International Journal of Erosion Control Engineering, Vol. 1, 2008

5

Debris Flow Induced by Deep-Seated Landslides at Minamata City, Kumamoto Prefecture, Japan in 2003

Takashi JITOUSONO1, Etsuro SHIMOKAWA1 and Yukiyoshi TERAMOTO1

1 Faculty of Agriculture, Kagoshima University (Kagoshima 8900065, Japan)

A large scale landslide occurred in the Atsumari River Catchment, Minamata City, Kumamoto Prefecture on July 20, 2003. This type of landslide is called a “deep-seated landslide.” The debris flow induced by this landslide killed 15 people. The deep-seated landslide was caused by the rising of groundwater level due to heavy rainfall, hydrogeomorphological formation of the underground area prone to storage of groundwater, and deeply weathered volcanic rocks. Some debris-flow disasters induced by the deep-seated landslides have occurred around this disaster site, such as the disasters at Nishiuchitate in Ebino City, Miyazaki Prefecture in 1972 and at Harihara in Izumi City, Kagoshima Prefecture in 1997. These areas are underlain by quaternary volcanic rocks, andesite, and tuff breccia. The characteristics of the deep-seated landslides and the debris flow induced by it were examined based on a field study in the Atsumari River Catchment. On the basis of the topographical, geological, and hydrological data of the volcanic areas around these sites, this study describes a method of predicting potential deep-seated landslide sites.

1. INTRODUCTION

Heavy rainfall occurred in Kyushu region from July 19 to 20, 2003. This rainfall caused a large scale landslide in the Atsumari River Catchment, Hougawachi, Minamata City, Kumamoto Prefecture (Taniguchi, 2003; Mizuno et al., 2003) inducing a debris flow, which killed 15 people. This large scale landslide was of the type called a “deep-seated landslide.” Recently, debris-flow disasters induced by deep-seated landslides have killed many people and also caused great property loss. Much attention should be paid to deep-seated landslides to prevent disasters.

In this paper, the characteristics of a deep-seated landslide and the debris flow induced by it are examined based on a field study in the Atsumari River Catchment. Furthermore, this study describes a method for predicting potential deep-seated landslide sites in the volcanic areas around this disaster site. 2. DEBRIS-FLOW DISASTER INDUCED

BY DEEP-SEATED LANDSLIDE

The debris-flow disaster induced by a deep-seated landslide occurred in the Atsumari River Catchment on July 20, 2003 (Figure 1). The Atsumari River is located in the upper reaches of the Minamata River running through Minamata City.

The Atsumari River has a small catchment with an area of 1.14 km2, and a river channel of length 2.52 kms (Figure 2). The lower reaches in the Atsumari River Catchment consists of an alluvial fan, which is used as a site for houses and farm. On the alluvial fan, the debris flow killed 15 people, injured 6 people, and destroyed 13 houses.

Deep-seatedlandslide

Atsumari Debris-flowdisaster

Photo by Asia Air Survey Co., Ltd. Fig.1 Deep-seated landslide and debris flow in the Atsumari River Catchment

International Journal of Erosion Control Engineering, Vol. 1, 2008

6

N

m0 500 1000

Kumamoto

Kagoshima

Minamata

Miyazaki

Minamata R.

Atsumari

Deep-seatedlandslide

Debris-flowdisaster

Hougawachi R.

Fig.2 Location and topography of the Atsumari River Catchment

Some debris-flow disasters induced by deep-seated landslides have occurred around this site (Shimokawa et al., 2004), such as those at Nishiuchitate in Ebino City, Miyazaki Prefecture in 1972 (Takahashi, 1974) and at Harihara in Izumi City, Kagoshima Prefecture in 1997 (Shimokawa and Jitousono，1999). Areas around these sites are underlain by quaternary volcanic rocks such as andesite and tuff breccia. This area of quaternary volcanic rocks, called the “Hisatsu volcanic province,” is widely distributed near the boundary in Kagoshima, Kumamoto, and Miyazaki Prefecture (Yamamoto, 1960). 3. RAINFALL

Heavy rainfall occurred in the southern Kyushu on July 20, 2003. Figure 3 shows the temporal variation in the hourly and cumulative rainfalls from July 19 to 20 at the Fukagawa Meteorological Observatory of Kumamoto Prefecture, about 2 km west-southwest of the Astumari district (Kumamoto Prefecture, 2003). The debris-flow disaster occurred around 4:20 a.m. on July 20, 2003. It rained intermittently with an intense fall of 87 mm from 3:00 a.m. to 4:00 a.m. and 91 mm from 4:00 a.m. to 5:00 a.m. on July 20, reaching a cumulative rainfall of about 270 mm from the beginning of rain to the occurrence of the disaster. The 1-h rainfall of 91 mm from 4:00 a.m. to 5:00 a.m., the 2-h rainfall of 178 mm from 3:00 a.m. to 5:00 a.m., and the 3-h rainfall of 220 mm from 2:00 a.m. to 5:00 a.m. were

the maximum recorded at the Fukagawa Meteorological Observatory since the beginning of the rainfall observations there (Kumamoto Prefecture, 2004).

