Geotechnical Earthquake Engineering and Soil Dynamics V GSP 293 98

The Takanodai Landslide, Kumamoto, Japan: Insights from Post-Earthquake Field

Observations, Laboratory Tests, and Numerical Analyses
Gabriele Chiaro, Ph.D.1; Mohammed Umar2; Takashi Kiyota, Ph.D.3;
and Christopher Massey, Ph.D.4
1
Dept. of Civil and Natural Resources Engineering, Univ. of Canterbury, Private Bag 4800,
Christchurch 8140, New Zealand. E-mail: gabriele.chiaro@canterbury.ac.nz
2
Institute of Industrial Science, Univ. of Tokyo, Be 205, 4-6-1 Komaba, Meguro-ku, Tokyo 153-
Downloaded from ascelibrary.org by University of Surrey on 04/15/23. Copyright ASCE. For personal use only; all rights reserved.

8505, Japan. E-mail: umar@iis.u-tokyo.ac.jp

3
Institute of Industrial Science, Univ. of Tokyo, Be 204, 4-6-1 Komaba, Meguro-ku, Tokyo 153-
8505, Japan. E-mail: kiyota@iis.u-tokyo.ac.jp
4
GNS Science, PO Box 30-368, Lower Hutt 5040, New Zealand. E-mail: C.Massey@gns.cri.nz

ABSTRACT
The 2016 Kumamoto earthquakes, Japan, caused a number of geo-disasters in the Aso
Volcanic Caldera, including a large-scale flow-type slope failure known as the Takanodai
landslide. Between April and October 2016, the authors conducted a series of post-earthquake
geotechnical damage surveys and field investigations in the affected areas, and retrieved samples
of volcanic soils to be characterized in the laboratory. In this paper, first, the Takanodai landslide
is described. Then, results of monotonic and cyclic undrained torsional simple shear tests carried
out on reconstituted specimens of Aso pumice are shown. Moreover, to provide insights into the
failure mechanism of the Takanodai landslide, results of dynamic soil response and seismic slope
stability analyses are presented. The numerical investigation confirmed that the pumice soil
deposit was responsible for the landslide triggering. Besides, the combination of pore pressure
build-up and large inertial forces were the key factors for the flow-type Takanodai slope failure.

INTRODUCTION
On 16 April 2016, a moment magnitude (Mw) 7.0 earthquake struck the Island of Kyushu,
Japan. It was preceded by two major foreshocks of Mw 6.2 (14 April) and Mw 6.0 (15 April). The
earthquake sequence contributed to devastation in the mountainous areas of the Aso Caldera,
where traces of medium to large scale landslides and rock falls were frequently observed (Dang
et al., 2016; Mukunoki et al., 2016; Kiyota et al., 2017; Chiaro et al., 2017a). Among many
landslides, the earthquakes caused the failure of a gentle slope near the Aso Volcanological
Laboratory in Minamiaso Village. This large-scale runout slope failure, known as the Takanodai
landslide (Mukunoki et al., 2016), was not classified as a potential earthquake-induced landslide
(Dang et al, 2016). However, it destroyed at least 7 houses and killed 5 people, threatened many
other houses and blocked several roads (Kiyota et al., 2017; Chiaro et al., 2017a).
Earliest field observations suggested that the key soil to cause the slope failure could be the
orange-colored pumice soil deposit (Mukunoki et al., 2016; Chiaro et al., 2017a). Moreover, the
sensitivity ratio of shear strength of this volcanic soil before and after the earthquake as well as
the water pressure buildup could be the cause for the flow type slope failure. However, further
site investigations and in-depth laboratory and numerical studies were required to clarify the
mechanisms for such a flow-type failure of the Takanodai gentle slopes.
Between April and October 2016, as a part of the NZSEE LFE Kumamoto Mission and J-
Rapid Kumamoto Project, the Authors conducted a series of geotechnical damage surveys and

© ASCE

Geotechnical Earthquake Engineering and Soil Dynamics V

 Geotechnical Earthquake Engineering and Soil Dynamics V GSP 293 99

field investigations in the Aso Caldera, and retrieved samples of volcanic soils at the Takanodai
site to be characterized in the laboratory. One of the primary objectives of this research effort
was to provide in-depth understandings into the failure mechanism of the Takanodai landslide
and evaluate the liquefaction potential of the Aso pumice believed to be the key soil responsible
for the activation of the landslide. To this scope, a series of monotonic and cyclic undrained
torsional simple shear tests were carried out on reconstituted specimens of the Aso pumice.
Moreover, a numerical investigation including dynamic soil response and seismic slope stability
analysis were performed by using the Quake/W (Quake/W, 2014) and Slope/W (Slope/W, 2012)
Downloaded from ascelibrary.org by University of Surrey on 04/15/23. Copyright ASCE. For personal use only; all rights reserved.

software.

Figure 1. Geology of Aso Caldera (https://gbank.gsj.jp/) and Takanodai landslide site.

