Original Article

Landslides (2006) 3: 149158

DOI: 10.1007/s10346-005-0031-y
Received: 5 March 2005
Accepted: 29 November 2005
Published online: 8 February 2006
Springer-Verlag 2005

H. Yoshimatsu S. Abe

A review of landslide hazards in Japan and assessment of their

susceptibility using an analytical hierarchic process (AHP) method

Abstract In spite of its small size, Japan suffers many landslide disasters due to intense rainfall and earthquakes. This article describes the
distribution and topography of these landslides, and a new method of
evaluating the susceptibility, the analytical hierarchic process (AHP).
The method assigns scores to each factor of micro-topography of
landslide-prone areas identified in aerial photographs, and assesses
the susceptibility of landslide from the total score. In addition, a
method of simulating sliding mass runout is briefly presented for the
designating sediment-related disaster warning areas.
Keywords Landslide distribution . Landslide susceptibility
assessment . Sliding mass runout
Types and damage situation of sedimentrelated disasters in Japan
Kinds and classification of sedimentrelated disasters
Japan has a total land area of 370,000 km2 , of which 70% is mountainous. Although this is only 0.1% of the earths land area, over 10% of all
active volcanoes (86) are located in Japan, and the country accounts
for about 10% of the worlds seismic energy emission. In addition,
Japan receives annual precipitation of 1750 mm, which is almost
double the annual global average, and is struck by approximately 11
typhoons every year. Thus, the country is prone to sediment-related
disasters.
Sediment-related disasters in Japan vary in morphology depending
on whether caused by rainfall, snow melt, earthquake, soil, geological
structure, topography, or other factors. Sediment-related disasters
can be classified either legally or academically. The legal classification is: debris flow, landslide and slope failure, based on differences
in damage features, investigation methods and prevention countermeasures. This classification is based on the fact that Japanese people
have long distinguished between flow, slide and collapse of earth
masses according to the speed of travel, distance traveled, size and
physical properties of traveling soil mass, possibility of recurrence,
and other traveling morphologies. Landslides and slope failures are
distinguished as shown in Table 1.
Sediment-related disasters are classified according to this table by
engineers of governmental bodies, and prevention and evacuation
measures are taken based on the Landslide Prevention Law and the
Law for Prevention of Steep Slope Failure Disaster. Municipal governments then inform local residents of the landslide-risk areas identified. Academically, the classification by Varnes (1978) is used.
States and characteristics of sediment-related disasters
The frequency of sediment-related disasters is affected by the occurrence, scale, and intensity of triggers, such as earthquakes, volcanic
activity, storms, and snow melting. Sediment-related disasters have
been more frequent in recent years probably due to abnormal local

downpours, which may be attributable to global warming, and frequent earthquakes such as the large shallow direct-hit earthquake in
Niigata-Chuetsu (M=6.8) in 2004. Recent sediment-related disasters
have also tended to affect city residents possibly due to residential development in mountainous areas, and some large-scale debris flows
caused by deep-seated collapses have killed many people. Earthquakes
have also caused many sediment-related disasters.
The number of each class of sediment-related disaster is shown in
Fig. 1. The number of disasters shown in this figure is by the legal
classification and includes small slope movements of less than 50 m
in length and 20 m in width. Many slopes consisting of pyroclastic
flow deposit and weathered plutonic granite have collapsed during
typhoons and storms in the rainy season. Landslides are few but
have caused serious damages due to their large scale. The damage
caused to people by each class of sediment-related disaster is shown
in Fig. 2. Debris flows, which are fast, have killed many people, while
landslides, which are slow, have affected few. Figure 3 shows the
generation morphology of sediment-related disasters for each trigger.
Many landslides are triggered by prolonged rain and snow melt, and
debris flows and slope failures are triggered by intense storms and
local downpours during typhoons.
Laws for preventing sediment-related disasters and offices in charge of taking
countermeasures
Risk areas where measures should be taken to prevent debris flows,
landslides and slope failures are designated by the ministers in charge
and prefectural governors under the Sabo Law, Forest Law, Landslide
Prevention Law, and the Law for Prevention of Steep Slope Failure
Disaster. Designated areas are those susceptible to sediment-related
disasters that may directly affect peoples lives and houses. In these
areas, countermeasures are taken by the national and/or prefectural
governments, and warning and evaluation systems are built. The
number of designated landslide prevention areas in Japan and their
areas are shown in Table 2. The Ministry of Land, Infrastructure and
Transport is responsible for risk areas where landslides may damage
rivers and highways, and the Ministry of Agriculture, Forestry and
Fisheries for areas where landslides may damage forests and farmlands. Thus, areas where landslides may directly damage people and
facilities are designated as landslide prevention areas, and there are
many such areas throughout Japan.
Distribution of landslide hazard areas in Japan and their topographic
and geological characteristics
The geology of Japan mainly consists of granite and metamorphic
rock of the Paleozoic to Mesozoic Eras, Tertiary sedimentary rocks,
volcanic rocks, and Quaternary volcanic sediment. These strata are
cut into blocks by a great number of faults, and the small land area
is divided into many geological zones arranged in mosaic patterns

