# IJEGE-11_BS-Nishiguchi-et-alii

*Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza*

*DOI: 10.4408/IJEGE.2011-03.B-053*

**PREDICTION OF RUN-OUT PROCESS FOR A DEBRIS FLOW TRIGGERED**

**BY A DEEP RAPID LANDSLIDE**

**INTRODUCTION**

In particular, deep rapid landsides have triggered large-

scale debris flows that have had serious impacts on hu-

mans. Therefore, it is important to predict the run-out

process of debris flows and to identify debris flow haz-

ard areas. A number of numerical simulation models

have been developed to describe the propagation and

deposition of debris flows (e.g., e

*et alii*, 1989;

*et alii*, 1993; t

*et alii*, 2002;

*et alii*, 2006). However, useful techniques

flows have not yet been developed.

(t

tial water are assumed to be stationary (Fig. 1a). When

sediments move in a debris flow, their motion under

these models is considered to be laminar, while that of

the interstitial water is turbulent (H

*et alii*, 1998)

sediments exhibiting laminar flow and a “fluid phase”

of interstitial water exhibiting turbulent flow.

flows have a broad grain size distribution; numerical

simulations based on these simplified models may be

**ABSTRACT**

not be applicable for large-scale debris flows. In this

study, we developed a technique for simulation of

large-scale stony debris flows. For this purpose, we

tested the hypothesis that the motion of fine sediment

in these debris flows is similar to that of the intersti-

tial water. We developed key parameters to simulate

large-scale debris flows, such as sediment concentra-

tion, fluid density, and representative particle diam-

eter based on the hypothesis. We also used a modified

version of the continuity equation for sediment in our

simulation. We conducted detailed field surveys of a

past debris flow in Japan and used topographic data

from LiDAR imagery, porosity measurements of soil

and weathered bedrock, and the grain size distribu-

tion of the debris flow sediments to test our model.

We also proposed a new process-based method for

determination of hydrographs at the lower end of the

landslide scar. Using these new data and methods, we

conducted numerical simulations of the past debris

flow, which reproduced well the observed erosional

and depositional pattern if when the concept of fine

sediment behaving like fluids was included in the nu-

merical simulation.

**K**

**ey**

**words**

**:**debris flow, deep rapid landside, numerical si-*mulation*

*Y. NISHIGUCHI, T. UCHIDA, k. TAMURA, & Y. SATOFUkA*

*5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011*

bris flows is similar to that of interstitial water, and to

develop a valid technique for the simulation of large-

scale stony debris flows. To test our hypothesis, we

conducted detailed field surveys of debris flows by

using topographic measurements from LiDAR data,

measuring the porosity of soil and weathered bedrock

and measuring the grain size distributions of debris

flow sediments. Further, we proposed a new process-

based method for the determination of debris flow hy-

drographs at the lower end of the landslide scar. We

also considered the effects of the uncertainty of sev-

eral parameters on the simulated result.

**THEORY**

sediment and a fine sediment, and that the motion of

the fine sediment in a debris flow is similar to that of

the interstitial water. We therefore considered these

fine sediments as a fluid phase. Further, we defined

a maximum diameter

*D*

*c*

this definition of

*D*

*c*

concentration at the lower end of the landslide scar

(Cd, Eq. 1), fluid density of the debris flow averaged

in time and space (

*r*, Eq. 2), and the representative

particle diameter of the debris flow (D, Eq. 3).

landslide at Mt. Ontake in Nagano in 1984 could be

represented by a fluid phase, whereas the motion of

the coarse sediment was that of a solid phase. Thus,

for this large-scale debris flow, both the fine sediments

and the interstitial water exhibited turbulent flow, and

only the coarse sediment exhibited laminar flow (Fig.