In this paper, main field observations, initial laboratory test and preliminary numerical
simulation results are presented and discussed. Furthermore, the importance of accounting for the
presence of problematic volcanic soil deposits (in addition to geometry consideration and
evidence of past landslide) in developing landslide hazard maps is discussed.

GEOLOGY OF THE MOUNT ASO VOLCANIC CALDERA

Munt Aso is the largest active volcano in Japan, its peak is about 1,592 m above sea level.
The Caldera varies in width from 18 to 25 km, and the central “cone” consists of five peaks
(volcanic craters/vents), the highest being Mt. Taka (1,592 m above sea level). The volcano (one
of the central vents) has been active over the past 70 years. The geology of the Aso Caldera is
shown in Figure 1, which is taken from the 1:50,000 scale geology map produced by the
Geological Society of Japan. The volcanic rocks forming the Aso Caldera are mainly late

© ASCE

Geotechnical Earthquake Engineering and Soil Dynamics V

 Geotechnical Earthquake Engineering and Soil Dynamics V GSP 293 100

Pleistocene non-alkaline felsic and mafic volcanic rocks, comprising ignimbrite, volcanic breccia
and some basalt lava flows. Most of the slopes within the inner caldera are covered by volcanic
pyroclastic soils (tephra), ranging from a few meters to tens of meters in thickness. These soils
are known to be sensitive to pore-water pressure changes and earthquake loading, and there have
been numerous past studies on landslides in these materials (e.g. Paudel et al., 2006; Chigira and
Suzuki, 2016).
Downloaded from ascelibrary.org by University of Surrey on 04/15/23. Copyright ASCE. For personal use only; all rights reserved.

Figure 2. (a) Google Earth image showing the location of the Takanodai landslide and its
impacts on houses and buildings (modified from Chiaro et al., 2017a); and (b) Cross-
section A-A’ through the Takanodai landslide (modified from Chiaro et al., 2017a).
The slopes forming the Inner caldera, range from gentle angles (<10 degrees) on the lower
slopes, to very steep (>60 degrees) upslope towards the central cone. The gentler lower slopes
are mantled in volcanic soils, mainly tephra, and including pumice and ash, which can be many
meters thick and underlain by pyroclastic deposits (ignimbrites) and lava flows. The upper slopes
are typically formed in rock (lavas and ignimbrites) with a shallow (up to 10 meters thick but
typically only a few meters thick) mantle of volcanic soil. The Takanodai landslide is located on
the gentler slopes formed in these volcanic soils. The landslide is reported to have occurred in
tephra, within which were several pumice layers (Chiaro et al., 2017a; Wang, 2017). Field
observations reports by Chiaro et al. (2017a), and Wang (2017) infer the landslide slide surface
to be at the boundary between a pumice layer and the underlying scoracious tephra.

© ASCE

Geotechnical Earthquake Engineering and Soil Dynamics V

 Geotechnical Earthquake Engineering and Soil Dynamics V GSP 293 101
Downloaded from ascelibrary.org by University of Surrey on 04/15/23. Copyright ASCE. For personal use only; all rights reserved.

Figure 3. View of the Takanodai landslide: (a) Looking downhill, with the damaged
Takanodai Housing Complex visible in the distance; and (b) Looking uphill, with the Aso
Volcanological Laboratory visible in the distance (photos taken in May 2016).

Figure 4. (a) Traces of pumice soil on the slip surface; and (b) Soils exposed in wall of a trial
pit excavated in the northern flank of the landslide (photos taken in May 2016).
FIELD OBSERVATIONS
Takanodai landslide characteristics: Field observations indicated that the Takanodai
landslide was a mobile earth slide that developed into a flow-type slide on a low angle slope
(around 12-15o), with a travel angle from landslide crown to debris toe of approximately 6o
(Figure 2(b)). The landslide moved in three different directions from a common source (Figure
2(a)). It was about 100 m in width, 300-600 m in length and 5-10 m in depth. As shown in Figure
3, relatively large intact blocks of soil, grass and trees travelled towards the toe of the slope
suggesting a translational movement of the soil. Tension cracks and scarps above the head scarp
adjacent the Aso Volcanological Laboratory were commonly observed, indicating the potential
for head scarp retrogression and/or new incipient landslides to occur. Figure 4(a) shows the
identified slip surface and the traces of orange-colored pumice soil that were noted on it.
Field investigation and soil sampling: Two small trial pits were excavated across the slip
surface (see the location in Figure 1). Soil exposed in the trial pit is shown in Figure 4(b), and a
borehole describing a typical soil profile up to the depth of 14 m is reported in Figure 5. The soil
consists of: (1) clay-like Kuroboku volcanic ash with organic contents (black color); (2) clay-like
Akaboku volcanic ash (red/brown color); (3) pumice soil (orange color); and (4) soft/weathered
rock.
One of the primary objectives of this research effort was to provide in-depth understandings

