter tests were conducted in the field using an open-ended pipe 0.085 m in diameter and 0.5 m in length. The ground surface around the sampling point was first cleared, and the end of the pipe was then penetrated 0.2 m into the soil. A volume of water was then poured into the permeameter pipe, and the change of water level was recorded at each time interval. In the original procedure of Phillip-Dunne permeameter, it is necessary to measure the infiltration times when permeameter is half full (

*t*

_{med}) and empty (

*t*

_{max}) (Regalado et al. [2005]). In this study, the parameters

*t*

_{med}and

*t*

_{max}were estimated by linear regression of time vs. water level. The test was repeated several times at each site. The following equation of Regalado et al. ([2005]) was used to estimate saturated hydraulic conductivity (

*k*

_{s}) and the Green-Ampts suction at the wetting front (

*ψ*):

where *r*_{i} is the internal radius of the pipe.

**Input parameters used in the Green-Ampt infiltration model**

Infiltration model parameters | Values |
---|---|

Saturated hydraulic conductivity, | 300 |

Suction head, | 240 |

Initial volumetric water content, Δ | 0.62 |

Volumetric water content at saturation, Δ | 0.71 |

Slope angle, | 30 |

### 3.2 Infiltration model

where *f(t)* = potential infiltration rate at time t, *F(t)* = cumulative infiltration at time t, *ψ* = suction head at the wetting front, Δ*w* = volumetric water content deficit, and *θ* = slope angle.

*z*

_{w}) had already surpassed the three-metre depth where the impermeable sandy clay is located. This suggests that water percolated into the pumice sand during rainfall, and collided with the impermeable sandy clay layer. This generated a temporary perched groundwater table, which consequently created fully saturated conditions in the lower part of the pumice sand.

### 3.3 Groundwater modelling

Many researches have been conducted to develop numerical model to predict groundwater in unconfined aquifer. Among many methods, Boussinesq equation is most often used to estimate groundwater (Bansal and Das [2010]). The performance of this method is reliable to predict experimental soil flume test (Steenhuis et al. [1999]; Sloan and Moore [1984]). This method generally formulated as a parabolic nonlinear equation, thus linearization process is used to derive analytical solution. Alternatively, finite difference numerical model can be used to utilize the aforesaid equation (Bansal [2013]).

*h*is the height of phreatic surface measured above the impermeable sloping bed in the vertical direction.

*k*

_{s}and

*S*respectively are the hydraulic conductivity and specific yield of the aquifer.

*R*is the net rate of recharge of infiltrated rainfall and

*θ*is the bed slope.

*C*

_{1}= (

*k*

_{s}cos

^{2}

*θ*)/

*S*and

*C*

_{2}= (

*k*

_{s}sin 2

*θ*)/2

*S*. Mac Cormack scheme is an explicit finite difference with predictor-corrector step. The predictor step is applied by replacing the spatial and temporal derivatives by forwards difference to obtain predicted value of

*h*, indicated as

*h*

^{ • }in Eq. 7.

*n*and

*t*are respectively spatial and time identifier, Δ

*x*and Δ

*t*are horizontal spatial interval and time interval, respectively. The corrector step is then obtained by replacing the space derivative by rearward difference, while the time derivative is preserved using forward difference approximation. Then Eq. 8 can be obtained.

*h*

_{n,t + 1}is simply obtained from arithmetic mean of ${h}_{n,t+1}^{\u2022}$ and ${h}_{n,t+1}^{\u2022\u2022}$ from Eqs. 7 and 8, respectively, and the Eq. 9 is obtained.

### 3.4 Laboratory tests

*C*

_{u}) greater than 6 (

*C*

_{u}= 8.19), and a coefficient of curvature (

*C*

_{c}) between 1 and 3 (

*C*

_{c}= 1.06).