Landslides 3 . 2005

149

Original Article
Table 1 Difference between landslide and slope failure (Japan Landslide Society, 2002)

Geology
Soil
Topography
Nature of movement
Rate of movement
Nature of Moving mass (blocks)
Causes, Triggering mechanism
Size
Warning Signs
Typical original gradient

150

Landslides
Often influenced by geology and geological structure
Move along slip surface(s) that consists of highly plastic clay
Occur on gentle to moderate slopes of 5 to 30 degrees
The upper slopes often have a flat-plateau like topography
Continuous, recurrence(repetitive occurrence)
Duration of a single episode is generally long
Generally slow to very slow 0.0110 mm/day (most common)
Little disturbance within a sliding block
Often move while retaining the original shape and characteristics
Generally influenced by excess groundwater, elevated groundwater table
Surface area is often large, ranging between 1 to 100 ha

Rapid Slope Failures

Little or no influence of geology and geological structure
Usually involve topsoil, residual soil and (highly)
weathered bedrock
Usually occur on slopes steeper than 30 degrees
Occur suddenly
Short duration
Very to extremely rapid 10 m/sec or faster
Incoherently move as highly disturbed mass

Generally influenced by rainfall intensity

Surface area is generally small, with an average volume of
about 440 m3
Often develop cracks, depressions, upheavals, groundwater fluctuation, etc. Hardly any warning signs
prior to sliding
Almost always fail suddenly
10 to 25 degrees
35 to 60 degrees

(Fig. 4). Japan is located where the North American, Eurasian, Pacific, and Philippine Plates collide, and lies on a subduction zone of
the plates where most of the worlds developed volcanoes are located
and earthquakes occur frequently. In terms of geology and geological
structure, landslides in Japan tend to occur in large-scale fault and
fracture zones, areas of Tertiary sedimentary soft rocks, and volcanic
zones (Fig. 5). Landslides are frequent in (1) the Tertiary green tuff areas along the Sea of Japan in the Tohoku, Hokuriku and Chugoku Areas, (2) fractured and metamorphosed zones of Paleozoic-Mesozoic
schist along the median tectonic line, which runs east to west in
Shikoku and Kii Peninsula, (3) the Tertiary coal-bearing formation
zones in northwestern Kyushu, (4) areas along the Fossa magna tectonic line, which runs from Itoigawa to Shizuoka, (5) the serpentine
area along the Mineoka fractured zone in southern Chiba, and (6)
the Sangun metamorphic zone in Okayama Prefecture and other
prefectures in the Chugoku Area. The landslide distribution shown
in Fig. 5 was investigated based on the slide movement styles in the
mass movement classification by Varnes (1978). The landslides vary in
morphology such as the features of debris, rock slide and clayey slide
due to hydrothermal alteration depending on the geology, geological
structure, and topography. In general, landslides in Tertiary zones are

composed of soft and weathered materials such as mud rocks, and

their slope inclinations are usually not steeper than 20 . On the other
hand, the slopes of landslides in fractured zones consisting of rocky
fragment materials are inclined at 2030 .

Fig. 1 Number of sediment disasters (number for 2004 is estimated)

Fig. 2 Number of dead and missing persons due to sediment disasters

Landslides 3 . 2005

Identification of landslide-susceptible areas and methods of assessing

landslide activity
The national and prefectural governments are in charge of identifying landslide susceptibility areas and taking preventive measures.
Main measures include monitoring sliding areas, researching sliding
features, constructing prevention works and building emergency systems in case of sudden landslide due to unpredictable weather and
earthquakes. Areas at particularly high risk of sliding are also identified and assessed. It is important to assess landslide intensity since it
can help mitigate damage beforehand by preventing sediment-related
disasters.
Therefore, the ministries and governmental divisions in charge
conduct periodical surveys on highways, steep slopes, and landslideprone areas based on the relevant laws (once every 15 years). The field
surveys are usually done by consultants, who visually inspect the site,
assess the susceptibility by giving scores based on statistical analysis of