© ASCE

Geotechnical Earthquake Engineering and Soil Dynamics V

 Geotechnical Earthquake Engineering and Soil Dynamics V GSP 293 102

into the failure mechanism of the Takanodai landslide and evaluate the liquefaction potential of
the Aso pumice deposit believed to be the key soil responsible for the activation of the landslide.
Therefore, undisturbed and disturbed samples of Aso pumice soils were retrieved to be
characterized in the laboratory. It is worth to mention that water seepage was clearly observed
within the pumice soil layer during the field survey, indicating that: (i) the pumice soil could be
fully saturated at the time of the earthquake; and (ii) the Akaboku ash soil (clay-like soil), within
which the pumice soil is sandwiched, is a much less permeable deposit than the pumice one.
Downloaded from ascelibrary.org by University of Surrey on 04/15/23. Copyright ASCE. For personal use only; all rights reserved.

Figure 5. Typical borehole data for the Takanodai landslide area.

LABORATORY INVESTIGATION
Test apparatus, material and procedure: Laboratory testing was carried out using the fully
automated torsional apparatus shown in Figure 6, which has been developed in the Institute of
Industrial Science, University of Tokyo (Kiyota et al., 2008). Such a device is capable of
achieving double amplitude shear strain levels exceeding 100% by using a belt-driven torsional
loading system that is connected to an AC servo motor through electro-magnetic clutches and a
series of reduction gears. Torque moment and axial load are measured by a two-component load
cell, which is installed inside the pressure cell. The axial load and torque moment capacities are 8
kN and 0.15 kNm, respectively. Difference in pressure levels between the cell pressure and the
pore water pressure are measured by a high-capacity differential pressure transducer (HCDPT)
with a capacity of over 600 kPa. On the other hand, volume change during the consolidation
process is measured by low-capacity differential pressure transducer (LCDPT). A potentiometer
with a wire and a pulley is employed to measure the rotation angle of the top cap and, thus, the
large torsional deformation. Shear stress amplitude is controlled by a computer, which monitors
the outputs from the load cell, computes the corresponding stress value and controls the device

© ASCE

Geotechnical Earthquake Engineering and Soil Dynamics V

 Geotechnical Earthquake Engineering and Soil Dynamics V GSP 293 103

accordingly. Stress and strain calculations are described in detail in Chiaro et al. (2013b) and
Chiaro et al. (2017b).
Downloaded from ascelibrary.org by University of Surrey on 04/15/23. Copyright ASCE. For personal use only; all rights reserved.

Figure 6. Torsional shear apparatus employed in this study.

Figure 7. Results of monotonic undrained torsional simple shear tests ASO#1 and ASO#2.
Hollow cylindrical specimens with dimension of 150 mm in outer diameter, 90 mm in inner
diameter and 300 mm in height were prepared (Table 1). The tests were performed on Aso
pumice samples collected from the Takanodai landslide site (Figure 1). The soil had an in-situ
dry density of approximately 0.6 g/cm3, a natural water content of 117%, and a specific gravity
of 2.34.
Similar to other volcanic soils, Aso pumice has a highly varying specific gravity of
individual grains caused by occluded air pockets. Accordingly, conventional specimen
preparation methods are not suitable (Hyodo et al., 1998). For instance, in the case of water
pluviation volcanic soils tend to segregate, with some grains floating on the water surface. Using

© ASCE

Geotechnical Earthquake Engineering and Soil Dynamics V

 Geotechnical Earthquake Engineering and Soil Dynamics V GSP 293 104

the wet tamping method, crushing of particles may become very significant. Air pluviation also
has been found to be problematic. By trial and errors, in this study, the following procedure was
found most appropriate to build hollow cylindrical specimens with uniform density. To minimize
the particle crushing during the specimen preparation, each specimen was prepared in 15 layers
of equal height by spooning the soil (prepared at its natural water content) into the mold and
subsequently applying a very gentle tamping until a target density was achieved.
Due to high internal void ratio, saturation of volcanic soil requires a rigorous de-airing
process (Hyodo et al., 1998; Pender et al., 2006). In this study, therefore, the double vacuum
Downloaded from ascelibrary.org by University of Surrey on 04/15/23. Copyright ASCE. For personal use only; all rights reserved.

method (Ampadu, 1991) was employed before and during percolation of water into the
specimen. By doing so, and applying a back pressure of 200 kPa, a high degree of saturation was
ensured i.e. Skempton’s ‘B’ value was greater than 0.95. After completing the saturation process,
the specimens were isotropically consolidated by increasing the effective mean stress state (p0’)
up to 100 kPa (representative of in-situ stress conditions).
As summarized in Table 1, several strain-controlled (shear strain rate of 0.5 %/min) cyclic
and monotonic undrained torsional simple shear tests were performed. In the case of cyclic tests,
the loading direction was reversed when the amplitude of shear stress reached the target value.
Whereas, in some monotonic tests, the static shear corresponding to the in-situ sloping stress
condition (i.e. shear stress component induced by slope inclination) was applied by means of
drained monotonic torsional shear loading before undrained monotonic shearing (Chiaro et al.,
2012; Chiaro et al., 2013a). In all the tests, during the undrained shearing phase, the vertical
displacement of the top cap was mechanically prevented with the aim to simulate as much as
possible the simple shear condition that ground undergoes during horizontal seismic excitations.