**Pumice sand properties obtained from in-situ measurement and laboratory tests**

Pumice sand properties | Values |
---|---|

Specific gravity | 2.664 |

Bulk density (g/cm | 1.503 |

Dry density (g/cm | 0.888 |

Relative density (%) | ≈50 |

Void ratio, | 2.00 |

Uniformity coefficient ( | 8.19 |

Coefficient of curvature ( | 1.03 |

Saturated hydraulic conductivity, | 4.42x10 |

Suction head, | 240 |

Volumetric water content deficit, Δ | 0.09 |

Internal friction angle, | 39.0 |

_{2}gas to replace the air inside it before it was saturated with de-aired water. For undrained triaxial tests, Skempton ([1954]) introduced the pore pressure coefficient

*B*to express the pore water pressure change Δ

*u*that occurs due to a change in confining pressure Δ

*σ*

_{3}. The Skempton’s

*B*value can be obtained by Eq. 13.

Full saturation of the specimens was confirmed if Skempton’s *B* value was greater than 0.95. After full saturation was reached, the specimens were loaded axially into a triaxial cell with differing confining pressures of 20 kPa and 50 kPa, with displacement velocity of 0.7 mm/minute. The undrained strength parameters obtained are summarized in Table 2.

**Summary of CTX tests conducted during this study**

No | Test ID | B values |
| K |
---|---|---|---|---|

1 | TND45-0 | 0.980 | 45 | 0 |

2 | TND45-0.2 | 0.970 | 45 | 0.2 |

3 | TND45-0.3 | 0.980 | 45 | 0.3 |

4 | TND45-0.4 | 0.980 | 45 | 0.4 |

5 | TND90-0 | 0.960 | 90 | 0 |

6 | TND135-0 | 0.980 | 135 | 0 |

7 | TND60-0 | 0.980 | 60 | 0 |

8 | TND75-0 | 0.990 | 75 | 0 |

9 | TND100-0 | 0.980 | 100 | 0 |

10 | TND120-0 | 0.980 | 120 | 0 |

The test procedure was performed in a manner to simulate the stress condition in the field. At first, the specimens were consolidated with specified confining pressure after full saturated condition has been attained. As an approximation of the irregular motion of earthquake loading, the amplitude of cyclic sinusoidal deviatoric axial stress was taken to be 65% of the maximum magnitude of shear stress induced by the actual earthquake, as proposed by Seed et al. ([1975]). The cyclic sinusoidal axial stresses were then applied to the specimens at a rate of 1 Hz until the ultimate failure state was achieved.

### 3.5 Pore pressure model

*E*

_{0}) at the origin; this slope is known as the maximum Young’s modulus. The secant Young’s modulus (

*E*

_{s}) is represented as the slope of the line connecting the original point with the tip of the loop associated with the axial strain amplitude (

*ε*

_{s}). The secant Young’s modulus (

*E*

_{s}) and axial strain amplitude (

*ε*

_{s}) are the key properties of the developed non-linear constitutive model. These parameters were obtained at every cycle loading, to be correlated with the pore pressure increase.

*G*

_{ s }and the shear strain amplitude,

*γ*

_{s}are defined by the equation:

where *μ* is the Poisson’s ratio equal to 0.5 for undrained conditions, and *E*_{s}, *G*_{s} and *γ*_{s} are the secant Young’s modulus, the secant shear modulus, and the shear strain amplitude, respectively. The initial shear modulus, *G*_{0}, can be referred to Eq. 14 by changing parameter *E*_{s} to *E*_{0}.

## 4Results and discussion

### 4.1 Result of static triaxial test

Static triaxial tests were conducted under 20 kPa and 50 kPa effective confining pressure to attain the shear strength of the pumice sand, and to understand its basic physical behaviour under stresses. The stress path (Figure 11) shows a large sector of contractive curve, implying the effect of pore pressure increase. It also indicates a low elastic threshold (ET), suggesting that the soil structure contracts easily under low shear stress. As the effective confining pressure decreases, the stress path exhibits dilation behaviour as it approaches the phase transformation line (PTL). This tendency of dilation after passing PTL suggests cyclic mobility behaviour when cyclic loading is applied (Ishihara [1985]). From the test, the internal friction angle (*φ*′) is equal to 39.0°.