87mm91mm

2003.7.19 2003.7.200 6 12 18 0 6 12

0

20

40

60

80

100

0

100

200

300

400

500

Hou

rly

rain

fall

(mm

)

Fukagawa Meteorological Observatory

Occ

urre

nce

of d

isas

ter

Cum

ulat

ive

rain

fall

(mm

)

Fig.3 Temporal variation in hourly rainfall and cumulative rainfall from July 19 to 20 at the Fukagawa Meteorological Observatory 4. DEEP-SEATED LANDSLIDE AND

DEBRIS FLOW

The large scale landslide that occurred in the middle reaches of the Atsumari River was deep seated on the hillslope underlain with deeply weathered volcanic rocks (Figure 4).

Photo by Asia Air Survey Co., Ltd. Fig.4 Deep-seated landslide in the Atsumari River Catchment

Figure 5 shows the topographical and geological maps of the deep-seated landslide sites based on the field and boring surveys (Kumamoto Prefecture, 2004). The scale of the landslide was 70 to 100 m in width, 170 m in length, and 15 m in the maximum depth. The slope on which the landslide occurred had an average angle of about 35 degrees (Figure 5(b)). The landform, before the landslide was straight in longitudinal profile and slightly concave in lateral profile. Geologically, the slope is underlain by deeply weathered andesite and underlying tuff breccia. The slip occurred near the boundary between the andesite and the tuff breccia acting as a permeable layer and an aquiclude, respectively. The tuff breccia forms a concave surface that tends to gather groundwater from a

International Journal of Erosion Control Engineering, Vol. 1, 2008

7

wide underground area around the landslide site. The lithology and weathering of the andesite exposed on the sliding surface is markedly different in the southern and northern parts of landslide scar. The southern part of landslide scar is underlain by the relatively fresh andesite with prismatic structure, while the northern is underlain by highly weathered andesite (Figure 6).

On the basis of aerial photogrammetry and field survey, it was found that a landmass of about 42,700 m3 slid, out of which, about 12,200 m3 remained in landslide scar. Furthermore, a landmass of about 30,500 m3 slid into a ravine of the Atsumari River (Kumamoto Prefecture, 2004).

It seems that the process of change of debris flow from the landslide was a continuous phenomenon, because the trace of the deposits of the landslide dam in the ravine under landslide scar was not found. The debris flow rushed downstream running aground alternately on the left and right banks of the ravine, and repeating erosion and deposition. The velocity and peak discharge of the flow were estimated to be 11.7 m/s and 769 m3/s respectively on the basis of the trace of the debris flow in the check dam at an altitude of 180 m (Kumamoto Prefecture, 2004). This debris flow was characterized by little boulders rushing downstream first, and bigger boulders rushing immediately after that, as per the information obtained from the inhabitants and the scars left behind by the debris flow in the lower reaches of the Atsumari River. It is presumed that the human damage in the lower reaches was caused by the first flow (Kumamoto Prefecture, 2004).

The sediment yield transported by the debris flow at the downstream was 89,600 m3, out of which 76,500 m3 deposited on the downstream, and 13,100 m3 flowed into the Hougawachi River (Kumamoto Prefecture, 2004).

440

430

420

410

400

390

380m

A

A’

B

B’

(a)

Atsumari R.

Boring m0 50 100

0 50 100 150350

400

450

Surface level before landslide

Alti

tude

(m)

Distance (m)

A

A’

(b)

Colluvium (ls)

Talus deposit (dt)

Highly weathered Andesite (An1(w1))

Highly weathered Andesite (An1(w2))

Relatively fresh Andesite (An1)

Highly weathered tuff breccia (Tf(w1))

Tuff breccia (Tf)

Fresh Andesite (An2)

Slip plane

0 20 40 60 80 100370

380

390

400

410

420

Alt

itud

e (m

)

Distance (m)

Surface level before landslide

B

B’

(c)

Fig.5 Topographical and geological maps of deep-seated landslide sites (Kumamoto Prefecture, 2004). Plan of the landslide (a), longitudinal profile (b), and lateral profile (c) with geological information