Table 1. List of undrained cyclic torsional simple shear tests performed in this study
Dry Cyclic shear Static shear
Mean effective stress
Test Type density stress* stress*
(kg/m3) (kPa) (kPa) (kPa)
ASO#1 Monotonic 560 - 0
ASO#2 Monotonic 600 - 25 98-100
ASO#3 Cyclic 630 20 0
ASO#4 Cyclic 580 25 0
*Corrected for the effect of membrane force as reported in Chiaro et al. (2017c)

Monotonic test results: Figure 7 displays the results of the two monotonic undrained
torsional simple shear tests (Aso#1 and Aso#2), in terms of effective stress paths and stress-
strain relationships.
As shown by Figures 7(a), under the stress conditions used in this study, Aso pumice soil
exhibited a purely contractive behavior, during which the mean effective stress (p’) decreases
while shear stress (τ) progressively increases up to a transient peak stress, in correspondence of
the instability line or IL line (Lade, 1993). The peak stress state is accompanied by an unstable
behavior with a sudden loss of strength and a large development of pore water pressure and shear
strains. In addition, by looking at the stress-strain responses shown in Figures 7(b), strain-
softening behavior (i.e. decrease in shear strength owing to shear strain increase) can be observed
after the transient peak stress. Moreover, Aso pumice soil showed a tendency to deform under a
nearly constant stress (residual shear strength).
To investigate the effects of slope inclination on the undrained response of Aso pumice soil,
the specimen Aso#2 was pre-sheared by applying 25 kPa of initial static shear. The overall

© ASCE

Geotechnical Earthquake Engineering and Soil Dynamics V

 Geotechnical Earthquake Engineering and Soil Dynamics V GSP 293 105

behavior of Aso pumice soil subjected to monotonic shearing loading did not change
significantly compared to that Aso#1without initial static shear, in the sense that it remained
purely contractive and strain-softening followed by large strain development was observed after
the peak stress state. However, an increase in both the peak shear strength and the residual
strength was noticed.
Downloaded from ascelibrary.org by University of Surrey on 04/15/23. Copyright ASCE. For personal use only; all rights reserved.

Figure 8. Results of cyclic undrained torsional simple shear tests ASO#3 and ASO#4.
0.4
Torsional simple shear
Cyclic stress ratio, cyc/p0'

D =6 Air pluviation; p0'=100 kPa

r 5%
0.3

Aso pumice
0.2
D =2 D =4
r 5% r 5-50
Experimental data %
Christchurch sand
0.1 Chiaro et al. (2017c)
Toyoura sand
Kiyota (2008); Chiaro et al. (2012, 2013a)
0.0
0.1 1 10 100 1000
Number of cycles to cause DA= 7.5%
Figure 9. Cyclic strength of Aso pumice (d ≈ 600 kg/m3).
Cyclic test results: Due to limited amount of soil available at the time writing this paper,
only two strain-controlled undrained shear tests were performed using cyclic stress ratio (CSR =
τcylic/p0’) of 0.20 and 0.25. Figure 8 shows the test results plotted in terms of effective stress
paths and associate stress-strain relationships.
In both tests, Aso pumice soil showed a cyclic mobility failure type where the effective stress
path progressively moves toward the zero-p’ condition (i.e. full liquefaction) but the cyclic shear
strain increased up to large values. In particular, initially the shear strain were very small and
then suddenly exceed values of double amplitude shear strain (γDA) of 50% or more within the
last few cycles of loading as p’ approached to zero. Hyodo et al. (1998) presented similar
behavior for the crushable Shirasu volcanic soil (from Southern Kyushu) subjected to cyclic
triaxial shearing.
Resistance against liquefaction: Conventionally, the resistance to liquefaction or to cyclic
strain accumulation of soils is evaluated as the number of cycles to develop a specific amount of
double amplitude shear strain (γDA) during cyclic loading. Accordingly, Figure 9 reports the
liquefaction resistance of Aso pumice soil corresponding to double amplitude shear strain (γDA)

© ASCE

Geotechnical Earthquake Engineering and Soil Dynamics V

 Geotechnical Earthquake Engineering and Soil Dynamics V GSP 293 106

of 7.5%, which is equivalent to double amplitude shear strain of 5% in triaxial tests. Despite
some differences in the cyclic behavior (i.e. stress-strain relationships and effective stress paths)
between crushable pumice soils and typical of hard-grained Toyoura and Christchurch sands
(Chiaro et al., 2012; Chiaro et al., 2017c), it appears that the liquefaction resistance of Aso
pumice can be closely related to that of typical less-crushable medium-dense sands tested under
the same conditions and using the same torsional shear apparatus.