### 4.2 Typical result of cyclic triaxial test

*r*

_{u}was used as the cyclic triaxial (CTX) test is conducted, by applying cyclic loads in the axial direction only (Žlender and Lenart [2005]). Equation 16 of corrected pore pressure ratios

*r*

_{u}is used incorporating the

*B*value.

where Δ*u* is the excess pore pressure, *σ*_{d} is the deviatoric axial stress, and ${\sigma}_{0}^{\prime}$ is the effective confining pressure.

*r*

_{u}= 0.8, which is the inflation point of the

*γ*

_{t}

*-r*

_{u}curve. Cumulative shear strain corresponding to

*r*

_{u}= 0.8 was considered to be the reference cumulative shear strain,

*γ*

_{r}, where, beyond this point, the pumice sand specimen underwent large strain without further significant pore pressure increase (Figure 17).

### 4.3 Effect of effective confining pressure on reference cumulative shear strain

*γ*

_{t}

*-r*

_{u}curves, at each given effective confining pressure, pore pressure ratio increased rapidly at low cumulative shear strain (Figure 18). However, each effective confining pressure responded differently to pore pressure ratio increase during cyclic loading. Pore pressure ratio of the specimen with low confining pressure increased rapidly in response to cyclic loading, whereas pore pressure ratio at high confining pressure had a slower response to cyclic loading. At a given cumulative shear strain, pore pressure ratio clearly increases with decreasing effective confining pressure (Figure 18).

*γ*

_{r}where

*γ*

_{r}increased with effective confining pressure, indicating that more cumulative strain was necessary to increase pore pressure at some level, as ${\sigma}_{0}^{\prime}$ increased. This suggests, therefore, that shallow saturated pumice sand deposits have higher risk of pore pressure increase during earthquakes than do deeper saturated sands.

### 4.4 Effect of initial shear stress on reference cumulative shear strain

*D*

_{r}≈ 50%). The ratio of initial shear stress to the initial effective confining pressure,

*K*, is defined as:

*γ*

_{t}

*-r*

_{u}curves of different initial shear stresses are presented in Figure 20. It shows the response in pore pressure increase of effective confining pressure 45 kPa, where the increase of initial shear stress reduced the response of pore pressure generation.

*K*(Figure 21). This suggests that, in a shallow deposit, soil mass with larger initial shear stress needs larger cumulative shear strain to increase pore pressure ratio to some certain value. From the viewpoint of slope stability analysis, this suggests that, particularly in the studied area where shallow deposit be present, gentle slopes are more prone to pore pressure build-up during earthquakes than are steep slopes.

### 4.5 Effect of effective confining pressure on stiffness degradation

*G*

_{s}and cumulative shear strain,

*γ*

_{t}with different effective confining pressures (Figure 22) clearly shows that the effective confining pressure contributed to the initial shear modulus,

*G*

_{0}, which occurred at low strain. During cyclic loadings, the shear modulus decreased rapidly irrespective of the effective confining pressure. Initial shear moduli,

*G*

_{0}, were estimated during curve fitting of the secant shear modulus,

*G*

_{s}versus cumulative shear strain, by using the

*G*

_{s}

*-γ*

_{t}curve fit equation:

*β*is a non-dimensional calibration parameter which involves effective confining pressure in the regression equation (R

^{2}= 0.966) in Eq. 20.

### 4.6 Effect of initial shear stress on stiffness degradation

The plot secant shear modulus, *G*_{s} vs cumulative shear strain, *γ*_{t} corresponding to initial shear stress ratio, *K* of 45 kPa effective confining pressure are shown in Figure 23. The effect of initial shear stress, *K*, on shear stiffness degradation is not clearly observable. The plot shows the same tendency, and insignificant differences of initial shear modulus, *G*_{0,} at every different values of *K*. Therefore, the influence of initial shear stress on stiffness degradation is negligible.