Southern part

Northern part

Fig.6 Geologic section photos in deep-seated landslide scar

International Journal of Erosion Control Engineering, Vol. 1, 2008

8

A similar debris-flow disaster induced by a deep-seated landslide occurred at the Harihara River in Izumi City, Kagoshima Prefecture, about 10 km southeast of the Astumari River on July 10, 1997, and killed 21 people. However, some differences were observed between the deep-seated landslides in the Atsumari River Catchment and the Harihara River Catchment. A remarkable difference is the rainfall condition at the time of occurrence of the deep-seated landslide. The deep-seated landslide occurred 4 h after it stopped raining in the Harihara River Catchment, whereas it occurred simultaneously with the maximum rainfall in the Atsumari River Catchment. The groundwater level was continuously observed at both deep-seated landslide sites (Jitousono et al., 2004; Kumamoto Prefecture, 2004). Figure 7 shows the comparison of the responses of the groundwater level to the storm observed at these sites. The rise of the groundwater level due to the storm in the Atsumari River Catchment is quicker than that of the Harihara River Catchment. It shows clearly that the groundwater level continued rising 4 h after it stopped raining, after which, the landslide occurred in the Harihara River Catchment (Jitousono et al., 2004). On the other hand, it is presumed that the groundwater level rose rapidly and the landslide occurred in the rainfall in the Atsumari River Catchment. This was due to the difference in the geological structures in the deep-seated landslide sites. That is, it is presumed that the stratum has many cracks and the groundwater flows quickly around the deep-seated landslide site in the Atsumari River Catchment.

Hourly rainfall (mm)

Ris

ing

rate

of g

roun

dwat

er le

vel (

m/h

)

Atsumari RiverHarihara River

0 20 40 60 80 1000

1

2

3

4

5

6

Fig.7 Comparison of responses of groundwater level to storm 5. PREDICTION OF DEBRIS-FLOW

DISASTER INDUCED BY

DEEP-SEATED LANDSLIDE

In order to prevent the debris-flow disaster induced by the deep-seated landslides, it is important to predict the potential site of deep-seated landslide. Potential sites of deep-seated landslide on the western flank of the Mt. Yahazu-dake in Izumi City, Kagoshima Prefecture are predicted (Jitousono et al., 2006). A deep-seated landslide occurred in the Harihara River Catchment on the western flank of Mt. Yahazu-dake on July 10, 1997. This landslide was caused by rising of groundwater level due to heavy rainfall and deeply weathered volcanic rocks (Jitousono et al., 2004). Some indicators for predicting the potential sites of deep-seated landslide were examined based on topographical and geological surveys, along with hydrological observations in the catchments on the western flank of Mt. Yahazu-dake.

Figure 8 is the spatial distribution map of the deep-seated landslide scars on the western flank of Mt. Yahazu-dake with lineaments on the basis of field surveys and the interpretation of aerial photographs. In addition, the distribution map of gentle slopes based on topography measurements was added to Figure 8. Most of the deep-seated landslide scars are located on the hillslope with aslope of 20 to 30 degrees under a large ridge with gentle slopes of altitude 150–200 m and 250–350 m. This zone is underlain with deeply weathered volcanic rocks, and the boundary between the andesite and the tuff breccia acts as a permeable layer and aquiclude. Moreover, the deep-seated landslide scars are located along the lineaments.

Figure 9 shows the relationships between the altitude measured stream water and specific discharge, electric conductivity, and concentration of silica in the Era River. Many springs were observed near the points which the discharge, electric conductivity, and silica concentrations along a longitudinal stream increased rapidly.

Figure 10 shows a map predicting the potential sites of deep-seated landslides on the basis of topographical and geological indicators, such as a hillslope with a slope angle of 20 to 30 degrees under a large ridge, deeply weathered volcanic rocks, boundaries between aquifer and aquiclude, lineaments, hydrological indicators such as the change points of stream discharge, electric conductivity, and silica concentrations along a longitudinal stream, and the distribution of springs. These results agree with the deep-seated landslide scars on the basis of the field surveys and the interpretation of aerial photographs.

International Journal of Erosion Control Engineering, Vol. 1, 2008

9

N

m0 500 1000

Rogi R.

Harihara R.

Mt. Yahazu-dake

Era R.

Tributaryof Era R.