NUMERICAL MODELING, DYNAMIC SLOPE RESPONSE AND STABILITY

ANALYSIS
Downloaded from ascelibrary.org by University of Surrey on 04/15/23. Copyright ASCE. For personal use only; all rights reserved.

The main types of landslide hazard identified at the Takanodai site originating from the
volcanic soil slope was earth/debris flow, which is a relatively fluid and rapid type of landslide.
Field observations also suggested that the key soil material in which the slide surface developed
was the pumice, which was saturated at the time of the field visit. On this basis, as the potential
triggering mechanism for the assessed source areas, earthquake-induced liquefaction of the Aso
pumice (i.e. failure through the pumice soil deposit) was anticipated. To confirm the assumed
failure mechanisms, a two-dimensional (2-D) numerical model of the Takanodai landslide was
created based on the slope geometry shown in Figure 2(b) and the typical soil profile shown in
Figure 5. The numerical model was implemented into Quake/W software to evaluate the
dynamic soil response. Then, Slope/W software was used to evaluate (i.e. back-analyze) the
seismic slope stability of the Takanodai slope.
Dynamic slope response assessment: 2-D dynamic slope response assessment using
Quake/W software (Quake/W, 2014) was carried out adopting the acceleration records of the
main earthquake (Mw 7.0), known to have triggered the Takanodai landslide. Specifically, the E-
W horizontal acceleration component from the strong motion station KMM005 was used (Figure
10). It should be noted that acceleration attenuation with distance from source to site, local site-
specific effects due to topography and impedance contrast between rock-clay and clay-pumice
layers were not considered in this pilot analysis. They will be object of future detailed
investigations.

Figure 10. E-W horizontal acceleration records from KMM005 ground motion station
(http://www.kyoshin.bosai.go.jp/) used in the numerical Quake/W- Slope/W analysis.
The equivalent linear soil behavior model was used for the assessment (Kramer, 1996). Pore
water pressure (PWP) generation was enabled. Model parameters for the Aso pumice, including
the PWP function, were obtained from the analysis of the test results shown in Figures 7 and 8.
On the other hand, the strain-dependent shear-modulus reduction and damping functions for the
rock materials and clayey deposits were based on data from the literature for similar volcanic
soils (e.g. Mayer et al., 2005). Compression (Vp) and shear (Vs) wave velocities from borehole

© ASCE

Geotechnical Earthquake Engineering and Soil Dynamics V

 Geotechnical Earthquake Engineering and Soil Dynamics V GSP 293 107

logs at the Mashiki Town allowed for the estimation of the stiffness of the stiff clay and the
bedrock layers. Then, combining the adopted Vs measurements with the actual soil density, the
small-strain shear moduli (Gmax) were estimated (Table 2).

Table 2. Soil properties used in the numerical analyses

Depth Bulk unit weight Gmax Cohesion Friction angle
Layer Soil Type
(m) (kN/m3) (MPa) (kPa) (o)
1a Kuroboku 0-0.6 11.2 27 15.3 23.7
Downloaded from ascelibrary.org by University of Surrey on 04/15/23. Copyright ASCE. For personal use only; all rights reserved.

2a Akaboku 0.6-3.0 12.9 27 15.3 23.7

1b Kuroboku 3.0-4.2 11.2 27 15.3 23.7
2b Akaboku 4.2-8.6 19.2 35 15.3 23.7
3 Pumice 8.6-9.8 11.3 50 0 40
2c Akaboku (Stiff) 9.8-12.5 19.2 100 150 23.7
4 Soft Rock >12.5 22.0 450 - -

Figure 11. (a) FOS under static conditions; and (b) FOS under seismic conditions.
Seismic slope stability analysis: The dynamic stress response computed with Quake/W
were evaluated using Slope/W Newmark function to examine the stability of the slope subjected
to earthquake shaking using a procedure similar to the Newmark (1965) method (Slope/W, 2012).
The soil properties used for the Slope/W assessment are summarized in Table 2.
For each trial slide surface, Slope/W uses: 1) the initial lithostatic stress condition to establish
the static strength of the slope (i.e., the static factor of safety); and 2) the dynamic stress (from
Quake/W) at each time step to compute the dynamic shear stress of the slope and the factor of
safety (FOS) at each time step during the modelled earthquake. Slope/W determines the total
mobilized shear arising from the dynamic inertial forces. The dynamically driven mobilized
shear force is then divided by the total slide mass to obtain an average acceleration (Kmax) for a
given slide surface at a given time step. The acceleration response for the entire potential sliding
mass represents the acceleration value that affects the stability at a given time step during the
earthquake.
Critical layer: As shown in Figure 11(a), the numerical analysis indicated that the lowest
FOS under static conditions was 2.74 (i.e. stable slope). However, under seismic conditions, the
FOS drastically dropped to 0.42 representative of the occurrence of slope failure. Moreover,
looking at the location and shape of the slip surface, it can be seen that the failure plan is within
the pumice soil, confirming that the pumice soil was the key deposit triggering the landslide.
Failure mechanism: The dynamic analysis indicated that the earthquake generated dynamic
shear stresses so high that the peak shear strength of the pumice layer (i.e. 37 kPa, as shown by