## 5Numerical analysis

### 5.1 Pore pressure model fitting

*γ*

_{t}

*-r*

_{u}relationship. The following equation was fitted to the data:

*α*is a non-dimensional calibration parameter of the pore pressure model, and

*γ*

_{t}is the cumulative shear strain. From a practical viewpoint,

*γ*

_{t}can be estimated over numerical analysis by considering a constitutive model that can simulate the non-linear stress–strain response of the soil undergoing rapid pore pressure increase, which will be discussed later in a subsequent section. Using linear regression, the calibration parameter

*α*was correlated with effective confining pressure, ${\sigma}_{0}^{\prime}$ and the initial shear stress ratio,

*K*. The following linear regression of

*α*involves parameter ${\sigma}_{0}^{\prime}$ and

*K*in the Eq. 22 (R

^{2}= 0.951).

In this study, the non-linearity of stress–strain were also implemented by correlating cumulative shear strain, *γ*_{t} and the secant shear modulus, *G*_{s}. The influence of effective confining pressure and initial shear stress on shear stiffness degradation was also examined through data obtained from the CTX tests.

The best fit curve showing the relationship of *G*_{s}*-γ*_{t} is expressed by Eq. 19. The equation implies that shear strain (γ < 10^{−5}) near zero will transform the equation to *G ≈ G*_{0}*,* and the increased cumulative shear strain will gradually decrease the secant modulus.

### 5.2 Rigid block on quasi-plastic layer and simulation procedure

*G*

_{s}involves in stress–strain interaction, whereas the unsaturated portion of the pumice sand layer was assumed to act as a rigid body. However, during earthquake loadings, increasing shear stress develops large shear strain, which reduces the secant shear modulus,

*G*

_{s}. Therefore, the next cycle of loadings may generate larger shear strain due to the previously reduced shear modulus. This process results in irrecoverable shear strain in the cyclic loading time history, which is the reason of the use of the term “quasi plastic”. Slope stability analysis was also incorporated with the model, in the manner of infinite slope assumption. To explain this process, Eq. 19 can be also be recast as follows:

*n*represents the number of calculation steps. At the first calculation step (

*n*= 1), the initial value of

*G*

_{s}is set as equal to G

_{0}. The strain is calculated based on the Eq. 24, which satisfies the elastic assumption:

*τ*

_{d}is shear stress taken from dynamic loadings at step

*n*. In the case of the earthquake, forces are calculated over a unit surface, and therefore shear stress

*τ*

_{d}is simply determined as:

*m*is the mass and

*a*

_{e}is the earthquake acceleration working parallel to the slope.

*t*

_{ n }is the time of

*n*

^{ th }step,

*u*

_{t}is total pore water pressure,

*u*

_{i}is initial pore water pressure defined in Eq. 28 just before earthquake happen.

where *γ* w is the unit weight of water (9.81 kN/m3), *h* is the height of phreatic line from the sliding surface obtained by groundwater simulation (Eq. 9) and *θ* is the slope angle.

*F*

_{s}) was calculated using the equation:

where *c*′ is effective cohesion, *φ*′ is the effective friction angle, *γ* is the unit weight of the sliding mass, *l* is the depth of the sliding mass and *k*_{h} is the coefficient of horizontal earthquake acceleration.

^{2}= 0.939; Figure 27), but some inaccuracies occur. These inaccuracies were probably due to error accumulation during the regression and calibration process.

### 5.3 Pore pressure simulation and slope stability analysis during actual earthquake

_{w}7.6 Padang earthquake. The results of pore pressure simulation and slope stability analysis are presented in Figure 29, as time history of earthquake acceleration, pore pressure, and factor of safety (

*FoS*). In each section, initial pore pressure generated from groundwater model was included. In the slope toe section, the pore pressure ratio presented by

*r*

_{u}curve increased rapidly in the first 17 s to unity, indicating the occurrence of liquefaction (Figure 29a). However, the factor of safety fell in accordance with the increase of pore pressure, reaching the failure state in 13 s, suggesting immediate failure due to pore pressure increase. This apparently shows that failure of the slope toe occurred before the maximum acceleration was reached, and that failure was strongly induced by pore water pressure increase. In contrast, failure of middle part and the crest slope occurred due to the combination of acceleration amplification and pore pressure build-up (Figure 29b and c, respectively).