Fig. 3 Causes of sediment disasters

past landslides and engineering experimental information related to

the state of the slope, geology, soil, properties to be protected, history
of disaster, and spring water according to past disasters.
Recently, a new law (Sediment Disaster Prevention Law) was enacted after the sediment disaster in Hiroshima in 1999 during a storm
(325 landslides, 24 people killed). The law requires governments to
understand areas with high landslide risk, communicate the risk,
construct a warning and evacuation system, control the construction
of new houses, etc., and promote the relocation of existing houses.
The governments designate and monitor risk areas as either sediment disaster warning areas or sediment disaster special warning
areas based on the degree of risk of landslides. The scale of such areas
is the length of the landslide from the toe or up to 250 m for warning
areas, and up to 60 m from the toe for special warning areas. When
an area is designated as a sediment disaster warning area, warning
and evacuation systems must be established for the area in the disaster prevention plan of the municipal government, and information
on sediment disasters must be given to residents. When an area is
designated as a sediment disaster special warning system, permission has to be obtained to construct houses, social welfare facilities,
schools and medical facilities in times of disaster; the safety of other
buildings against sediment-related disaster should also be confirmed;
and some buildings are advised to be moved to safe areas. Obviously,
permission is needed for development activities and for selling and
purchasing houses and housing land.
Landslide morphologies for all of Japan are identified in national
research institutes and universities by analyzing the topography using aerial photographs. The landslide distribution was investigated
based on the slide movement styles in the mass movement classification by Varnes (1978). Although there is no uniform method for
expressing the results of topographic analysis of aerial photographs,
the standards shown in Fig. 6 are now being widely used for reading landslide micro-topography. This method estimates the degree

Table 2 Summary of designated

landslides threatened area (Japan Landslide
Society, 2002)

Number of area
Total area (ha)

of landslide activity and its susceptibility by dissecting the landslide

topography and clarifying movement units. A landslide form with
the dissected old landslide morphology is relatively stable, and areas with clear landslide morphology and movement units are judged
unstable. The analytical results are drawn on topographic maps of
1/50,000 in principle, and are partially used for risk assessment by
the governments. Analysis of aerial photographs can only identify
landslide morphology in two dimensions based on the concept of a
surface plane, and does not take account of data regarding underground structures and changes due to landslide movement. Thus, the
method has limitations in assessing the hazard, but has been effective
in practice for quickly understanding the states of damage caused by
sediment-related disasters originating in the Niigata-Chuetsu Earthquake (M6.8, Fig. 7) that occurred in Niigata Prefecture in 2004. The
large and shallow direct-hit earthquake disrupted the traffic network
and caused damages especially in mountainous districts, making it
difficult to grasp the overall damage, so aerial photographs were taken
and analyzed to clarify the damage situation. The characteristics of
these landslides are recorded on highly precise topographic maps of 1m grids for topographic surveys using a laser profiler, and hazardous
areas have been identified by comparative analysis with the geological
structure (Fig. 8).
Recent studies on susceptibility assessment method
Susceptibility assessment of landslide intensity
Various methods of using aerial photographs for monitoring present
landslide movements, history of landslides and predicting movements
in the near future are being investigated in universities and as joint
studies.
Recently, the analytical hierarchic process (AHP) method has been
used to identify areas susceptible to landslide, which assigns scores to
each factor of micro-topography of landslide-prone areas found in
aerial photographs and assesses the susceptibility of landslide from the
total score. The AHP method decomposes the process of subjective
judgment of people into a layer structure, and expresses the process
quantitatively. Here, the susceptibility of landslide means the possibility of another landslide in areas where landslides have occurred in
the past by analyzing aerial photographs. From aerial photographs,
landslide factors are identified based on the following:
1. Degree of erosional dissection of sliding scarp: The oldness of
landslide topography can be estimated from the degree of erosional
dissection of the sliding scarp. Areas that show clear sliding scarps
may be unstable.
2. Surface ground morphology of landslide zone: The landslide zone
is divided into small units by clear ground level differences and
cracks. Ground level differences and cracks on slopes further reduce the shearing resistance of the slopes by weathering. Landslide
zones that have been stable over a long period of time show unclear ground level differences and cracks and gentle and smooth
ground surface. The stability of a slope can be roughly estimated
by observing the characteristics of the ground surface. Here, crack