© ASCE

Geotechnical Earthquake Engineering and Soil Dynamics V

 Geotechnical Earthquake Engineering and Soil Dynamics V GSP 293 108

the monotonic tests Aso#2 in Figure 7) was most likely exceeded during the seismic event. In
addition, a partial but significant build-up of PWP within the pumice layer was also computed,
which is consistent with the test results shown in Figure 7a. This suggests that a flow-failure
mechanism similar to that shown by monotonic test Aso#2 (Figure 7), characterized by an abrupt
development of large shear strain occurred in the Takanodai landslide.

DISCUSSION AND ADDITIONAL FIELD OBSERVATIONS

Landslide hazard criteria: As reported by Dang et al. (2016), the Kumamoto prefectural
Downloaded from ascelibrary.org by University of Surrey on 04/15/23. Copyright ASCE. For personal use only; all rights reserved.

government had created “landslide hazard maps” based on national standards for landslide
prevention using the following criteria:
i) steep areas at least 5m high with a slope of 30o or more;
ii) areas below a rapid mountain stream that has formed as alluvial fan; and
iii) areas where landslides have occurred or are at risk of occurring.
The slopes surrounding the Takanodai Housing Complex did not meet any of these criteria
and, thus, were not included in the landslide hazard maps.
These criteria, however, are merely based on consideration of slope geometry and evidence
of past slope failure. They do not account for the presence or not of problematic soils (i.e.
liquefiable and/or crushable volcanic soils), the thickness of such challenging volcanic deposits
and the depth from ground surface of critical layers. As shown by this preliminary study, if well-
planned field investigations, laboratory tests and numerical dynamic analyses (accounting for
slope geometry, actual soil profile and soil properties, liquefaction potential of volcanic soils) are
used, dangerous slopes similar to that of Takanodai one, could have been identified and,
therefore, improved earthquake-induced landslide hazard maps in volcanic areas could have been
obtained.
Cyclic behavior of volcanic soils: Worldwide, large-magnitude earthquakes usually trigger
a large number of slope failure in momentous areas (Yin et al., 2009; Chiaro et al., 2015; Dellow
at al., 2017). However, earthquake-induced slope failures in volcanic areas are of major concern
due to the presence of problematic volcanic soils (Yamanouchi, 1968; Ishihara and Harada,
1996; Wang, 2017). Nevertheless, to date, very limited research has been undertaken to
comprehensively understand the cyclic undrained behavior of challenging volcanic soils and
their failure mechanisms. In addition, due to peculiar properties i.e. high crushability and
compressibility, high internal void ratio (Wesley, 2001; Pender at al. 2006) existing correlating
to estimate the liquefaction characteristics of hard grained (less crushable) soils are not
applicable to highly crushable volcanic soils. Therefore, it is essential to conduct comprehensive
investigations that enable the understanding of the failure mechanism of volcanic soil under
cyclic shear loading and establishment of soil-specific liquefaction criteria suitable for this
problematic soils.
Immediate post-earthquake recovery and engineering mitigation strategies: As
mentioned earlier, the Takanodai landslide destroyed at least 7 houses and killed 5 people in the
Takanodai Housing Complex. Also, it made impassable a few main roads nearby the residential
complex of Tokyu Country Town (Figure 12a).
Closely after the earthquake, a new road section passing over the landslide debris deposit was
constructed (Figure 12b). The volcanic soil deposit at the toe of the landslide, which was
obstructing the road (Figure 12a), was compacted and the road was built on it. This prevented not
only the removal of a large amount of accumulated soil debris, but also the possibility of further
movement/slide of the unstable slope debris directly above the road.

© ASCE

Geotechnical Earthquake Engineering and Soil Dynamics V

 Geotechnical Earthquake Engineering and Soil Dynamics V GSP 293 109

A segment of the Takanodai slope (No. 7 in Figure 1) did not slide during the earthquake.
Such slope, however, is posing a great threat to many houses of in the Tokyu Country Town. To
protect those houses, the installation of a system of protection against additional potential slope
failure in the Takanodai area was set up (Figure 12c). This will mitigate the impact of the
expected landslide (if any) and save lives and additional economic loss. Yet, it will not prevent
the landslide to occur. Therefore, stabilization work of the slope should be carried out as soon as
feasible.
Downloaded from ascelibrary.org by University of Surrey on 04/15/23. Copyright ASCE. For personal use only; all rights reserved.