1c). Further, n

*et alii*, (1998) and e

*et*

*alii*, (1998) performed numerical simulations of past

large-scale debris flows with particular attention to the

motion of fine sediments. The results of their simula-

tions matched the observed deposited area and thick-

ness of the debris-flow deposits if it was assumed that

increases of fluid density were dependent on the pro-

portion of fine sediments considered as a fluid phase,

or when coarse and fine sediments were considered

independently to account for their different behaviour

and the different continuity equations between coarse

and fine sediments were presented.

deposition of debris flows, most parameters have been

determined by back-calculation or by calibration to

match field observations; in many cases, mean values

have been used. Consequently, the hypothesis that the

motion of fine sediments in debris flows is similar to

that of interstitial water has not been fully examined.

*Fig. 1 - Conceptual diagram of static condition, small-*

*scale debris flow and large-scale debris flow*

*Fig. 2 - Definition of Dc*

**PREDICTION OF RUN-OUT PROCESS FOR A DEBRIS FLOW TRIGGERED BY A DEEP RAPID LANDSLIDE**

*Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza*

muddy deposits and stony deposits (Fig. 4). Local peo-

ple who were interviewed said that the muddy debris

flow occurred first and was followed by the stony de-

bris flow 10–20 min later. The distance from the lower

end of the landslide scar to the lower end of the debris

flow deposits was about 1,600 m.

ment was around 42,700 m

and the volume of collapsed sediment that remained

in the landslide scar was 12,220 m

m

be 59,100 m

mated to be 76,500 m

Although the relative volumes of muddy debris flow

and stony debris flow were not measured, aerial pho-

tographs and field observations indicate that more than

half of the sediment volume was stony debris flow.

**FIELD INVESTIGATION**

*TOPOGRAPHY*

run-out area was estimated by calculating the differ-

*P(D*

*c*

*)*is the ratio of sediment smaller than

*D*

*c*

*ρ*

*w*

*ρ*

*s*

*d (D*

*c*

*)*is the weighted aver-

*D*

*c*

*w*is the water

*w*and is water

content of the debris flow averaged in time and space.

*t*is time,

*i*is erosion and deposition rates,

*u*and

*v*are velocities in the

*x*and

*y*direction, respectively,

*h*

is flow depth of debris flow, and

*C*

***

**MATERIALS AND METHODS**

*STUDY SITE*

20 July 2003 induced a deep rapid landslide and the

resultant mass of collapsed sediments caused a large-

scale debris flow that hit the village along Atsumari

River, killing at least 15 people. The study area is un-

derlain by andesite and weathered tuff breccia. The

slip surface of the landslide was the interface between

*Fig. 3 Location of our study site*

*Fig. 4 - Aerial photograph after the debris flow*

*Y. NISHIGUCHI, T. UCHIDA, k. TAMURA, & Y. SATOFUkA*

*5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011*

(i.e., the interface between weathered andesite and

weathered tuff breccia) was around 0.34.

*PARTICLE SIZE DISTRIBUTION OF THE DE-*

BRIS FLOw

BRIS FLOw

tional photograph of the deposits. We also identified

the proportions of sediments smaller than 7.5 cm and

larger than 30 cm by using the photograph. We com-

bined the sieve test result of sediments smaller than

7.5 cm and the observed result of particle size distri-

bution of sediments larger than 30 cm with the particle

size distribution of sediment between 7.5 cm and 30

cm at the rate of their respective proportions (Fig. 8).

**NUMERICAL SIMULATION**

*MODEL*

debris flows in alluvial fans (n

*et alii*, 2008).

ments ranging from bed load to stony debris flow. The

equations from Eq.6 to Eq.11 are governing equations

in the Kanako2D simulation. Eq.6 is continuation

digital elevation model (DEM), both generated from

LiDAR data (Fig. 5). In the depositional area, it was

evaluated from aerial photography.

after the debris flow were obtained from aerial pho-

tography and LiDAR data, respectively. The magni-

tude of the elevation change was calculated as the

difference of these (Fig. 6).

