Debris flow disaster in the Harihara River, Izumi City, 1997

 

HAJIME NAKAGAWA, TAMOTSU TAKAHASHI and YOSHIFUMI SATOFUKA

 

Disaster Prevention Research Institute, Kyoto University

Gokasho, Uji City, Kyoto 611-0011, Japan

Telephone: +81-774-38-4119, Fax: +81-774-32-6039,

e-mail:nakagawa@sabom.dpri.kyoto-u.ac.jp

 

 

ABSTRACT

A severe debris flow disaster occurred on the Harihara River, Izumi City, Kagoshima Prefecture, Japan about midnight of June 10, 1997. Twenty-one persons were killed, and 18 houses destroyed by this flow which, due to heavy rainfall, caused a massive landslide. Residents were aware that an extraordinary phenomenon was occurring in the river but did not take refuge, resulting in much material and human damage. A numerical simulation model to explain the behavior and the depositional processes of the debris flow is developed. The actual phenomena such as a sediment deposition area and a thickness of deposits were fairly well explained by the calculation.

 

Keywords: debris flow, numerical simulation, sediment disaster, landslide, Harihara

 

INTRODUCTION

A severe debris flow occurred in the Harihara River basin, Izumi City, Kagoshima Prefecture, Japan about midnight of June 10, 1997. Twenty-one persons were killed, and 18 houses destroyed by this flow which, due to heavy rainfall of 356 mm on the day before, recorded at the Izumi sewage purification center, caused a massive landslide of a sediment volume of 160,000 m Minami et al., 1997).

The slope on which the landslide occurred originally had an average angle of 26 degrees, which is not very steep, and the catchment area of the slope apparently was not large enough to hold a large volume of water. Before the disaster, on May 13, 1997, an earthquake (M6.2) had occurred in the northwestern part of Kagoshima Prefecture. Many surface landslides occurred on mountain slopes near the Harihara district, and slopes in the Harihara district also must have been affected by the strong seismic forces. Water channels within the landslide may have been cut by the earlier seismic activity, causing ground water to store in the slope. This would have caused the water table to rise, causing the landslide. If this was the origin of the landslide, similar landslides would have occurred on several nearby slopes.

To determine why such a large landslide occurred only on this slope, requires investigations by people in many areas of research. If the numbers of such sediment disasters are to be decreased, we need to know the times of occurrence, places, and scales of landslide under the recorded rainfall. Unfortunately, so far research has been insufficient to determine these factors. It is important however, that residents of sediment hazard-prone areas be made aware of simulations made under various landslide conditions. A numerical simulation model that explains the behavior and depositional processes of a debris is presented and is applied to the debris flow that occurred in the Harihara River basin.

 

SUMMARY OF THE DEBRIS FLOW DISASTER

The Harihara River basin which covers an area of 1.55 km has two typical tributaries. The 2.3 km main channel originates at the altitude of 445 m and debouches into the Yatsushiro Sea. At the outlet of the channel's ravine, an altitude of about 50 m, a relatively large debris fan has developed (Fig.1).

 

 

Fig.1 Harihara River basin

 

The community of Harihara was located on this debris fan. A part of the fan and the mild slope areas had been intensely cultivated with orange trees.

A sabo dam was under construction in the main channel, and the main part had been completed just before the disaster. A deep-seated landslide occurred on the mountain slope at about 00:40 on July 10, 1997 (Fig.1). Half of the sediment volume of the landslide was estimated to be caught by the sabo dam and half discharged from the dam (Kagoshima Prefectural Government, 1997).

 

 

Photo 1 Harihara district just after the disaster (photo courtesy of Kokusai Aerial Photo Co. Ltd.)

 

Photo 1 shows the Harihara district just after the disaster. The slope angle of the surface on which the landslide occurred was about 26 degrees before the disaster. The landslide's dimensions were about 200 m long, 80 m wide, and 27 m maximum depth, and the landslide volume is about 165,000 m (Moriwaki et al., 1998). Geologically, the slope was constituted of weathered andesite with underlying tuff breccia. A large part of the landslide mass consisted of this weathered andesite. The sliding mass first surged down along the Harihara River then flowed into an irrigation pond with the maximum storage capacity of about 9,000 m (Shimokawa et al., 1998). At that time, the landslide mass not only pushed the water out of the pond but absorbed some of it. This water produced increased fluidity in part of the landslide mass, causing it to flow further down to the sabo dam as a debris flow. This debris flow overtopped the sabo dam, flooding and depositing sediment on the Harihara fan, which destroyed 18 houses and killed 21 persons. A large part of the mass moved straightforward along the failed slope and buried the irrigation pond in sediment, but a large amount of sediment was deposited in a high mound on the left bank just upstream of the sabo dam.