Figure 12. Tokyu Country Town: (a) main road blocked by the landslide, as appeared in
May 2016; (b) new road section passing over the landslide debris, as seen in September
2016; and (c) new system of protection against an anticipated landside (see Figure 1).
CONCLUSION
Following reconnaissance damage surveys, a thin deposit of crushable pumice soil was
understood to be responsible for the Takanodai landslide occurred in Minami Aso during the
2016 Kumamoto Earthquakes. However, there were debates in regard to the potential failure
mechanism among researchers. To provide insights into the failure mechanisms of the Takanodai
landslide, well-organized field investigations, advanced laboratory tests and numerical analyses
were carried out by the Authors. The following main conclusion can be drawn from this study:
1. Monotonic and cyclic undrained torsional simple shear tests revealed that Aso pumice
has the tendency to show a flow-type failure behavior under shear loading, characterized
by an abrupt develop of large shear strains exceeding 50%. Such behavior is led by a
significant build-up of excess pore water pressure, causing the effective stress to drop
considerably.
2. Dynamic soil response and seismic slope stability analyses confirmed that Aso pumice
layer was responsible of the activation of the landslide (i.e. failure developed within the
pumice soil layer). The FOS under static conditions was 2.7 and dropped to a values of
0.4 during the earthquake shaking.
3. Simultaneous pore pressure build-up and large inertial forces most likely was the cause of
the flow-type Takanodai slope failure.
This paper presents background information (including data, numerical model and
descriptions) to help develop and inform discussions on the failure mechanism of the Takanodai
landslide. The results of this study are not exhaustive and farther advanced numerical studies
(effective stress analysis) are required before the actual Takanodai landslide failure mechanism
can be confirmed.

ACKNOWLEDGMENTS
The Authors would like to acknowledge the financial support from the New Zealand Society

© ASCE

Geotechnical Earthquake Engineering and Soil Dynamics V

 Geotechnical Earthquake Engineering and Soil Dynamics V GSP 293 110

for Earthquake Engineering (NZSEE LFE Mission), the University of Canterbury (UC CNRE
Strategic Grant), the University of Tokyo, GNS and the Japan Science and Technology Agency
(JST J-Rapid Kumamoto Project). Contribution of Kevin Chew and Joon Su Kim in conducting
the dynamic soil response and seismic slope stability analyses is gratefully appreciated. The
ground motion data were obtained from http://www.kyoshin.bosai.go.jp/. The Geological Map of
the Aso Volcano was obtained from https://gbank.gsj.jp/ .

REFERENCES
Downloaded from ascelibrary.org by University of Surrey on 04/15/23. Copyright ASCE. For personal use only; all rights reserved.

Ampadu, S. I. K. (1991). Undrained behavior of kaolin in torsional simple shear. Ph.D. Thesis,
Dept. of Civil Engineering, University of Tokyo, Japan.
Chiaro, G., Alexander, G., Brabhaharan, P., Massey, C., Koseki, J., Yamada, S., and Aoyagi, Y.
(2017a). “Reconnaissance report on geotechnical and geological aspects of the 2016
Kumamoto Earthquake, Japan.” Bulletin New Zealand Soc. Earth. Eng., 50(3), 365-393.
Chiaro, G., De Silva, L. I. N., and Koseki, J. (2017b). “Modeling the effects of static shear on the
undrained cyclic torsional simple shear behavior of liquefiable sand.” Geotech. Eng. J.,
SEAGS, 48(4), 1-9.
Chiaro, G., Kiyota, T., and Koseki, J. (2013a). “Strain localization characteristics of loose
saturated Toyoura sand in undrained cyclic torsional shear tests with initial static shear.”
Soils Found., 53(1), 23-34.
Chiaro, G., Kiyota, T., and Miyamoto, H. (2017c). “Liquefaction potential and large deformation
properties of Christchurch liquefied sand subjected to undrained cyclic torsional simple shear
loading.” Proc., 19th Int. Conf. Soil Mech. Geotech. Eng., Seoul, South Korea, 1497-1500.
Chiaro, G., Kiyota, T., Pokhrel, R. M., Goda, K., Katagiri, T., and Sharma, K. (2015).
“Reconnaissance report on geotechnical and structural damage caused by the 2015 Gorkha
Earthquake, Nepal.” Soils Found., 55(5), 1030-1043.
Chiaro, G., Koseki, J., and De Silva, L. I. N. (2013b). “A density- and stress-dependent elasto-
plastic model for sands subjected to monotonic torsional shear loading.” Geotech. Eng. J.,
SEAGS, 44(2), 18-26.
Chiaro, G., Koseki, J., and Sato, T. (2012). “Effects of initial static shear on liquefaction and
large deformation properties of loose saturated Toyoura sand in undrained cyclic torsional
shear.” Soils Found., 52(3), 498-510.
Chigira, M., and Suzuki, T. (2016). “Prediction of earthquake-induced landslides of pyroclastic
fall deposits.” Landslides and Engineered Slopes. Experience, Theory and Practice. Aversa,
S., Cascini, L., Picarelli, L., and Scavia, C. (Eds.), CRC Press, 93-100.
Ishihara, K., and Harada, K. (1996). “Cyclic behavior of partially saturated collapsible soils
subjected to water permeation.” Ground Failures under Seismic Conditions, ASCE, GSP 44,
34-50.
Dang, K., Sassa, K., Fukuoka, H., Sakai, N., Sato, Y., Takara, K., Quang, L. H., Loi, D. H., Tien,
P. V., and Ha, N. D. (2016). “Mechanism of two rapid and long runout landslides in the 16
April 2016 Kumamoto earthquake using a ring-shear apparatus and computer simulation
(LS-RAPID).” Landslides, 13(6), 1525–1534.
Dellow, S., Massey, C., Cox, S., Archibald, G., Begg, J., Bruce, Z., Carey, J., Davidson, J., Della
Pasqua, F., Glassey, P., Hill. M., Jones, K., Lyndsell, B., Lukovic, B., McColl, S.,
Rattenbury, M., Read, S., Rosser, B., Singeisen, C., Townsend, D., Villamor, P., Villeneuve,
M., Wartman, J., Rathje, E., Sitar, N., Adda, A. Z., Manousakis, J., and Little, M. (2017).
“Landslides caused by the Mw7.8 Kaikoura Earthquake and the immediate response”.