*POROSITY OF SOIL AND BEDROCk*

the landslide scar. We measured bulk density and

soil water content using a gamma-radiation density

gauge and a nuclear radiation moisture gauge, re-

spectively (Fig. 7). Mean porosity of soil and bed-

*Fig. 5 - Method for calculation of flow width from DEM*

*and DSM*

*Fig. 6 - Longitudinal profile and width of debris flow*

*Fig. 7 - Depth profiles of porosity from two boreholes near*

*the landslide scar*

*Fig. 8 - Particle size distribution of debris flow*

**PREDICTION OF RUN-OUT PROCESS FOR A DEBRIS FLOW TRIGGERED BY A DEEP RAPID LANDSLIDE**

*Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza*

*C*

*∞*

*is equilibrium sediment concentration in the*

*δ*

*e*

*δ*

*d*

*q*is discharge of debris flow per

unit width.

*DATA PREPARATION*

flow deposits. How we determined the main simula-

tion parameters is provided in Table 1.

survey data (Fig. 6). Our field survey data also showed

that the maximum erosion depth was 5 m, indicating

that the initial depth of the movable bed layer (

*D*

*s*

eroded area, but movable sediment remained in some

areas after the debris flow, suggesting that the mean

initial depth of the movable bed layer might have been

more than 5 m. Consequently, we used two values (5

*i*is erosion/deposition rate. Then, the continu-

ation equation for determining the debris flow is as

follow.

tion flow.

*g*is gravitational acceleration,

*θ*

*wx*

*θ*

*wy*

*τ*

*x*

*τ*

*y*

*τ*

*x*

*et alii,*2001).

*n*

*m*

*is Manning's coefficient. The river bed*

u is replaced by v in Eq.9. Equation for determining

change in bed surface elevation is as follow

*i*is described as Eq.11. (n

*et alii*, 2001).

*Tab. 1 - Determination of main simulation parameters*

*Y. NISHIGUCHI, T. UCHIDA, k. TAMURA, & Y. SATOFUkA*

*5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011*

ticle diameter of the debris flow were determined from

Eqs. 1, 2, and 3, respectively. We assumed that the soil

and weathered bedrock of the landslide material were

water saturated. Our field investigations at boreholes

1 and 2 showed that porosity of landslide soil and bed-

rock was 0.34, so we used 0.34 for the water content

of soil and weathered bedrock. We determined

*C*

*d*

end of the landslide scar) from Eq. 1 with

*w*= 0.34.

*D*

*c*

*ρ*were constant in both space and time.

*ρ*in the numerical simu-

lation represents the fluid density of the debris flow

in the run-out area. According to our field observa-

tions, the upper part of the river was eroded by the

debris flow, so we assumed that the debris flow was

composed of both collapsed sediments and river bed

sediments. Therefore, we decided to use an average of

the water content of the soil and weathered bedrock

that yielded the landslide and that of the river bed ma-

terials for

*w*in Eq. 2. Although we did not measure the

water content of river bed material, previous studies

reported it to be around 0.4, so we used the average of

0.34 and 0.4 (i.e., 0.37) for

*w*in Eq. 2.

density, and the representative diameter of debris flow

particles. To ascertain the effect of different values for

the maximum diameter of sediment that behaves like

a fluid (

*Dc*), we considered five cases (

*D*

*c*

flow at Atsumari river showed that a large proportion

of fine material were included. Although these fine

materials might have colloidal properties, in this pa-

per, we did not consider these effects.

volume of stony flow to total flow (

*k*). We used aerial

photography and field data to determine that

*k*was be-

tween 0.5 and 1.0. Therefore, we used values of 0.5,

0.7, and 0.9 for

*k*in our simulations.

mated as follows. Movement of a landslide mass can

be expressed as

*L*is longitudinal length of the landslide scar

(120 m),

*v*

*m*

*t*

*s*

*h*

*m*

*V*

*s*

*B*

*m*

end of the landslide scar could be described by the

resistance law developed by t

*I*

*m*

*a*

*i*

*α*are constants (0.042 and 17.8°,

*C*

*d**

*is the maxi-*

*ρ*is fluid density at the lower end of landslide scar,

which can be described as

*D*

*c*

Atsumari River, maximum 10-minute rainfall of 26

mm has been recorded before the debris flows. The

water discharge attributed to the rainfall was esti-

mated at 30 m

charge of this volume is negligible in comparison to

the discharges presented by the hydrographs of Fig. 9.