Ministry of agriculture, forestry and fisheries

Ministry of land, infrastructure Rural development bureau Forestry agency
and transport
3,329
1,868
1,725
114,023
107,061
97,927

Total
6,922
319,011

Landslides 3 . 2005

151

Original Article

Fig. 4 Major engineering geologic divisions of Japan (Japan Landslide Society, 2002)

elements are used to judge characteristics, such as clearness and

continuity.
3. Occurrence position of landslide zone: A large-scale landslide develops into two or more landslide zones due to weathering of
landslide masses and changes in underground water channels during landslide movement. The position of a landslide zone to be
152

Landslides 3 . 2005

investigated in a large-scale landslide area is important and the

positional relationship to the toe of the landslide area must be
understood. The toe of a landslide zone is very unstable since it
has undergone compression stress. Landslide zones located at the
toe of a large-scale landslide area are especially unstable because
the material tends to be composed of clayey soil and the toe zone

Fig. 5 Distribution of designated landslide-threatend zone and estimated hazard zone (modified from Fujita 2002)

has received compression stress by landslide movements in the

past.
4. Ground level differences, open cracks and their locations: Ground
level differences and open cracks can be used to predict landslides. They also supply underground water to landslide zones.
In particular, those at the head serve as routes by which surface
and rain water can flow to the underground. Zones where these
signs are clearly visible are in danger of landslides. Here, crack
elements are used to investigate the location within the landslide
area.
5. Conditions at toe of landslide zone: The toe portion of a landslide
zone which has developed into a tongue-like form is usually very

Table 3 Ranking of landslide susceptibility by AHP score

Susceptibility level

AHP score

Level 1 (high)
Level 2 (slightly high)
Level 3 (slightly stable)
Level 4 (stable)

62<AHP score
38<AHP score
24<AHP score
AHP score<24

Percentage to total number

(%)
5
25
30
40

unstable. When the toe of a landslide zone is strongly eroded, the

susceptibility of landslide is high. Small-scale landslides located at
the toes of large-scale landslides consist of soil mass that has been
crushed during the large-scale landslides and are unstable. Here,
Landslides 3 . 2005

153

Original Article
Fig. 6 Landslide Maps (Shimizu et al. 1996)

toe elements are used to judge the erosion of the toe of the landslide
to be analyzed. The layer structure for assessing the susceptibility is
shown in Fig. 9. An example of landslide susceptibility assessment
using scores obtained from the AHP structure (Table 3) is shown
in Fig. 10.
The layer structure of AHP in Fig. 9 was prepared by
1. Collecting landslides that recently showed sudden activities and
those that did not move;
2. Extracting topographic, geological and movement factors that are
important for assessing the susceptibility of landslide by conducting brain storming and statistical analysis;
3. Determining the AHP layer structure and weighting coefficients
based on the extracted factors, results of statistical analysis, and the
experience and knowledge of researchers and engineers;

154

Landslides 3 . 2005

4. Assessing the landslide data collected again using the determined

layer structure and weighting coefficients, and revising the layer
structure and coefficients to determine the final AHP layer structure and weighting coefficients.
Thus, the AHP layer structure in this method depends on the
collected landslide data, researchers and engineers involved, and their
experience and knowledge.
Topographic maps with contour lines at 1-m intervals, which are
prepared using laser profilers, and orthorectified imagery are increasingly used for identifying landslide hazard areas. Although this type
of analysis of digital photographs suffers elevation errors caused by
trees as obstacles, it is very useful since the photographs are clear,
various topographic factors are easy to analyze, topographic maps
are easy to prepare, and the method is applicable to GIS.

Fig. 7 Terano Landslide (L=360 m,

W=230 m, h=28 m, V=110103 m3 )

Fig. 8 Terano Landslide map (Niigatachuetsu earthquake, M=6.8, Oct. 23, 2004, Yagi et al. 2005)

All these landslide analyses focus only on the morphological features of the ground surface and do not consider underground structure elements, which are also important. As described above, landslides in Japan are strongly affected by the geology and geological
structure. In other words, similar landslide forms are observed in
areas of similar geology and geological structure. Thus, geology and

geological structure are indispensable indices for assessing landslide

susceptibility. Figure 11 shows the results of landslide susceptibility
evaluation based on topographic analysis using aerial photographs,
the results of field reconnaissance and landslide susceptibility scores
determined by analyzing the characteristics of past landslides in this
area from topographic and geological data. Arrows in the figure show
Landslides 3 . 2005

155

Original Article
Fig. 9 Net of Class Structure of AHP
method (Aganogawa R.W.O. 2002)