© ASCE

Geotechnical Earthquake Engineering and Soil Dynamics V

 Geotechnical Earthquake Engineering and Soil Dynamics V GSP 293 111

Bulletin New Zealand Soc. Earth. Eng., 50(2), 106-116.

Hyodo. M., Hyde, A. F. L., and Aramaki, N. (1998). “Liquefaction of crushable soils.”
Geotechnique, 48(4), 527-543.
Kiyota, T., Sato, T., Koseki, J., and Mohammad, A. (2008). “Behavior of liquefied sands under
extremely large strain levels in cyclic torsional shear tests.” Soils Found., 48(5), 727-739.
Kiyota, T., Ikeda, T., Konagai, K., and Shiga, M. (2017). “Geotechnical damage caused by the
2016 Kumamoto Earthquake, Japan.” Int. J. Geoeng. Case Histories, 4(2), 78-95.
Kramer, S. L. (1996). Geotechnical Earthquake Engineering. Prentice Hall, NJ, U.S., pp. 653.
Lade, P. V. (1993). “Initiation of static instability in the submarine Nerlerk berm.” Canadian
Downloaded from ascelibrary.org by University of Surrey on 04/15/23. Copyright ASCE. For personal use only; all rights reserved.

Geotech. J., 30(6), 895-904.

Meyer, V., Larkin, T., and Pender, M. (2005). “The shear strength and dynamic shear stiffness of
some New Zealand volcanic ash soils”, Soils Found., 45(3), 9–20.
Mukunoki, T., Kasama, T., Murakami, S., Ikemi, H., Ishikura, R., Fujikawa, T., Yasufuka, N.,
and Kitazono, Y. (2016). “Reconnaissance report on geotechnical damage caused by an
earthquake with JMA seismic intensity 7 twice in 28h, Kumamoto, Japan.” Soils Found.,
56(6), 947-964.
Newmark, N. (1965). “Effects of earthquakes on dams and embankments.” Geotechnique, 15,
139-160.
Paudel, P., Omura, H., Kubota, T., and Inoue, T. (2006). “Spatio-temporal patterns of historical
shallow landslides in a volcanic area, Mt. Aso, Japan.” Geomorphology, 88, 21–33.
Pender, M. J., Wesley, L. D., Larkin, T. J., and Pranjoto, S. (2006). “Geotechnical properties of a
pumice sand.” Soil Found., 46(1), 69-81.
Quake/W (2014). Dynamic Modeling with Quake/W. An engineering methodology. October 2014
Edition, GEO-SLOPE International Ltd., pp. 173.
Slope/W (2012). Stability modelling with Slope/W. An engineering methodology. July 2012
Edition, GEO-SLOPE International Ltd., pp. 246.
Wang, G (2017). “Fluidized landsliding phenomena during earthquakes”. Proc., JTC1 Workshop
on Advances in Landslide Understanding, Barcelona, Spain, pp. 4.
Wesley, L. D. (2001). “Determination of specific gravity and void ratio of pumice materials.”
Geotech.l Test. J., ASTM, 24(4), 418–422.
Yin, Y., and Wang, F., and Sun, P. (2009). “Landslide hazards triggered by the 2008 Wenchuan
earthquake, Sichuan, China”. Landslides, 6(2), 139–152.

© ASCE

Geotechnical Earthquake Engineering and Soil Dynamics V