Consequently, we didn’t include water discharge at-

tributed to rainfall in the upper drainage area with the

hydrographs of Fig. 9.

**PREDICTION OF RUN-OUT PROCESS FOR A DEBRIS FLOW TRIGGERED BY A DEEP RAPID LANDSLIDE**

*D*

*s*

*D*

*s*

when

*D*

*s*

*D*

*s*

*D*

*c*

*D*

*c*

*D*

*s*

*k*and

*D*

*s*

*δ*

*e*

*δ*

*d*

*et alii*, 2008; n

**RESULTS**

*D*

*c*

slide scar to the lower end of the debris flow deposits

was about 600 m, which is about 40% of the observed

travel distance. For this case, there was no erosion by

the debris flow and deposition started immediately

downstream of the landslide scar. As

*D*

*c*

(Fig. 10). When

*D*

*c*

the lower end of the landslide scar to the lower end

of the eroded area (hereafter referred to as “erosion

distance”) also increased with increasing

*D*

*c*

*D*

*c*

depths of erosion and deposition agreed well with our

observations (Fig. 10).

total volume (

*k*) and initial depth of the movable bed

layer (

*D*

*s*

*D*

*c*

*k*is 0.7, the calculated volume and thickness

of the deposit well agreed with the observed volume

and thickness of the deposit. However, for

*k*in the

range from 0.5 to 0.9, the differences of the simulated

travel and erosion distances were unremarkable (Fig.

11) which indicates that total volume of sediment at

the upper end of the simulation did not influence the

erosion and deposition distances of the debris flow, al-

though the erosion and deposition distance were influ-

enced by

*D*

*c*

end of the simulation, this uncertainty had little im-

pact on the results of erosion and deposition distances

*Fig. 9 - Hydrographs for different values of Dc*

*Fig. 10 - Observed and calculated results for different val-*

*ues of Dc*

*Fig. 11 - Observed and simulated results for different val-*

*ues of k and Ds*

*Y. NISHIGUCHI, T. UCHIDA, k. TAMURA, & Y. SATOFUkA*

**CONCLUDING REMARKS**

pothesis that the fine sediments in such debris flows

behave as fluids. Focusing on the maximum diam-

eter of sediment particles that behave as fluids (

*D*

*c*

simulation of the debris flow; these are sediment con-

centration, fluid density, and representative particle

diameter. We also used a modified version of the gen-

eral continuity equation for sediments. We conducted

detailed field surveys and used topographic data from

LiDAR imagery, porosity measurements of soil and

weathered bedrock, and the grain size distribution of

the debris flow sediments to test our model. Further,

we proposed a new process-based method for deter-

mination of hydrographs of a debris flow at the lower

end of the landslide scar. We also considered the effect

of the uncertainty of several parameters on the simula-

tion results. We then simulated the debris flow at the

Atsumari River on 20 July 2003. The conclusions of

this study can be summarized as follows:

- The simulated erosion and deposition distances of

when the concept of

*D*

*c*

was little agreement with the observed erosion and

deposition distances. Therefore, the hypothesis

that fine sediments in large-scale debris flows be-

have like fluids was confirmed for the July 2003

debris flow at the Atsumari River.

*D*

*c*

particle size below which particles behaved like

fluids for the 2003 debris flow at Atsumari River

by comparison of the settling velocity of the

*D*

*c*

and the turbulent velocity of the interstitial water.

**DISCUSSION**

*D*

*c*

*value of 20 mm, we compared the*

of interstitial water, and the settling velocity of sedi-

ment of 20 mm diameter.