This analysis shows that the slope would fail due to earthquake shaking, even without pore pressure increase. However, because the Tandikat landslide occurred during rainfall and underwent flow mobility, “dry” failure did not occur. Hence, “wet” failure, where the sliding zone reached a fully saturated condition is more realistic. The immediate and rapid failure of the slope before earthquake acceleration reached its peak shows that the saturated slope mass of loose pumice sand needs only slight energy to generate shear strain, which then increased the pore pressure to a critical and catastrophic level. This suggests that earthquake of smaller magnitude than the M7.6 2009.9.30 Padang earthquake can still lead to disaster if the required condition of sliding zone saturation due to rainfall is attained.

## 6Conclusions

Based on field observations, the pumice sand was the main material of the extensive landslide mass. The low density and high porosity of the pumice sand contributed to the slope failure induced by earthquake during rainfall.

The possibility of rainfall saturation of the pumice sand deposit was assessed using the Green-Ampt method. The results suggested that the sliding zone of the pumice sand deposits less than 3 m had a high probability of saturation by rainfall infiltration. High permeability and high water content due to antecedent rainfall facilitated rainfall water percolation into the ground.

Vulnerability of saturated pumice sand to pore pressure increase was confirmed by static and stress-controlled cyclic triaxial tests, which showed contractive behaviour of the pumice deposits, as indicated by excess pore pressure rise at small strains. Immediate pore pressure build-up occurred when fully saturated specimens were tested.

CTX test results showed that effective confining pressure greatly influenced reference cumulative shear strain. The test results showed that reference cumulative shear strain increased linearly with effective confining pressure, suggesting that risk of pore pressure increase during earthquake was greater in saturated shallow pumice sand deposits than in than thicker deposits.

The effect of initial shear stress on reference cumulative shear strain was also examined. The results indicate that soil mass with larger initial shear stress needs larger cumulative shear strain to increase pore pressure ratio to a certain value. This suggests that pore pressure increase during earthquake is more probable on gentle slopes than it is on steep slopes.

The phenomenon of stiffness degradation of pumice sand during cyclic loading was also considered. The effect of effective confining pressure on stiffness degradation was obvious, while the effect of initial shear stress was unclear. These results indicated that the effective confining pressure contributed to the initial shear modulus, *G*_{0}, which is the initial value of the secant shear modulus *G*_{s}. During cyclic loading, the shear modulus decreased rapidly irrespective of the effective confining pressure.

Stability analysis of the Tandikat landslide using a rigid block on a quasi-plastic layer assumption and the actual earthquake acceleration suggested that slope failure occurred due to pore pressure build-up. The factor of safety decreased rapidly before earthquake acceleration peaked. At that time, the energy of the earthquake had not reached its maximum level, suggesting that failure would probably occur on saturated sliding zone even during smaller earthquakes. This finding emphasises the high risk of catastrophic earthquake-triggered landslides in tropical regions with high rainfall.

## Declarations

### Acknowledgements

We appreciate the help of Prof. Dwikorita Karnawati and Dr. T. Faisal Fathani of Gadjah Mada University for providing additional data (SPT and geological logging) used in this study. We are also grateful to Mr. Rahindro Pandhu Mahesworo, S.T, M.T, the Head of Engineering Seismology Data, BMKG Indonesia, for providing the earthquake accelerogram data used in this study. We also express thank to Prof. Barry Roser of Shimane University and Austin O. Chukwueloka for their review of an early draft of the manuscript.