Fig. 10 Example of landslide susceptibility assessment by AHP method (Aganogawa R.W.O. 2002)

zones that are highly prone to landslide during an earthquake. This

type of comprehensive susceptibility assessment has just started, so
the distribution of points should be improved and an analytical
method that uses both statistical and simulation methods should
be developed.
The AHP method described here is for assessing landslide susceptibility and not for assessing the range over which landslide masses
flow and accumulate. Methods for analyzing the runout of landslide
156

Landslides 3 . 2005

masses and the range over which the masses accumulate are also
needed to mitigate landslide risk, and are briefly described in the
following section.
Predicting the runout area of landslide mass
When designating sediment disaster warning areas under the
recently-enacted Sediment Disaster Prevention Law, it is important
to estimate the area affected by a moving landslide mass.

Fig. 11 Example of landslide risk evaluation in Mogami area in Yamagata Prefecture

Landslides 3 . 2005

157

Original Article

Fig. 12 Features of landslide runout length

Runout areas have been estimated based on the fact that the runout
length of most landslides is no more than double the landslide scar
length (Fig. 12).
A numerical analysis is also conducted to evaluate accurately the
area at risk of being reached by landslide masses that travel long distances. The governing equation of the numerical analysis analyzes
the continuum and momentum equations using the finite difference method or the finite element method. The equation of MohrCoulomb mixture theory (Iverson and Denlinger, 2001) is used to
evaluate the dissipative stress in the analysis. This equation can also
be applied to deep-seated landslides since it has a term to express
bottom friction that stops landslide movement. An analysis using the
finite difference method is shown in Fig. 13 (Zhang et al., 2004).
Combining the methods for identifying landslide susceptibility
areas, such as AHP, and modeling of landslide mass runout enables
comprehensive measures for mitigating the risk of landslide to be
assessed. Landslide risk mitigation measures will involve assessing
the landslide susceptibility of each slope using the AHP method and
estimating the runout length and area over which landslide masses
will travel.
Conclusion
In Japan, the Sediment Disaster Prevention Law was enacted in 2001
to mitigate the risk of sediment-related disasters. For the first time
in the world, the law requires warning and evacuation systems to
be constructed and regulates the construction of buildings in landslide risk areas. Studies are being conducted on quantitative methods
for assessing landslide susceptibility areas that are more precise than
conventional methods. The authors presented an assessment method
that uses landslide topography and microtopography and a method
for predicting areas that will be affected by runout of landslide masses.
Some of these results are already being used by governments to carry
out landslide prevention measures, but the development of risk analysis using GIS and digital elevation morphological data has just started
in Japan and future development is awaited.
References
Aganogawa River Work Office (2002) Report of landslide risk evaluation maps (in Japanese): 171
Fujita T (2002) Landslides and geology (in Japanese), Kokonshoin: 125
Iverson RM, Denlinger RP (2001) Flow of variably fluidized granular masses across three-dimensional
terrain: 1. Coulomb mixture theory. J. Geophys. Res. 160(B1):537552
Japan Landslide Society (2002) Landslides in Japan. National Conference of Landslide Control: 164
Shimizu F, Oyagi N, Inokuchi T (1996) Landslides map Part 1 (in Japanese). N.R.I. for Earth Science and
Disaster Prevention: 112
Varnes DJ (1978) Slope Movement Types and Processes. In Special Report 176: Landslides: Analysis
and Control. In: Schuster RL, Krizek RJ (eds) TRB, National Research: 1133
Yagi H, Yamasaki T, Moriiwa T (2005) Distribution of landslides due to Niigata-Chuetsu-Earthquake in
2005 (in Japanese). Symposium of Niigata-Chuetsu Earthquake Disasters: 311
Zhang C, Yoshimatsu H, Iwahori Y, Abe S (2004) Numerical simulation of grain-fluid flow due to slope
collapse (in Japanese). J. Japan Landslide Soc. 41(1):917
H. Yoshimatsu (
)
Sabo Technical Center,
4-8-1, Kudan-minami, Chiyoda-ku, Tokyo 102-0074, Japan
e-mail: Yoshimatsu@stc.or.jp
Tel.: +81-3-5276-3273
Fax: +81-3-5276-3393
S. Abe
Okuyama Boring Co., Ltd.,
10-39, Shinmei-chou, Yokote City, Akita 013-0046, Japan

Fig. 13 Flow simulation by FDM of landslide occurred in HIROSAKI-City (Zhang et al. 2004)

158

Landslides 3 . 2005