*ω*

*0*

*v*is kinematic viscosity (0.01 cm

*D*

*c*

*ω*

*0*

*U**) can be calculated from

*h*

*f*

*I*

*f*

area. As a consequence,

*U**was 164 cm/s (Tab. 2).

*v*

*f*

*et alii*, 1998).

*ρ*

*f*

*ρ*

*f*

*u*

*f*

*h*

*f*

tively, when

*D*

*c*

*z*is bed elevation (here,

*α*can be expressed as

*k*

*f*

*C*

*df*

bris flow when

*D*

*c*

*ν*

*f*

mm in diameter, which verifies that a particle size of

20 mm can be suspended and in turbulent motion in a

debris flow. This means that

*D*

*c*

*Tab. 2 - Determination of settling, friction, and turbulent*

*velocities*

**PREDICTION OF RUN-OUT PROCESS FOR A DEBRIS FLOW TRIGGERED BY A DEEP RAPID LANDSLIDE**

**REFERENCES**

*Constitutive equation of debris flow.*Annuals of Disaster Prevention Research

**32**(B-2

**):**487-501, Japan.

*Numerical simulation of debris flow at the Gamaharazawa in the Hime river*

*basin.*Annual Journal of Hydraulic Engineering, JSCE,

**42:**919-924, Japan.

*Pore-water pressure distribution of solid-water phase flow in a rolling*

*mill.*Journal of the Japan Society of Erosion Control Engineering,

**50**(6): 11-16, Japan.

*Flow of variably fluidized granular masses across three-dimensional terrain:*

*1. Cou-*

*lomb mixture theory*. Journal of Geophysical Research,

**106**(B1): 537-552.

*Debris flow disas-*

*ters at Harihara River*. Izumi City, Kagoshima Prefecture,1997. Annuals of Disaster Prevention Research Institute, Kyoto

University,

**41**(B-2): 287-298, Japan.

*Sediment disaster caused by the heavy rainfall in the Camuri*

*Grande River Basin, Venezuela,*

*1999-Reproduction of sediment runoff, flooding, and evaluation of effectiveness of the sabo*

works by means of numerical simulation.Annuals of Disaster Prevention Research Institute, Kyoto University,

works by means of numerical simulation.

**44**(B-2)

**:**

207-228, Japan.

*Development and Application of debris flow numerical simulation system equipped with a graphical user*

*interface.*Ph.D Thesis, Kyoto University, Japan

*Development of “kanako 2D (Ver.2.00),” a user-friendly one- and*

*two-dimensional debris flow simulator equipped with a graphical user interface.*International Journal of Erosion Control

Engineering,

**1**(2): 62-72, Japan.

*Study on preventing debris flow disaster on heritage existing moun-*

*tainous area applying numerical simulation.*Disaster Mitigation of Cultural Heritage and Historic Cities,

**3:**135-142, Japan.

*Two- dimensional water flood and mudflow simulation.*J. Hydraulic Engi-

**119**(2)

**:**244-261.

*- Comparison of 2D debris-flow simulation models with field events.*

**10:**241-264.

*Settling velocities of gravel, sand and silt particles.*American Journal of Science,

**25:**325-338.

*Two-layer model for analysis of deposition and erosion processes of*

*debris flows.*Annual Journal of Hydraulic Engineering, JSCE, 46, 677-682, Japan.

*Mechanism of occurrence of mud–debris flows and their characteristics in motion.*Annual of Disaster

**20**(B-2)

**:**405-435, Japan.

*Mechanics and countermeasure of debris flow*. Kinmirai-sha, Japan.

*Formation of debris flow on varied slope bed.*Annuals of Disaster Prevention Research

**29**(B-2)

**:**345-359, Japan.

*Prediction of stony debris flow induced by severe rainfall.*Journal of the Japan Society of

**44**(3): 12-19, Japan.

*Dynamical law of debris flows in inertial regime.*Annual of Disaster Preven-

**39**(B-2)

**:**333-346, Japan