A house on the left side of the Harihara fan was destroyed (Photo 2). As there are no large stones or sediment in its remains, we judge that it was destroyed by flood water discharged from the sabo dam. In contrast, a house near the Harihara River was buried by a large amount of sediment transported by the debris flow (Photo 3).

 

 

Photo 2 Destroyed house located on the left side of the Harihara fan

 

 

Photo 3 A house near the river whose first floor is deeply buried by a large amount of sediment

 

These findings mean that water had to be stored at the sabo dam and that it was discharged by the sudden intrusion of the debris flow into the dam, but it is not clear whether the water came from the irrigation pond or was the rainfall runoff.

 

REPRODUCTION OF THE DEBRIS FLOW

 

BASIC EQUATIONS

Although the landslide mass may not have behaved as a debris flow from the onset of movement, we used a system of basic equations applicable to a debris flow, an immature debris flow, and a turbulent flow because the mechanism and formulation of the landslide's mass changing into the debris flow have yet to be clarified. The basic equations used to calculate the development and deposition of a debris flow and its flooding are the depth-averaged two-dimensional momentum and continuity equations (Nakagawa et al., 1997);

(1)

(2)

(3)

The continuity of the coarse particle fraction sustained in the flow by the action of particle encounters is

(4)

(5)

where and are the and components of the flow flux; and the and components of the mean velocity; is the flow depth; the erosion or deposition thickness measured from the original bed elevation; and are the and components of the inclination of the original bed surface; is the specific density of the debris flow; the momentum correction coefficient; and are the and components of the resistance to flow; is the erosion () or deposition (< 0) velocity; the total solid fraction in the bed; the degree of saturation in the bed (applicable only for erosion, when deposition takes place substitute ); the volume concentration of the coarse fraction in the flow; the volume concentration of the fine fraction in the interstitial fluid; and are the volume concentrations of the coarse and fine fractions in the original bed; and is the volume concentration of the coarse fraction in the static bed produced by deposition of the debris flow.

An immature debris flow occurs when is less than. Equations (10) and (11), which are nothing but the Manning formula, apply when is less than 2% or is larger than 30. The momentum correction coefficient, , is equal to 1.25 for a stony type debris flow and to 1.0 for both an immature debris flow and a turbulent flow.

The bottom resistance for a two-dimensional flow is described as follows: For a fully developed stony debris flow;

(6)

(7)

For an immature debris flow;

(8)

(9)

For a turbulent flow;

(10)

(11)

where is the mean diameter of the coarse fraction; the density of the coarse particle; the density of the interstitial muddy fluid; the resistance coefficient; the numerical coefficient (0.04); the dynamic internal angle of friction (); and the energy gradient given by .

We introduce the following deposition equations: For a fully developed debris flow;

(12)

and for an immature debris flow and a turbulent flow;

(13)

(14)

For a debris flow the equilibrium solid concentrations are,

(15)

(16)

and for an immature debris flow and for a bed load, the respective equilibrium solid concentrations are

(17)

(18)

where is the density of clear water; the equilibrium bed load transport rate. In this study the equation used is

(19)

where is the resistance coefficient; and are the nondimensional tractive and nondimensional critical tractive forces; and is a constant. Equation(17) is applicable only when has a value less than , as calculated by eq.(15).

, , and are constants; and the equilibrium velocity that continues the run down with neither deposition nor erosion as given by

(20)

where is the critical slope given by

(21)

The equation for the bed surface elevation is

(22)

 

CONDITIONS FOR CALCULATIONS

As the initial conditions for the calculation of landslide behavior, we assume that the landslide mass is a continuous fluid with a coarse sediment fraction of 50%, that movement of the mass is restricted until the start of calculation, and that there is simultaneous removal of that restriction with the start of calculation. The landslide's behavior and its flooding and depositing process were computed using a two-dimensional numerical simulation and a staggered first order up-wind finite difference

scheme. The grid sizes are and m, and the time interval sec. The numerical constants and the other parameters adopted in the calculation were , , , , cm, , kg/m, kg/m, and . These sediment properties were estimated from field survey findings. The calculated case is that 9,000 m of water was stored in an irrigation pond and 200 m of water in the Harihara sabo dam.