## Authors’ Affiliations

## References

- Aydan Ö:
*A reconnaissance report on the Pariaman-Padang earthquake of September 30, 2009*. Japan Society of Civil Engineers, Japan; 2009.Google Scholar - Bansal RK (2013) Modelling of groundwater flow over sloping beds in response to constant recharge and stream of varying water level. (in press). [last accessed: June 29, 2014], [http://www.ijm2c.ir/index.php/ijm2c/article/view/108/150]Google Scholar
- Bansal RK, Das SK:
**An analytical study of water table fluctuations in unconfined aquifers due to varying bed slopes and spatial location of the recharge basin.***J Hydrol Eng*2010,**15**(11):909–917. 10.1061/(ASCE)HE.1943-5584.0000267View ArticleGoogle Scholar - Chen L, Young MH:
**Green-Ampt infiltration model for sloping surfaces.***Water Resource Res*2006,**42:**1–9.Google Scholar - EERI (2009) Learning from earthquakes: the MW 7.6 Western Sumatra earthquake of September 30, 2009. [last accessed: April 8, 2013], [http://www.eeri.org/site/images/eeri_newsletter/2009_pdf/Padang-eq-report-NL-insert.pdf]
- Green WH, Ampt GA:
**Studies on soil physics: 1. Flow of air and water through soils.***J Agric Sci*1911,**4:**1–24. 10.1017/S0021859600001441View ArticleGoogle Scholar - Green RA, Mitchell JK, Polito CP:
*An energy-based excess pore pressure generation model for cohesionless soils. Proceedings of the John Booker Memorial Symposium. Sydney, New South Wales, Australia, November 16–17, 2000. A.A*. Balkema Publishers, Rotterdam; 2000.Google Scholar - Hsu S-M, Ni C-F, Hung P-F:
**Assessment of three infiltration formulas based on model fitting on Richard’s equation.***J Hydrol Eng*2002,**7**(5):373–379. 10.1061/(ASCE)1084-0699(2002)7:5(373)View ArticleGoogle Scholar - Hyodd M, Hyde AFL, Aramaki N:
**Liquefaction of crushable soils.***Géotechnique*1998,**48**(4):527–543. 10.1680/geot.1998.48.4.527View ArticleGoogle Scholar - Ishihara K (1985) Stability of natural deposits during earthquakes. Proceedings, 11
^{th}International Conference on Soil Mechanics and Foundation Engineering, San Francisco, Vol. 1. pp 321–376Google Scholar - Jafarian Y, Towhata I, Baziar MH, Noorzad A, Bahmanpour A:
**Strain energy based evaluation of liquefaction and residual pore water pressure in sands using cyclic torsional shear experiments.***Soil Dynam Earthq Eng*2012,**35:**13–28. 10.1016/j.soildyn.2011.11.006View ArticleGoogle Scholar - Lee KL, Albaisa A:
**Earthquake-induced settlement in saturated sands.***J Soil Mech Found Div ASCE*1974,**100:**387–406.Google Scholar - Lee CJ, Sheu SF:
**The stiffness degradation and damping ratio evolution of Taipei Silty Clay under cyclic straining.***Soil Dynam Earthq Eng*2007,**27:**730–740. 10.1016/j.soildyn.2006.12.008View ArticleGoogle Scholar - Lenart S:
*The use of dissipated energy at modeling of cyclic loaded saturated soils. Earthquake Engineering: New Research*. Nova Science Publishers, Inc., New York; 2008.Google Scholar - Marks S, Larkin TJ, Pender MJ:
**The dynamic properties of a pumiceous sand.***NZNSEE Bull*1998,**31**(2):86–100.Google Scholar - Muñoz-Carpena R, Regalado CM, Álvarez-Benedí J, Bartoli F:
**Field evaluation of the new Philip–Dunne permeameter for measuring saturated hydraulic conductivity.***Soil Sci*2002,**167:**9–24. 10.1097/00010694-200201000-00002View ArticleGoogle Scholar - Orense RP, Pender MJ (2013) Liquefaction characteristics of crushable pumice sand, Proceeding of the 18
^{th}International Conference on Soil Mechanics and Geotechnical Engineering. 2–6 September, Paris. pp 1559–1562Google Scholar - Petersen M, Harmsen S, Mueller C, Haller K, Dewey J, Luco N, Crone A, Lidke D, Rukstales K (2007) Documentation for the Southeast Asia seismic hazard maps. Administrative report, U.S. Geological Survey. p 67Google Scholar