Gentle slope

Deep-seated landslide scarLineament

100m200 300 400

500

600

687m

Kumamoto

Kagoshima

Atsumari

MiyazakiMt. Yahazu-dake

Fig.8 Spatial distribution map of deep-seated landslide scars with gentle slopes and lineaments in the western flank of Mt. Yahazu-dake (Jitousono et al., 2006)

Altitude m

Cat

chm

ent a

rea

km2

1

2

3

4

Spec

ific

disc

harg

e

m3/(s 100km2)･

0

0.1

0.2

Ele

ctri

c co

nduc

tivity

mS/m

6

8

10

Con

cent

ratio

n of

sili

ca mg/L

50 100 150 200

10

20

Fig.9 Relationships between altitude measured stream water and specific discharge, electric conductivity, and concentration of silica in the Era River (Jitousono et al., 2006)

N

m0 500 1000

Deep-seated landslide scar

Potential sites ofdeep-seated landslide

Topographical and geological indicators

Hydrological indicators

Rogi R.

Harihara R.

Mt. Yahazu-dake

Era R.

Tributaryof Era R.

100m

200 300 400

500

600

687m

Fig.10 Spatial distribution map of the potential sites of deep-seated landslides in the study area (Jitousono et al., 2006) 6. CONCLUSION

A large scale landslide occurred in the Atsumari River Catchment, Minamata City, Kumamoto Prefecture on July 20, 2003. This type of landslide is called a “deep-seated landslide.” The debris flow induced by this landslide killed 15 people.

The characteristics of the deep-seated landslides and the debris flow induced by it were examined based on a field study in the Atsumari River Catchment. In addition, this study described a method for predicting potential deep-seated landslide site in the volcanic area around this disaster site. The results are summarized as follows:

1) The deep-seated landslide in the Atsumari River Catchment was caused by rising of groundwater level associated with heavy rainfall, the hydrogeomorphological formation of the underground area prone to storage of groundwater, and deeply weathered volcanic rocks.

2) The process of change of the debris flow due to the landslide was a continuous phenomenon, because the trace of deposits in the landslide dam in the ravine under the landslide scar was not found.

3) It was ascertained that effective indicators for predicting potential sites of deep-seated landslides were the partial distributions of gentle slopes, boundaries between aquifer and aquiclude, the change points of stream discharge, electric conductivity, silica concentrations along a longitudinal stream, and the distribution of springs. REFERENCES Jitousono, T., Shimokawa, E. Sako, M. and Teramoto, Y.

(2004): Hydrogeomorphological characteristics of a deep-seated landslide in the Harihara River Basin, Izumi City, Kagoshima Prefecture, Japan, Journal of the Japan Society of Erosion Control Engineering, Vol.56, No.5, pp.15-26 (in Japanese with English abstract).

International Journal of Erosion Control Engineering, Vol. 1, 2008

10

Jitousono, T., Shimokawa, E. and Teramoto, Y. (2006): Potential site prediction of deep‐seated landslide on the western flank of Mt. Yahazu‐dake, Izumi City, Kagoshima Prefecture, Japan, Journal of the Japan Society of Erosion Control Engineering, Vol.59, No.2, pp.5-12 (in Japanese with English abstract).

Kumamoto Prefecture (2003): Rainfall data (in Japanese). Kumamoto Prefecture (2004): The report about the debris-flow

disaster in Minamata City on July 20, 2003, pp.1-84 (in Japanese).

Mizuno, H., Sugiura, N., Terada, H., Uchida, T., Haramaki, T., Sokabe, M., Sakurai, W., Nishimoto, H., Osanai, N., Takezawa, N. and Doi, Y. (2003): The debris flow disasters caused by localized rainfall of seasonal rain front in Kyusyu region in July, 2003 (prompt report), Journal of the Japan Society of Erosion Control Engineering, Vol.56, No.3, pp.36-43 (in Japanese with English abstract).

Shimokawa E. and Jitousono T.（1999）：A study of the change from a landslide to debris flow at Harihara, Izumi City,

Southern Kyushu, Journal of Natural Disaster Science, Vol.20, No.2, pp.75-81.

Shimokawa, E., Jitousono, T. and Teramoto, Y. (2004): Deep-seated landslides in the area of the volcanic rocks called the “Hisatsu volcanic province,” Proceedings of the Annual Meeting for the Japan Society of Erosion Control Engineering in 2004, pp.8-9 (in Japanese).

Takahashi, M. (1974): Study on the landslide at Nishiuchitate of Miyazaki pref., Journal of the Japan Society of Erosion Control Engineering, Vol.26, No.4, pp.24-31 (in Japanese).

Taniguchi, Y. (2003): Debris disaster caused by local heavy rain in Kyushu area on July 20th, 2003 (prompt report) - Minamata debris disaster, Journal of the Japan Society of Erosion Control Engineering, Vol.56, No.3, pp.31-35 (in Japanese with English abstract).

Yamamoto, T. (1960): Volcano-geological and petrological studies of Hisatsu Volcanic Area, Bulletin of the Kyushu Institute of Technology, pp.1-87 (in Japanese).