 

CALCULATED RESULTS AND DISCUSSIONS

The calculated amounts of sediment deposited and the area flooded by water 1, 2, and 3 minute(s) after the landslide, are shown in Fig.2.

 

 

Fig.2 Calculated sediment deposition and water flooded area

 

The dark gray area shows the sediment deposited and the light gray one the water flooded area. The black spots are houses destroyed by the debris flow. Several houses are shown to have been destroyed by a debris flow only one minute after the landslide occurred, and the sediment deposit area has not changed much two minutes after the landslide.

 

 

Fig.3 Calculated maximum areas of sediment deposition and water flooding

 

Figure 3 shows the calculated maximum areas of sediment deposition and water flooding 20 minutes after the landslide (i.e. the final stage of the debris flow). The recorded sediment deposition and water-flooded areas also are shown in the figure. The calculated sediment deposition area corresponds comparatively well with the area in which houses were destroyed and is mainly along the Harihara River, whereas the actual area spreads straightforwardly on the fan. The calculated height of the sediment deposited on the left bank just upstream of the sabo dam is about 82 m, which reasonably well expresses the actual situation.

 

 

Fig.4 Calculated thicknesses of accumulated sediment

 

Figure 4 shows the calculated sediment accumulation. The depth of the deposit just upstream of the sabo dam is more than 5 m, and about 2-5 m along the Harihara River downstream of the dam, whereas on the left side of the fan it is less than 1 m. These values well express the sediment deposition phenomena shown in Photos.3 and 2.

 

 

Table 1 Comparison of the calculated and observed sediment volumes of the deposits

 

Table 1 shows a comparison between the calculated and observed (Minami et al.,1997) sediment volumes of the deposits. The observed sediment volumes deposited downstream and upstream of the sabo dam are both about 80,000 m. The calculated results show similar values.

 

CONCLUSIONS

How the debris flow was generated and developed from the landslide mass is not clear, but the depositional process of the flow is considered to be fairly well expressed by the numerical simulation model presented here. The simulation model, however, is not precise enough because it is not clear how the bottom shear stresses of the landslide mass changes when the mass becomes a debris flow. Moreover, there are limitations when expressing the mixing process of the debris flow with water using the depth-averaged two-dimensional model. These are urgent problems that must be solved.

 

ACKNOWLEDGEMENTS

This work was financially supported by Grant-in-Aid for Scientific Research No.09600003 (Prof. Etsuro SHIMOKAWA, Kagoshima University) from the Japanese Ministry of Education, Science, Culture and Sports. We thank Dr. Yasuto TACHIKAWA, Associate Professor, DPRI, Kyoto University and Mr. Yasuhiro, SATO, Undergraduate Student, Kyoto University for their help in the flood runoff analysis, Dr. Motoyuki USHIYAMA for the rainfall data, and the many other persons who kindly made useful suggestions. We also are grateful to Kokusai Aerial Photo Co. Ltd. for providing the aerial photos.

 

REFERENCES

Kagoshima Prefectural Government (1997) Materials for discussion: the 1st committee on the debris flow in the Harihara River (in Japanese)

Minami, N., Yamada, T. and Mizuno, H. (1997) Sediment volume and the deposition area of the debris flow in the Harihara River July 10, 1997, Izumi City, Kagoshima Prefecture, Japan, Jour. Japan Society of Erosion Control Engineering, Vol. 50, No.3, 81-82 (in Japanese).

Moriwaki, H., Sato, T. and Chiba, M. (1998) Report on the Harihara River debris flow disaster of July 10, 1997 in Kagoshima Prefecture, Japan, Natural Disaster Research Report No.35, National Research Institute for Earth Science and Disaster Prevention, Science and Technology Agency, Japan, 1-69 (in Japanese).

Nakagawa, H. and Takahashi, T. (1997) Estimation of a debris flow hydrograph and hazard area, Proc. of the 1st Intrn. Conf. on Debris-Flow Hazards Mitigation, 64-73.

Shimokawa, E., Jitousono, T. and Ogawa, S. (1998) Debris flow disaster in the Harihara River, Izumi City, Research Report on Natural Disasters, Supported by the Japanese Ministry of Education, Science, Culture and Sports (Grant No.09600003), 19-30 (in Japanese).