- Philip JR:
**Approximate analysis of falling-head lined borehole permeameter.***Water Resource Res*1993,**29:**3763–3768. 10.1029/93WR01688View ArticleGoogle Scholar - Regalado CM, Ritter A, Álvarez-Benedí J, Muñoz-Carpena R:
**Simplified method to estimate the Green–Ampt wetting front suction and soil sorptivity with the Philip–Dunne falling-head permeameter.***Vadose Zone Journal*2005,**4:**291–299. 10.2136/vzj2004.0103View ArticleGoogle Scholar - Seed HB, Idriss IM, Makdisi F, Banerjee N:
*Representation of irregular stress time histories by equivalent uniform stress series in liquefaction analyses*. University of California, Berkley; 1975.Google Scholar - Seed HB, Martin PP, Lysmer J:
**Pore water pressure change during soil liquefaction.***J Geotech Eng Div ASCE*1976,**102**(4):323–346.Google Scholar - Sipayung SB, Lely QA, Bambang DD, Sutikno A:
**The analysis of rainfall pattern in Indonesia based on global circulation model (GCM) output.***Jurnal Sains Nusantara*2007,**4**(2):145–154. in Indonesian in IndonesianGoogle Scholar - Skempton AW:
**The pore pressure coefficients A and B.***Geotechnique*1954,**4:**143–7. 10.1680/geot.1954.4.4.143View ArticleGoogle Scholar - Sloan PG, Moore ID:
**Modeling subsurface stormflow on steeply sloping forested watersheds.***Water Resources Res*1984,**20:**1815–1822. 10.1029/WR020i012p01815View ArticleGoogle Scholar - Steenhuis TS, Parlange JY, Sanford WE, HeiligA SF, Walter MF:
**Can we distinguish Richards’ and Boussinesq’s equations for hillslopes?: The Coweeta experiment revisited.***Water Resources Res*1999,**35**(2):589–593. 10.1029/1998WR900067View ArticleGoogle Scholar - Sugano T, Yanagisawa E:
*Cyclic undrained shear behaviour of sand under surface wave stress conditions*. 1992.Google Scholar - Suzuki M, Yamamoto T:
*Liquefaction characteristic of undisturbed volcanic soil in cyclic triaxial test*. 2004.Google Scholar - Wang FW, Sassa K:
**Relationship between grain crushing and excess pore pressure generation by sandy soils in ring shear tests.***J Nat Disaster Sci*2000,**22**(2):87–96. 10.2328/jnds.22.87View ArticleGoogle Scholar - Wang FW, Muhammad Wafid AN, Zhang F:
**Tandikek and Malalak flowslides triggered by 2009.9.30 M7.6 Sumatra earthquake during rainfall in Indonesia, Geoscience Report.***Shimane Univ*2010,**29:**1–10.Google Scholar - Wu P, Hara M, Hamada JI, Yamanaka MD, Kimura F:
**Why a large amount of rain falls over the sea in the vicinity of western Sumatra Island during nighttime.***J Appl Meteorol Clim*2009,**48:**1345–1361. 10.1175/2009JAMC2052.1View ArticleGoogle Scholar - Yamazaki F, Towhata I, Ishihara K (1985) Numerical model for liquefaction problem under multi-directional shearing on horizontal plane, Proceeding of Fifth International Conference on Numerical Methods in Geomechanics. 1–5 April 1985, Nagoya, Japan. p 399Google Scholar
- Žlender B, Lenart S:
**Cyclic liquefaction potential of lacustrine carbonate silt from Julian Alps.***Acta Geotechnica Slovenica*2005,**2005**(1):23–31.Google Scholar

## Copyright

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited.