You are on page 1of 28

Article

Risk assessment of
simultaneous debris flows
in mountain townships

Progress in Physical Geography


37(4) 516542
The Author(s) 2013
Reprints and permission:
sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0309133313491445
ppg.sagepub.com

Peng Cui
CAS Key Lab. of Mountain Hazards and Earth Surface Process, Institute of Mountain Hazards and Environment,
Chinese Academy of Sciences, China

Qiang Zou
CAS Key Lab. of Mountain Hazards and Earth Surface Process, Institute of Mountain Hazards and Environment,
Chinese Academy of Sciences, China; University of Chinese Academy of Sciences, China; and Southwest
University of Science and Technology, China

Ling-zhi Xiang
Chongqing Jiaotong University, China

Chao Zeng
CAS Key Lab. of Mountain Hazards and Earth Surface Process, Institute of Mountain Hazards and Environment,
Chinese Academy of Sciences, China; and University of Chinese Academy of Sciences, China

Abstract
Many mountain towns in China are located on the joint alluvial fans of multiple and adjacent past debris flows,
making them vulnerable to large, multiple, and simultaneous debris flows during heavy rainfall. Without
emergency management planning, such flows, often appearing with interconnecting and chain-reaction
processes, can lead to extensive loss of life and property. In the Wenchuan earthquake-affected area, such
disasters are common. We analyzed the compound effects of simultaneous debris flow events, and proposed
three quantitative methods of debris risk assessment based on kinetic energy, flow depth, and inundation
depth. Validated using a field study of actual debris flow disasters, these analyses are useful in determining
the type, quantity, distribution, economic worth, and susceptibility of hazard-affected objects in a region. Subsequently, we established a method to determine the vulnerability of different hazard-affected objects, particularly concerning the susceptibility indexes of buildings or structures. By analyzing the elements underlying
hazard formation conditions, damage potential, and the socio-economic conditions of mountain townships, we
proposed a systematic and quantitative method for risk analysis of mountain townships. Finally, the proposed
method was applied to a case study of Qingping Township, which was affected by 21 simultaneous debris flows
triggered by a 50-year return period precipitation event. The proposed method analyzed the superposition and
chain-reaction effects of disasters and divided the affected area of the township into three risk zones. The analysis indicated that the calculated risk zones coincide with the actual distribution and severity of damage in the
debris flow event, which suggests that the risk assessment is consistent with results from the actual disaster.

Corresponding author:
Peng Cui, 9#, Section 4, Renminnanlu Road, Chengdu 610041, Sichuan, Peoples Republic of China.
Email: pengcui@imde.ac.cn

Cui et al.

517

Keywords
debris flow, hazard analysis, risk assessment, vulnerability, Wenchuan earthquake

I Introduction
On 12 May 2008, a devastating mega-earthquake
of magnitude 8.0 struck the Wenchuan area in
northwestern Sichuan Province, China. The
Wenchuan earthquake caused subsequent debris
flows to be more active and occur on a larger
scale than in previous years (Cui et al., 2010;
You et al., 2010). Large-scale and simultaneous
debris flows caused by heavy rainfall commonly occur in such earthquake-prone areas.
Some large mountain towns in western China
are located on the alluvial fans of several adjacent debris flow catchments, and their similar
geological and geomorphological conditions
can cause simultaneous debris flows to develop
in these gullies during periods of locally heavy
rainfall. Moreover, these simultaneous debris
flows often lead to a complex process of damage
accompanied by multiple hazards, including
direct dynamic impact destruction, debris accumulation, and subsequent damage induced
by lifeline destruction and chain-reaction disasters that occur due to river blockages. The
destruction of Qingping Town, Yingxiu Town,
and Longchi Town by simultaneous debris
flows in August 2010 represents typical examples of this phenomenon (Table 1). Therefore,
it is vital to develop an accurate risk evaluation
of the effects of multiple, simultaneous, or conterminous debris flows on mountain towns.
Debris flow risk assessment plays a crucial role
in disaster prevention and mitigation. In recent
years, several studies have examined hazard and
risk assessment of debris flows (Hurlimann
et al., 2006; Jakob and Hungr, 2005). Generally,
debris flow risk analysis focuses on two scales:
regional studies and on-site or local-scale
studies. Debris flow risk analysis on a regional
scale provides a risk awareness of potential
regional hazards, and may satisfy the needs of

macroscopic disaster mitigation. This scale of


analysis is predominantly used in combination
with methods such as geographic information
systems (GIS) (Huggel et al., 2003), statistical
analysis (Mark and Ellen, 1995; Rickenmann,
1999; Cannon et al., 2010), simple dynamic
approaches (Archetti and Lamberti, 2003;
Gret-Regamey and Straub, 2006), and interpretation of aerial photographs or satellite images
(Bisson et al., 2005; Pradhan, 2010). Detailed
on-site studies at a local scale instead focus on
analyzing the damage processes, destruction
range, and potential losses from debris flows.
Scientists have explored various methods and
models to develop accurate risk assessment for a
single debris flow. Liu and Mo (2003) proposed
a debris flow risk evaluation model based on 14
impact factors, establishing five risk levels.
They related debris flow hazard (H) to an appropriate number of variables, including magnitude, frequency, watershed area, the length of
the main channel, relative height of watershed,
incision density, and the percentage of the
unstable channel bed. The overall vulnerability
(V) was estimated by means of empirical functions dependent on population and the total
amount of property. This was the first major
study on debris flow risk assessment in China,
but the damage process and the resistance of
hazard-affected objects on different locations
of alluvial fans were not taken into account.
Calvo and Savi (2009) proposed a method for
formal risk analysis in debris flow prone areas.
In their method, a Monte Carlo procedure was
applied to quantify debris flow hazards, and
three different vulnerability functions were
introduced to analyze a case of damaged buildings. Based on an analysis of the hazards of debris flows and the individual vulnerability and
value of hazard-affected objects, Gentile et al.

518

Progress in Physical Geography 37(4)

Table 1. Examples of typical simultaneous debris flows in earthquake-affected areas.


Date

Location

1314 August Qingping,


2010
Mianzhu
County

1314 August Longchi,


2010
Dujiangyan
County

1314 August Yingxiu,


2010
Wenchuan
County

Description
Debris flows, triggered by locally intensive rainfall, simultaneously
appeared in 21 catchments around Qingping Township, causing 12
deaths. The debris (about 400 million m3) blocked the Mianyuan River
and silted an area 3.5 km in length and 400500 m wide. The debris
raised the riverbed over 5 m, to a maximum thickness of 13 m in some
areas. Furthermore, the debris flow destroyed 479 houses recently
reconstructed after being destroyed in the Wenchuan earthquake.
Debris flows occurred in 43 catchments around the town of Longchi in
Dujiangyan County, destroying or seriously silting the buildings along
the nearby watercourse. In total, the debris flows damaged 182
houses, destroyed more than 311 ha of farmland and 3.35 km of
roads, and killed 2682 poultry.
Following heavy rain, a debris flow occurred in Hongchun Gully along
the upper reaches of the Min River, blocking water flow in the river.
Eventually, the backwater of the barrier lake submerged the Yingxiu
hydropower station, the Shaohuoping Bridge (a suspension bridge
upstream of the debris barrier), and national road G213.
Furthermore, the debris forced the main course of the Min River to
shift to the right, flooding Yingxiu town. The township had just been
reconstructed on the first river terrace after the Wenchuan
earthquake, and the new residential buildings and water supply
facilities were seriously damaged.

(2008) evaluated four risk degrees of debris


flows and implemented a risk mapping model.
However, the aforementioned studies focused
on analyzing the risk caused by a single debris
flow. Current risk analysis methods for a single
debris flow normally do not consider the compound disaster effects of multiple debris flows.
As is well known, natural disasters involve various interacting components of natural systems;
ignoring such interconnections, and their
mutual effect(s) on the different types of risks
within a single area, can lead to underestimation
of the synergy produced by the joint functions of
several processes (Kappes et al., 2010; Marzocchi et al., 2012; Perles-Roselloy et al., 2010).
Although hazard and vulnerability analysis
methods are well established for risk assessment
of single natural disasters, assessment of the
compound hazards of simultaneous disasters
poses a variety of challenges due to their widely
different characteristics (Kappes et al., 2010).

To date, some scientists have conducted preliminary research on multi-hazard risk assessment and mapping. Carrasco et al. (2003)
applied the Bayes conditioning probabilistic
method and GIS techniques to establish hazard
zoning in the Jerte Valley, Spain, which is subject to frequent landslides and floods. Considering six primary hazards, including shoreline
erosion, riverine flooding, storms, landslides,
seismicity and volcanism, and man-made structures, De Pippo et al. (2008) adopted a semiquantitative method to quantify, rank, and map
the distribution of hazards along the northern
Campanian coastal zone in Italy. PerlesRoselloy et al. (2010) proposed different methods for producing multi-hazard maps, including
a synthetic-hazard map, an aggregate-hazard
map, a stability map, and an accumulated risk
map. These studies improve knowledge of risk
assessment for multiple hazards, notwithstanding
that they focused on various hazards on an

Cui et al.

individual basis. However, each type of hazard displays unique damage characteristics due to idiographic, dynamic, and movement processes.
Consequently, it is difficult to consider the
compound and chain-reaction effects based on
dynamic processes of each hazard in a synthetic-risk assessment for multiple hazards.
Although some semi-quantitative hazard and risk
analysis methods are well established for several
natural hazards (Granger et al., 1999; Marzocchi
et al., 2009; De Pippo et al., 2008), few methods
are available for analyzing the simultaneous
occurrence of multiple debris flows. Therefore,
a method to quantitatively evaluate the risks of
simultaneous debris flows in mountain townships
is necessary in order to develop appropriate risk
management strategies.
This paper explores a new risk analysis
method for mountain towns subject to a group
of large-scale and simultaneously occurring
debris flows, by focusing on analysis of the
superposition effect and of the chain reaction
triggered by multiple debris flows. We propose
a set of hazard evaluation indicators based on
this method to describe the resulting damage.
Finally, we apply the proposed method to conduct a case study of Qingping Township, which
was significantly damaged by debris flows on
13 August 2010.

II Hazard characteristics of
simultaneous debris flows in
mountain townships
1 Simultaneous occurrence of debris flows
In China, many mountain towns are located on
the joint alluvial fans of several adjacent debris
flow catchments. According to investigation
and statistical analysis, more than 1000 such
towns are in Western China. Table 2 lists some
typical cities and towns in the region affected
by several adjacent debris flows. Dongchuan
Township resides just within the debris flow
hazardous area (Figure 1). Due to the similar

519

geological and geomorphological conditions


of these adjacent debris flow gullies, debris
flows triggered by regional rainfall can simultaneously occur in a group in these gullies and
result in interconnecting and superpositioned
disasters. This characteristic of debris flows is
particularly evident in areas that have conditions of steep terrain and abundant loose debris.
For instance, a field investigation shows that a
recorded individual rainfall of 96.3 mm and a
rainfall intensity of 77.3 mm in 40 minutes during the evening of 7 August 2010 triggered
simultaneous debris flows in Sanyanyu and
Luojiayu Ravines. The Sanyanyu and Luojiayu
debris flows had peak discharges of 1485 m3/s
and 390 m3/s, respectively, and the resultant
flows destroyed the urban area situated on the
alluvial fans, caused 1765 deaths and blocked
the Bailong River, which resulted in inundation
of the township (Table 2). This case is typical
for interconnecting and superposition disasters
of multiple debris flows. During the rainstorm
of 1314 August 2010, 43 debris flows broke
out in the area of Longchi Township of Dujiangyan County, and debris flows appeared in 21
watersheds around the town of Qingping in
Mianzhu County. Both towns suffered heavy
losses from the simultaneous occurrence of
debris flows.

2 Hazard formation conditions of


debris flows
Due to limiting terrain conditions, it is difficult
to find a flat and open area upon which to build
towns in many mountainous areas. The joint
debris flow alluvial fans in alpine gorge regions
generally have advantageous conditions such as
a gentle slope and thus convenient traffic and
water infrastructure. Thus, these regions
become preferred sites for town construction.
As is well known, old deposits from all sizes
of debris flows gradually form these fans. So,
owing to the quasi-periodicity of debris flow
development (Chen, 2002), large-scale debris

520

Lanzhou (capital
of Gansu
Province)

Zhouqu

Zhuanglang

Beichuan

Jiuzhaigou

Songpan

Huaying

Yaan (capital of
Yaan
prefecture)

Name of
city or town

No.

E106.44, N30.26

E102.97, N29.97

Sichuan

E103.61, N32.64

E104.23, N33.27

E104.44, N31.89

E106.06, N35.20

Sichuan

Sichuan

Sichuan

Sichuan

Gansu

E104.38, N 33.81

E103.51, N36.04

Gansu

Gansu

Location

Province

15

Number of
debris flows

Table 2. Cities and towns affected by adjacent debris flows.

Luwang Gully,
Ganxi Gully

Jianzi Gully, Cuoji Gully,


Pianyanzi Gully,
Guanyin Gully

Guanmiao Gully, Bala


Gully, Cuoji Gully,
Houshan Gully
Shangyao Gully, Songlin
Gully, Dongyu Gully,
Zhongzhao Gully,
Taping Gully

Weijia Gully, Xijia Gully,


Huashiban Gully

Wenjia Gully, Shijia


Gully, Lijiazui Gully

Hongshui Gully,
Wuquanshan Gully,
Huangyu Gully, Jian
Gully, etc.
Luojiayu Gully, Sanyanyu
Gully, Dazai Gully

Name of debris
flow catchments

(continued)

On 7 August 1978, 14 debris flows simultaneously broke out in the


Xujiawan district of Lanzhou city, causing tens of deaths and
losses of more than 3 million yuan. In addition, a debris flow in
Hongshui Gully on 20 July 1964 killed 157 people.
Two simultaneous debris flows occurred in the Sanyanyu and
Luojiayu catchments on 8 August 2010, ruining Zhouqu
Township (including three villages), causing 1765 deaths, and
leaving 22,667 others homeless. The dumped debris blocked the
Bailong River and formed a barrier lake, which inundated half of
Zhouqu Township for over 20 days. The hazard destroyed 4321
houses and 21 buildings.
On 27 April, debris flows simultaneously broke out in Wenjia Gully,
Shijia Gully, and Lijiazui Gully. The catastrophe buried more than
4000 houses, killed 800 people and 2600 poultry, and totally
destroyed 670 ha of farmland.
Debris flows in Weijia Gully, Xijia Gully, and Huashiban Gully
around Beichuan Township were simultaneously triggered by
intense rainfall on 24 September 2008, silting half of the township,
burying earthquake remains, and causing 21 deaths and
enormous property loss.
During heavy rainfall on 18 July 1984, debris flows simultaneously
occurred in Guanmiao Gully, Bala Gully, and Cuoji Gully, causing
35 deaths and losses of more than 54.43 million yuan.
In 1988, debris flows in Shangyao Gully destroyed 34 houses and
more than 8.7 ha of farmland. The debris flows surrounding
Songpan Township threatened the safety of more than 2025
people as well as having the potential for an enormous amount of
property damage.
Debris flows simultaneously occurred in Jianzi Gully, Cuoji Gully,
Pianyanzi, and Guanyin Gully on 2 July 1986. This event resulted
in four deaths, nine badly wounded people, and losses of 20
million yuan.
A death toll of 169 persons resulted from simultaneously occurring
debris flows in Luwang Gully and Ganxi Gully on 2 November 1979.
One street of Yaan city was ruined, resulting in very heavy losses.

Description

521

Ningnan

Maerkang (capital
of Aba Tibet
Autonomous
prefecture)

Jinchuan

Dongchuan

Qiaojia

10

11

12

13

Name of
city or town

No.

Table 2. (continued)

Yunnan

Yunnan

E102.92, N26.90

E103.12, N26.06

E 102.03 N 31.48

E102.22, N31.92

Sichuan

Sichuan

E102.76, N27.07

Location

Sichuan

Province

More than 10

Number of
debris flows

Shiyang Gully, Nilagu


Gully, Shen Gully,
Zhuguoshi Gully,
Tianba Gully and
lalihe Gully
Shuilian Gully, Laoshu
Gully, Shihuiyao
Gully, Baini Gully,
Fuxing Gully,
Gaoliangdi Gully,
Heini Gully and Xifan
Gully

Dalangzu Gully,
Narisichazu Gully
Minjingzhongdui Gully
Sanjiacun Gully,
Chamuqiao Gully, etc.
Babuli Gully, Caijia Gully

Yangjuan Gully,
Yingyang Gully, Shenjia
Gully

Name of debris
flow catchments

Simultaneously occurring debris flows in Shuilian Gully, Baini Gully,


Fuxing Gully, Gaoliangdi Gully, and Heini Gully during 2427
August 1980 destroyed 71 houses, more than 171 ha of farmland,
and 800 m of road. More than 2000 t of grain was lost due to the
event.

Jinchuan Township has suffered from debris flows many times. A


debris flow on 22 June 1926 killed 17 people and destroyed nine
mills. Debris flows on 4 July 1976 washed away a small factory and
five houses, and debris flows on 22 July 1980 seriously damaged a
hospital and a food-processing plant.
In 1964, simultaneously occurring debris flows in Shen Gully and
Shiyang Gully resulted in eight deaths and the destruction of 20
ha of farmland.

In 1954, a debris flow in Yangjuan Gully destroyed more than 10


houses and more than 6.8 ha of farmland. A second event in the
same year, which had debris flows occurring simultaneously in
Yingyang Gully and Shenjia Gully, ravaged one bridge on Heishui
River and destroyed 11 houses and more than 20 ha of farmland.
Simultaneous adjacent debris flows in Yangjuan Gully and
Yingyang Gully destroyed a total of 14 houses and 12 ha of
farmland during 527 June 1983.
Debris flows broke out in multiple gullies in 1949, 1953, and 1957,
respectively, causing heavy losses. Debris flows threatened the
security and property of more than 1300 people.

Description

522

Progress in Physical Geography 37(4)

Figure 1. Dongchuan Township, located on the joint alluvial fan of six debris gullies, repeatedly suffered from
debris flows before it was controlled (SPOT-5 image). In 1964, debris flows simultaneously broke out in Shen
Gully and Shiyang Gully. The catastrophe resulted in eight deaths and the destruction of 300 ha of farmland.

flows can break out at the same location in the


future under appropriate conditions. Once this
happens, the town on the fans will suffer serious
damage. In the Wenchuan earthquake affected
area, a large number of landslides triggered by
earthquakes provided abundant loose material
for debris flows. Although the debris flows in
some gullies had not broken out for a long time,
after the Wenchuan earthquake debris flows
became more active (Cui et al., 2011a; Hu
et al., 2010; Liu et al., 2010). Moreover, some
gullies changed into debris flow gullies, wherein
if the intensity of rainfall reaches a critical
amount it triggers the loose material in those
channels to form debris flows. Compared with
debris flows before the Wenchuan earthquake,
the magnitude and frequency of debris flows after
the earthquake have increased, to the extent that it
has become a major hazard for mountain towns
located on debris flow fans (Cui et al., 2011a). For
instance, because of locally intensive rainfall on
24 September 2008, debris flows simultaneously
occurred in Weijia Gully, Xijia Gully, and

Huashiban Gully around Beichuan Township


(Figure2). According to field investigations, the
cross-sectional peak discharge reached 225
m3/s. This scale of debris flow is much larger than
what occurred before the earthquake as the result
of similar rainfall intensity. The giant debris flow
silted Qushan Town, the old seat of Beichuan
County, which is regarded as a reserved earthquake relic, and caused 21 deaths and enormous
property loss (Figure 3; Table 2).

3 Hazard characteristics of simultaneous


debris flows
As shown in Figure 4, when a mountain town
suffers from multiple debris flows there are particular characteristics to the interaction effects
of the hazards:
1. Compound and overlapping disasters can
result due to simultaneous debris flows.
Many sites in a town may be hit by interconnecting disasters, or one location can

Cui et al.

523

Figure 2. The spatial distribution of debris flows of Xijia Gully, Huashiban Gully, and Weijia Gully. Beichuan
Township is located in the dangerous zone of the three debris flows.
Source: National Administration of Surveying, Mapping, and Geoinfomation.

Figure 3. The debris flow activity and its disastrous results in Qushan Town of Beichuan County after the
earthquake. (A) Qushan Town after the earthquake (before debris flow). (B) Qushan Town after the debris
flow event on 24 September 2010. The debris flow ruined and buried many buildings.

suffer various successive ones. For


instance, two different potential debris
flows overlap point A in Figure 4, while
point B is affected by a debris flow and
can be subsequently inundated by flooding due to an upstream dam failure. In
another scenario, debris flow may silt up
the river channel and shift its course,
inducing a riverbed change that leads
to a cross-section reduction and a rise
in water level that can eventually flood

river terraces and floodplains (point C in


Figure 4).
2. Multiple hazards can result in a chain reaction of disasters due to the initial debris
flow. Not only can the debris flow move
directly through the town and cause impact
and silting damages for the township,
but large-magnitude flows can also block
rivers and lead to an upstream inundation
hazard. When the makeshift dam collapses, it can result in flood disasters

524

Progress in Physical Geography 37(4)

and national road 213 upstream of the barrier. The


dumped debris forced the main channel of the
river to shift to the right past the barrier, causing
flooding in the brand new Yingxiu Township,
reconstructed after the Wenchuan earthquake
(Figure 5). This compound and interconnecting
disaster caused extremely heavy losses (Table 1).
At present, the threat of large-magnitude
and simultaneous debris flows confronts many
mountain towns. However, previous studies
have paid little attention to the issues of chain
reactions, continuity, and the overlapping
characteristics of potential disasters when analyzing the risk to these mountain townships,
and there has been little research analyzing the
urban hazards of debris flows occurring in a
group after the Wenchuan earthquake. Therefore, it is important to explore a new method
to solve this issue.

Figure 4. Hazard characteristics for a mountain


township due to debris flows occurring in a group.

downstream. The above successive disasters form a disaster chain called debris
flowbarrier lakeflood.
3. The third characteristic is the continuation of the disaster. Simultaneously
occurring large-magnitude debris flows
are likely to cause huge compound catastrophes and disaster chains. Those interconnecting and superposition hazards
often expand spatially and continue
successively.
The debris flow in Hongchun Gully is a typical
example of these three hazard characteristics. The
debris flow occurred on 14 August 2012, delivering a huge amount of debris that blocked the
watercourse of the Min River. The impounded
water behind the debris barrier submerged Yingxiu Hydropower station, Shaohuoping Bridge,

III Risk analysis methods for


mountain townships suffering from
simultaneous debris flows
1 Method for hazard analysis
In order to analyze the hazard characteristics of
debris flows occurring in a group around mountain townships, we propose a systematic and
quantitative hazard analysis method supported
by numerical simulation of debris flow movement and flood analysis. The proposed model
is expressed as:
D De Dh Di Df

Here, D is the total hazard degree, De is the


hazard caused by the impact force of the debris
flow indexed to the maximum kinetic energy
value in each grid during the whole debris flow
movement process, Dh is the hazard caused by
debris flow silting indexed to flow depth, Di is
the inundating hazard of the barrier lake
indexed to the inundated backwater depth, and
Df is the dam-failure flood hazard indexed to
the highest water level of the flooding.

Cui et al.

525

Figure 5. Disaster chain induced by the debris flow that occurred in Hongchun Gully. The debris flow on 14
August 2010 destroyed and blocked national road G213, delivered debris, and formed a huge barrier in the
channel of Min River which pushed the main stream toward the right bank, and then caused a flood disaster in
the newly reconstructed Yingxiu Township.
Source: Sichuan Bureau of Surveying, Mapping, and Geoinfomation.

526

a Numerical approach debris flow processes.


Flow velocity is a key parameter for identifying
the impact force of a debris flow, while the flow
depth can reflect the silting hazard (Kienholz,
1999; OBrien et al., 1993; Rickenmann,
2001; Wei et al., 2006). When discussing debris
flow deposits, the debris flow motion equation
includes three important variables: mud depth,
the x-velocity component, and the y-velocity
component:
Du
gSsx  gSfx
Dt
2
Dv
gSsy  gSfy
Dt
Here, u and v are x-component and y-component velocities respectively (m/s), g is acceleration due to gravity(m/s2), Ssx is the bottom slope
of the deposition area in the x-direction, Ssy is
the bottom slope of the deposition area in the
y-direction, Sfx is the friction gradient of
the debris flow in the x-direction and Sfy is the
friction gradient of the debris flow in the
y-direction.
According to OBrien et al. (1993), Sfx and Sfy
can be calculated using:
p
B
2B u kc u u2 v2
Sfx

sgnu
m h2
gh
m h
p 3
B
2B v
kc v u2 v2

Sfy
sgnv
2
m h
gh
m h
where B is the yield stress (N/m2), m is
the density of the debris flow (t/m3); h is the
flow-depth (m), B is the viscous coefficient
(N.s/m2), and kc is the roughness coefficient.
The model treats debris flow masses as
aggregates of many small particles, each of
which has its own mass and velocity. To solve
equation (2) numerically, Hu and Wei (2005)
improved the particle model originally developed by Wang et al. (1997), while Cui et al.
(2011c) discussed the method and approximated the debris flow movement by using the
forward difference for each particle. The

Progress in Physical Geography 37(4)

difference equations can thus be expressed as:


un1
 unk
n;k
k
gSsx
 gSfxn;k
t
vn1
 vnk
n;k
k
gSsy
 gSfyn;k
t

are the values of u and v for the


where ukn1 , vn1
k
k-th particle at time n1, respectively, and unk ,
n;k
n;k
, Ssy
, Sfxn;k , and Sfyn;k are the values of u,
vnk , Ssx
v, Ssx, Ssy, Sfx, and Sfy for the k-th particle at time
n.
Particle movement can be traced using the
MAC (marker-and-cell) computational technique (Hu and Wei, 2005). A digital elevation
model (DEM) grid of the real topography of the
debris flow gully is generated using GIS, and
provides division of the computational cells.
Therefore, each grid has a value for flow depth,
velocity, and elevation.
b Methods for determining the impact force and the
silting depth of debris flow. We can simulate the
debris flow movement process in an alluvial
area by adopting the numerical approach in section III.1.a, using the DEM of the studied area to
identify the spatial distributions of velocity and
flow depth for each grid square during the
movement process. In order to determine silting
damage, we adopted the maximum flow depth
of the debris flow to index the silting hazard.
The flow depth of each grid (i, j) can be calculated using the following formula:
Dh maxh

Nni;j V
A

where Nni,j is the number of particles on grid


(i, j) at time n, V is the average volume of each
particle (m3), A is the grid area (m2), h is the
flow depth (m), and Dh is the silting degree or
maximum deposit depth (m). Generally, larger
values of Dh represent a more serious hazard.
In order to easily calculate the value of Dh, we
simplify the process of overlapping deposits of
joint and opposite debris flows as the accumulation of sequential debris flows. Therefore, we

Cui et al.

527

can identify the deposition range and the thickness of each debris flow in the same coordinate
system, and obtain the total thickness of deposition by summing the thickness of each overlapping part.
In order to characterize quantitatively the largest impact hazard at each grid position, we index
the impact force and damage capability of a debris flow by applying the maximum kinetic
energy. This process allows us to calculate the
impact degree of each grid square for the entire
dynamic debris flow process. The simulation
starts from the gully mouth (at the top of the
alluvial fan) and is implemented by determining:


De A  max u2 v2 h
t>0
Nn

1 Xij
uk
u
Nnij k1

Nn

1 Xij
vk
Nnij k1

where De is the maximum value of the kinetic


energy of the debris flow (Nm), u and v are the
velocities (m/s) in the x and y directions, respectively, h is the flow depth (m),  is the density of
the debris flow (t/m3), and A is the grid area
(m2), Nni,j is the number of particles on grid
(i, j) at time n, uk, vk are the velocities (m/s) for
the k-th particle on grid (i, j), respectively.
c Method for hazard analysis of a debris flow barrier
lake. In general, flow depth is an important
element for identifying the indirect hazard of
dam-failure floods and river-blocking backwaters. In order to obtain the distribution of
submerged range and depth in mountain towns,
we consider two steps: (1) calculating the
reservoir storage of the barrier lake; and (2)
analyzing the process of the dam-failure flood
based on the DEM.
The possibility of river blockage caused by a
debris flow can be calculated for a given severity of rainfall. Cuis previous experiments

proposed the following fundamental discriminant in order to identify whether a debris flow
has the capacity to block the main channel of
a river (Cui et al., 2006):
CM 1:1891  cos 2 3:677B =M
 1nQM M =QB B  12:132
CF 0:8831  cos 2 2:587B =M

 1nQM =QB  8:572

7
8

where CM is the momentum criterion of the river


blockage, CF is the discharge criterion of the river
blockage, QM and QB are the flow discharges
of stream flow in the main channel and the
debris flow in the tributary (m3/s), respectively,
M and B are the maximum velocities of the river
flow and the debris flow (m/s), respectively, B is
the debris flow density in the tributary (t/m3), M
is the water density in the main channel (t/m3), and 
is the intersection angle between the main channel
and the tributary ( ). If CM  12.132 or CF 
8.572, the debris flow is likely to block the river
channel.
There are two methods for calculating the
height of the barrier dam: a numerical simulation and an empirical formula. Cui (Cui et al.,
2011c) suggested a method to calculate the
debris flow process for a given rainfall condition. By combining the peak discharge (Qp)
obtained by field investigation and theoretical
calculation, the flow depth distributions at
each grid square during the movement process
and final silting range can be obtained by the
simulation method mentioned in section
III.1.a. The deposit depth at the river channel
calculated by equation (5) indicates the height
of the debris barrier dam. On the other hand, if
we know the flow duration of the debris flow,
we can calculate the total runoff using the
empirical formula:
19
9
Vc TQp
72
where Vc is the total runoff of the debris flow
(m3) and T is the flow duration time of the debris

528

Progress in Physical Geography 37(4)

flow (s). The height (H) of the debris barrier


dam can be estimated by:
H 2Vc=LBU BD
10
where L is the length of the barrier dam in the
flow direction of the debris flow (m), BU is
the top width of the barrier dam (m), and BD is
the bottom width of the barrier dam (m).
Consequently, the submerging hazard degree
(Di) of the barrier lake, characterized by the
inundated depth, can be calculated as:
Di H0 H  Hi

11

where H0 is the bottom elevation of the barrier


dam (m); H is the height of the debris barrier
dam (m); Hi is the elevation of the selected point
in the area of the town (m); and Di is the inundated depth for the selected point in town (m).
d Method for hazard analysis of dam-failure floods.
The hazard degree (Df) of a dam-failure flood is
acquired by calculating flood discharge and water
level. We select the worst-case scenario of a complete collapse of the barrier dam to calculate the
magnitude of the dam-failure flood in order to
maximize the value of the potential risk. If the
dam-failure flood goes beyond the designed flow
capacity of anti-flood engineering, the mountainous town may be in risk. The inundated depth and
area can be calculated using the towns topographical data and Schoklitschs formula (Schoklitsch,
1948) to calculate peak discharge at the dam site:
 1
8 p Bu 4 3
g
bH 2
12
Qmax
27
b
where Qmax is the maximum discharge at the dam
site (m3/s), BU is the width at the top of the barrier
dam (m), b is the width of the sluice channel (m),
and g is gravitational acceleration (9.8 m/s2).
The maximum discharge at each cross-section
of the river downstream of the dam site is:
W
13
QLM W
L
Qmax VK

where QLM is the largest flood discharge at each


cross-section (m3/s), L is the distance from the
dam site to the river cross-section (m), W is the
water storage of the barrier lake (m3), Qmax is
the largest flood discharge at the dam site
(m3/s), and VK is an empirical coefficient equal
to 7.15 in mountainous areas, 4.76 in hilly areas,
and 3.13 in the plains (Li, 2006).
To calculate the discharge and water level of
a dam-failure flood at each cross-section, we
use equations (12) and (13), and the terrain surrounding the river. Using these results and the
townships DEM, the inundated area (Af) and its
flood depth (Df) is determined by:


Q

nQ
LM
0
R2=3 I 1=2
Af
14

2=3
1=2

Q

nQ
LM
0 R
I Bi
15
Df
where Q0 is the flow capacity of each selected
river cross-section (m3/s), R is the hydraulic
radius (m), I is the hydraulic slope, Bi is the
width of the river cross-section (m), and n is
roughness of river channel.
After identifying all the information concerning potential debris flow and flood inundation
damage, the distribution of the debris flows
destructive power, the debris flow depth, and
the dam-failure flood depth can be calculated
using GIS spatial analysis functions. This process thus provides a complete hazard zoning for
mountain townships. In general, the collapse of
an upstream debris dam often occurs after debris
flow has deposited debris in the township. In
order to calculate the inundation range of the
upstream dam-failure flood, we start with the
method proposed by Cui and Hu (Cui et al.,
2011c; Hu and Wei, 2005) to calculate the range
and depth of the debris flow deposit within the
urban area, then input the changed topographic
conditions of the silted township as a new
variable for calculating the inundation range and
flood depth of the dam-failure flood from
upstream. If one location suffers from several
hazards (e.g. impact force, silt buildup, and flood
inundation), the largest hazard value is selected

Cui et al.

to express the hazard degree in one hazard process, and then the sum of each individual largest
value of different hazard processes will represent
the total degree of hazard for that location.

2 Method for vulnerability analysis


Scientists with different scientific backgrounds
have a different understanding regarding the
definition of vulnerability (Hufschmidt, 2011;
Papathoma-Kohle et al., 2012). Social scientists
focus more on the characteristics of the society,
relating vulnerability only to the social context
(Bohle and Glade, 2007). On the other hand,
natural scientists and engineers often describe
vulnerability as the degree of loss to an element
at risk (Totschnig et al., 2011; UNDRO, 1984).
In addition, some scientists think there is neither
a common definition for vulnerability nor a
standard methodology for vulnerability assessment (Alcantara-Ayala and Goudie, 2010;
Papathoma-Kohle et al., 2011).
In this paper, we primarily analyze the resistance of hazard-affected objects for vulnerability
assessment of the debris flow. Moreover, there
are five main aspects taken into consideration:
1. identifying the types of hazard-affected
objects;
2. summing the quantity and distribution of
all kinds of hazard-affected objects;
3. establishing the vulnerability assessment
model;
4. calculating vulnerability degree;
5. zoning sub-vulnerability areas.
Through analyzing the spatial and spectral characteristics of the hazard-affected objects in
remote sensing images, we identified the shape,
size, image, shadow, and texture of the hazardaffected objects. Then we applied the Chinese
National Land Survey Technique Rules (TDT
1014-2007) as criteria for determining the types
of hazard-affected objects, and we used field
investigation for verification. We extracted the
types, quantity, and spatial distribution of the

529

hazard-affected objects from remote sensing


images of the area surrounding the township,
which provides the data for calculating degree
of vulnerability. Degree of vulnerability is
related to the economic value and the susceptibility index of hazard-affected objects, defined by:
V V u  C

16

where V is the vulnerability degree, V(u) is the


economic value of a type of hazard-affected
object, and C is the susceptibility value of the
corresponding object. V(u) is calculated by
multiplying the unit price P of a type of hazardaffected object by its size (area or length) N:
V u P  N

17

The susceptibility (C) indicates the degree of


resistance of the hazard-affected object against
a debris flow. Values for the variable range from
0 to 1, with larger susceptibility values indicating
a more vulnerable hazard-affected object. The
structure and material of a hazard-affected object
determine its susceptibility and damage mode.
Furthermore, the location of a hazard-affected
object in a debris flow gully also influences its
potential damage level. In the main area of the
debris flow gully, hazard-affected objects often
suffer from impact damage, while objects located
in outlying areas of the debris flow fan mainly
suffer from sediment damage. However, most
sediment-affected objects cannot be recovered
or reused either; although the objects maintain
their structure, their function is hard to recover
because of the change in terrain. Thus, the proposed method provides a susceptibility calculation for each hazard-affected object.
For hazard-affected objects sedimented by
debris flow, the susceptibility index (CD) can
be indexed according to the ratio of the debris
flow depth to construction height:
CD

hD
HC

18

Where hD is the debris flow depth (m) and HC is


the construction height (m), such as the height

530

Progress in Physical Geography 37(4)

Table 3. Susceptibility indexes of buildings or structures.


Types of
structures
Adobe
construction
Timber structure

Susceptibility Susceptibility
grades
values
Characteristics*
V

0.9*1.0

IV

0.8*0.9

Brick-wood
structure
Brick-concrete
structure

III

0.5*0.8

II

0.2*0.5

Steel reinforced
concrete
structure

0.1*0.2

Small-scale debris flows can entirely destroy this type of


structure.
Small-scale or medium-scale debris flows can seriously
damage this type of structure.
Small-scale or medium-scale debris flows can partially
destroy this type of structure.
Small-scale or medium-scale debris flows do not generally
affect this type of structure, but it can be partially
destroyed by large-scale debris flow.
This type of structure is not generally affected in smallscale or medium-scale debris flows, but it can be partially destroyed by a devastating debris flow of huge
magnitude.

*According to the Specification of Geological Investigation for Debris Flow Stabilization (DZ/T 0220-2006), four grades
of debris-flow magnitude are classified by the total runoff as small-scale debris flow for the total runoff less than
1104 m3, medium-scale debris flow for between 1104 m3 * 10104 m3, large-scale debris flow for between
10104 m3 * 100104 m3, and mega debris flow for larger than 100104 m3.

of a bridge or house. When HhDC  1, CD will


have a value of 1 as the building or structure
is completely buried.
The susceptibility index (CI) varies with
respect to hazard-affected objects impacted by
debris flows such as building structures or constructions. To quantitatively determine the susceptibility value of different constructions
needs a large number of failure trials or experiments. Moreover, a significant number of real
structure damage survey cases are needed to
confirm the index value (CI). For our method,
we classify all prone buildings or structures in
a debris flow area. In this paper, we carried out
an actual investigation of a disaster situation in
western China, analyzing the data of destroyed
building or structures to obtain the susceptibility
indexes in Table 3 and the probability values for
different building structures.

3 Risk assessment method of debris flows


The definition of disaster risk proposed by
the UN Department of Humanitarian Affairs

(UNDHA, 1992) quantifies risk as:


RDV

19

where R is the risk level, D is the hazard level,


and V is the vulnerability level.
a Process of risk analysis. We use the following
steps to carry out a risk analysis of debris flows
for mountain townships (Figure 6). First, we
build a spatial data set of the debris flow basin
that includes DEM data. Next, we ascertain the
rheological properties and kinematic parameters of the debris flow. From this information, we can calculate the hazard degree by
applying the hazard analysis method described
above. After identifying the types, quantity, and
distribution of hazard-affected objects, we can
then calculate the susceptibility of the hazardaffected objects by combining field survey data
with the described vulnerability analysis
method. Finally, the evaluation formula provides the degree of risk, from which we can
assign risk levels and risk zonation.

Cui et al.

531

Figure 6. Flow chart of the risk assessment process for debris flows.

b Data normalization method. According to equation (19), the degree of risk is the product of the
hazard degree and the degree of vulnerability.
Since the dimensions of the hazard index and
the vulnerability index are different, we must
normalize the hazard and vulnerability data.
We normalize the indexes by adopting the
following approach:
Hi
Vi
0
0
; Vi
20
Hi
Hmax
Vmax
0

where Hi indicates the normalized value of the


hazard degree, Hi is the initial hazard index,
0
Hmax is the maximum hazard index, Vi indicates
the normalized value of the vulnerability
degree, Vi is the initial vulnerability index, and
Vmax is the maximum vulnerability index.
The determination of Hmax and Vmax is significant for risk assessment. In general, the values
of Hmax and Vmax are determined by referring to
losses of past events and typical case studies.

The hazard value of the largest disaster in the


study area is regarded as Hmax, and a similar
process can be used to determine the value Vmax.
c Grading method of evaluation results. When
implementing a risk evaluation, the evaluation
indicators are divided into several grades to
describe the degree of hazard. The indicator
values are evenly partitioned between the maximum and minimum values; or, based on thresholds provided by experts, may be divided into
three levels of high, medium, and low (Hu and
Wei, 2005; Rickenmann, 2001; Wei et al.,
2006). We automatically divide the evaluation
indicators such as hazard degree, vulnerability
degree, and risk degree into three grades by
applying a probabilistic method using ArcGIS
and the following condition:
M r i  1V r < r < M r iV r
21

532

where M(r) and V(r) are the mean value and variance of the evaluation indicators, respectively
(where i is the number of grades). If the value of
grid r falls in the range of [M(r), M(r) V(r)], the
indicator is regarded as medium level. In the same
way, the range >M(r) V(r) belongs to the high
category, and the range <M(r) identifies the low
category. Note that if more grades are required, i
can be assigned a larger number such as 4 or 5. This
method generates reasonable results and avoids
the negative effects caused by abnormal data.
d Risk mapping. Two kinds of hazards must be
determined to develop a debris flow risk map for
mountain townships: the impact and sediment
hazards directly resulting from the debris flow
and the indirect hazards such as dam-failure
floods and backwater inundation induced by
a debris flow barrier lake. Features such as the
distribution of risk degree, levels of risk, and
location of the risk zone are key factors in risk
mapping. Applying GIS-based multi-map algebra analysis (David, 2001) using ArcGIS 9.3,
we can calculate the distribution of the degree
of risk based on the results of the hazard and vulnerability analyses. Using the calculated data and
grading method described above, we establish
the three risk levels of high, medium, and low
risk, combining grids belonging to the same level
of risk and applying different colors to represent
each of the different levels. We can then use this
data to construct a map of risk zonation for debris
flows by constructing polygons in ArcGIS.

IV Case study: risk assessment for


Qingping Township
1 Background of the debris flow
The topography of Qingping Township is characterized by high mountains with steep slopes
that average over 25 , while the slopes in the
upper reaches of the valley or in the debris
source areas reach 35 45 . The longitudinal
profiles of the gully channels typically range
between 105% and 400%. Therefore, the terrain

Progress in Physical Geography 37(4)

conditions in the region are especially conducive to debris flow formation.


Qingping Town is located at the front of the
Longmenshan fault belt, an active geological
tectonic structure. The Qingping-Baiyunshan
fault, which is a branch of the Longmenshan
fault, stretches through the township from
northwest to southeast. The rock strata in the
region are heavily fragmented due to active tectonic activity caused by the Longmenshan fault.
The exposed strata are mainly from the Triassic,
Permian, Devonian, Cambrian, and Sinian
Periods. Shale, mudstone, sandstone, limestone,
and interbedded soft and hard rocks dominate
the lithology of the area. Furthermore, the rupture of 5.12 Wenchuan earthquake extended
from southwest to northeast and passed through
Qingping Township. The earthquake intensity
was identified as XI degree in the Qingping
area. The earthquake triggered a large number
of landslides in the region and engendered abundant unconsolidated soil for future debris flows.
In Wenjia Gully, a gigantic landslide provided
6107 m3 of loose fragmented material accumulation, which became a rich source for a
subsequent debris flow (Xu, 2010).
Qingping Township frequently suffers rainstorms due to the influence of the monsoon in the
region, and the intensive rainfall generated in these
stormsis the major triggering factor ofdebrisflows.
After the Wenchuan earthquake, it was common
for large-scale debris flows to occur simultaneously during highly intensive rainfalls. For
instance, 21 debris flows occurred around Qingping Township during the rainstorm from 18:00 of
12 August to 4:00 of 13 August 2010 (i.e. the 813 event), when 230 mm of precipitation fell
within 10 hours (Cui et al., 2011b) (Figure 7).

2 The 8-13 debris flow disaster


During this heavy rainfall, 21 debris flows occurred
simultaneously and ruined Qingping Township
(Table 1; Figure 8). The damage that resulted from
the catastrophe can be classified into three

Cui et al.

533

Figure 7. Distribution of the debris flows that occurred simultaneously during a rainstorm on 13 August
2010. The clustered debris flows hit Qingping Township heavily.

categories. The first is the damage from the debris


flows, such as those in Luojia Gully, Dongzi Gully,
Wawa Gully, Linjia Gully, and Taiyang Gully,
which directly destroyed or silted structures. The
second category is damage from the debris flows
that partly blocked the river channel and caused
flooding, such as the debris flows in Shaoyao Gully
and Zoumaling Gully. The most serious category
is the debris flow in Wenjia Gully that entirely
blocked the Mianyuan River, causing a damfailure flood disaster for Qingping Township.

3 Risk calculation
a Deposition range simulation. Through field survey data and visits with local people, we obtained
the basic parameters of the debris flows: length
of main channel, area of catchment, channel
slope, roughness coefficient of each channel,

and particle size (Table 4; Figure 9). We then


used these results and the measured grain-size
distribution (Figure 9) to calculate the other
necessary parameters of each debris flow. Equation (22) (Chen et al., 2003) gives the densities
of the debris flows, while equations (23) and
(24) (Wu et al., 1993) provide the viscous coefficients and yield stresses of each debris flow.
We calculated the peak discharges of the debris
flows using equations (25) and (26). We then
simulated the debris flow motion process on the
alluvial area with the support of a DEM with a
55 m grid and the basic parameters listed in
Table 4.
  1:32  103 x7  5:13  102 x6 8:91x5
 55x4 34:6x3  67x2 12:5x 1:55
22

534

Progress in Physical Geography 37(4)

Figure 8. Scenario of the debris flow disaster in Qingping Township on 13 August 2010.
Source: Land and Resources Department of Sichuan Province.

  27:0 

 0:1
;   2:0 t=m3
1:635
  4:42 

;   2:0 t=m3
 0:1
0:60

23

  7:41 

;   2:2 t=m3
1:394
  45:0 

;   2:2 t=m3
 0:1
2:096

24

 0:1

QP 1  QB  DU

25

  ! =  S

26

In the above equations,  is the density of the


debris flow (t/m3), x is the percentage of clay
content in the resulting deposit,  is the viscous
coefficient of the debris flow,  is the yield
stress of the debris flow, QP is the peak discharge of the debris flow (m3/s), QB is the peak
discharge of the water flow (m3/s), is correction coefficient of the peak discharge of the
debris flow, DU is the blocking coefficient of the
debris flow channel, ! is the water density
(t/m3), and s is the density of the solid components in the debris flow (t/m3).

The simulated debris flow generates the spatial distribution of the velocity and flow depth
(Figure 10a) at each grid in the alluvial area,
with the deposition range provided by supporting GIS-based data conversion analysis using
ArcGIS 9.3. The simulated results show that
the debris flows in Luojia Gully, Wawa Gully,
Dongzi Gully, Linjia Gully, and Taiyang Gully
directly destroy buildings and bury roads, and
the debris flows in Wenjia Gully and Zoumaling Gully obviously block the Mianyuan River
and damage roads, farmland, and houses along
riverbanks. Using equation (12), we calculated
the maximum flood discharge at the dam site,
and the maximum discharge of river crosssection downstream of the barrier using equation (13). We then combined these results with
equations (14) and (15) by using ArcEngine
and C# to get the inundated area and flood
depth of each grid, as shown in Figure 10b.
b Hazard degree calculation. According to the
simulation results above, each component De,
Dh, Di, Df can be acquired through the hazard
analysis method in subsection III.1. Then, by
applying the normalization method from equation (20), the hazard degree (D) in the alluvial
area is calculated by applying equation (1), with
division values of 0.6 and 3.0 as given by

Cui et al.

535

Table 4. Basic parameters of numerical simulation for debris flows.


Parameter
Basic parameters
of gullies

Rheological
parameters

Kinematic
parameters

Linjia Taiyang Wenjia Zoumaling Luojia Dongzi Wawa


Gully Gully
Gully
Gully
Gully Gully Gully
Area of catchment
1.09
0.40
7.68
(km2)
Length of main
1.51
0.83
4.38
channel (km)
Average channel
250.41 484.34 322.66
slope (%)
Channel roughness
0.44
0.41
0.44
coefficient
2.10
2.10
2.10
Density (t/m3)
2.08
2.08
2.08
Viscous coefficient
(N.s/m2)
86.09 86.09
86.09
Yield stress (N/m2)
Time-interval (s)
0.20
0.20
0.20
103
1177
Peak discharge (m3/s) 233

5.46

1.30

0.20

0.56

3.10

1.10

0.48

0.92

183.13

465.98 566.02 367.19

0.44

0.44

0.41

0.44

2.15
2.48

2.15
2.48

2.10
2.08

2.10
2.08

162.50
0.20
774

162.50
0.20
323

Figure 9. Grain-size distribution of the debris flow deposit in Qingping Township.

86.09 86.09
0.20
0.20
62
130

536

Progress in Physical Geography 37(4)

Figure 10. Calculated results and risk map of Qingping Township. (a) Distribution of flow depth in the debrisflow affected area. (b) Distribution of flow depth of the dam-breaking flood. (c) Hazard zonation map. (d) Vulnerability zonation map. (e) Distribution of risk degree in Qingping Township. (f) Risk map of Qingping Township.

Cui et al.

537

Figure 10. Continued.

equation (21). The resulting three hazard zones


are the high hazard zone (>3.0), the medium
hazard zone (3.0*0.6), and the low hazard zone
(<0.6), as shown in Figure 10c. The high hazard
area accounts for 1,150,000 m2 and 45.9% of the
total hazardous area. This high hazard area is
located in the debris mainstream area, located
along main roads and Mianyuan River. This
region is so flat that it is also the most dangerous
area for potential inundation from a dam-failure
flood. The medium hazard area is 574,000 m2
in size, or 22.9% of the total area. The region is
located around the mainstream area where buildings and transport facilities are liable to suffer
from both silting and inundation damage. The
low hazard zone is 780,000 m2, accounting for
31.2% of the total area, and is mostly located in
the higher elevations of Qingping Township
where the hazard degree is relatively small and
potential damage from debris flows and a damfailure flood is minimal.

c Vulnerability degree calculation. Using the vulnerability analysis method in section III.2, the
hazard-affected objects were classified into four
categories based on the different characteristics
of the geographic elements in the area surrounding the township: buildings, roads, cropland,
and grassland. After identifying the shape, size,
image, shadow, and texture of each type of
hazard-affected object from 0.5 m resolution
aerial panchromatic images, the quantity of the
objects was determined.
In order to calculate the vulnerability degree,
we had to determine the integrated economic
value and susceptibility of the hazard-affected
objects. The value of V was determined for the
area surrounding the various hazard-affected
objects by applying the analytical tools available with ArcGIS, which can calculate area and
perimeter for each polygon. The economic value
of each hazard-affected object was obtained from
local government documents and on-site

538

investigation. Using these data, the integrated


economic value (V) of each hazard-affected
object is produced by unit price multiplying its
amount (area or length). To obtain the susceptibility C of each hazard-affected object, we performed a susceptibility analysis of the building
structure by applying equation (18) and adopting
Table 3 for an approximate value, using the median value for simplification. Finally, the vulnerability degree of each hazard-affected object
was calculated through equation (16). By applying the normalization and grading method from
equation (20) and formula (21), the vulnerability
was then graded into three levels: low, medium,
and high vulnerability.
Analysis of the vulnerability, given in Figure
10d, shows that the high vulnerability area occupies 904,000 m2, or 36.1% of the total hazard area,
located in the deposition area of the debris flow
and a 100200 m wide urban area along the towns
main roads. Residential buildings, streets, infrastructure, and civil facilities dominate this area.
The medium vulnerability area covers 861,000
m2, or 34.4% of the total hazard area, and is
located in the higher elevations and both sides of
the suburban roads. The main hazard-affected
objects in this area are roads, residential houses,
cropland, and the nearby grassland. The low vulnerability area covers 739,000 m2, which accounts
for 29.5% of the total area, and is located in the
watercourse areas and marginal urban areas. In
this zone, the hazard-affected objects are mainly
cropland, bottomland, and suburban roads.
d Risk degree calculation. The hazard value and
vulnerability value were first normalized using
equation (20). Then, using the normalized results
of the hazard degree and the vulnerability degree,
we calculated the risk degree using equation (19),
the results of which can be seen in Figure 10e.

4 Validation of calculation results


For the giant disaster in Qingping Township, the
risk degree is divided into three levels of high,

Progress in Physical Geography 37(4)

medium, and low risk, with risk gradation


threshold values of 0.35 and 0.04 as calculated
by equation (21). Using these threshold values,
we can divide the debris flow affected area of
Qingping Township into three risk zones: high,
medium, and low. A risk zoning map built
according to the risk mapping method is illustrated in Figure 10f. Statistical analysis shows
that the high-risk zone accounts for 33.4% of the
total hazard area, or 837,000 m2, while the
medium-risk zone is 792,000 m2 (31.6%) and
the low-risk zone is 875,000 m2 (35.0%). On the
other hand, through analyzing of the real quantity of damaged objects for the 8-13 debris flow
disaster in Qingping Township, the statistical
results in Table 5 show that the calculated risk
zones are in agreement with the actual distribution and damage severity experienced in the
respective hazardous areas. This validation indicates that the risk assessment results for Qingping Township are consistent with results from
the actual 8-13 debris flow disaster of 2010, and
suggests that this risk analysis method can
describe the compound and chain-reaction disaster of multiple debris flows. Thus, this method is
suitable for providing debris flow risk analysis
and risk management of mountain townships.

V Discussion and conclusion


Many mountain towns in China are located on
joint alluvial fans formed by several adjacent
debris flows. Due to the similar geological and
geomorphological conditions, debris flows may
simultaneously occur in more than one of these
catchments during periods of heavy rainfall,
often causing multiple, overlapping, and interconnecting damage processes that lead to
serious disasters. We highlighted and analyzed
this previously unattended issue based on a
study of disaster data, with which we characterized the simultaneous debris flows in mountain
townships, using parameters such as compound,
interconnection, superposition, and chainreaction effects. Large-magnitude debris flows

539

Medium-risk zone

Low-risk zone

Total

134,000
6800
495,000
107,000

0.7
1.5
0.6
4.7

25,000
1000
66,000
35,000

26,000
900
63,000
36,000

3.8
11.1
4.8
2.8

5700
400
12,300
14,000

6000
400
12,700
15,600

5.0
0.0
3.1
10.3

165,700
8300
576,300
161,000

166,000
8100
570,700
158,600

0.2
2.5
1.0
1.5

Simulated Actual
Deviation Simulated Actual Deviation Simulated Actual Deviation Simulated Actual Deviation
damaged damaged percentage damaged damaged percentage damaged damaged percentage damaged damaged percentage
value
value
(%)
value
value
(%)
value
value
(%)
value
value
(%)

Houses (m2)
135,000
Roads (m)
6900
Cropland (m2) 498,000
112,000
Forest and
grassland (m2)

Hazardaffected
objects

High-risk zone

Table 5. Statistics of the hazard-affected objects in three risk zones in Qingping township.

540

generally lead to compound hazards for mountain townships, including direct impacts and
silting destruction by the debris flow itself, and
indirect damage from backwater inundation
and dam-failure floods induced by debris flow
barrier lakes created in nearby rivers. In addition, these chain-reaction effects induce these
disasters to expand spatially and continue successively due to a positive feedback effect.
Reasonable and quantitative assessment of
the risks of multiple debris flows is complex and
beyond the ability of existing risk analysis
methods based on single debris flow. Accordingly, we established indicators to describe
compound disasters and chain-reaction effects,
and provided a quantitative method for analyzing the hazards associated with simultaneous
debris flows in mountain townships. The capacities for impact damage, silting damage, and
flooding damage were quantified using the
maximum values of kinetic energy, flow depth,
and inundated depth, respectively.
In recent studies on vulnerability, scientists
have made great progress in producing vulnerability curves or functions (Fuchs et al., 2007,
2012; Jakob and Hungr, 2005; Papathoma-Kohle
et al., 2012; Totschnig et al., 2011). In this paper,
we further considered the spatial distribution and
resistance of hazard-affected objects, and developed a method to determine vulnerability of
various hazard-affected objects, with particular
emphasis on the susceptibility indexes of buildings or structures. In the proposed method, five
steps are adopted to analyze the vulnerability of
different hazard-affected objects. The necessary
information about the potential hazard-affected
objects, in terms of type, quantity, and distribution, can be extracted from panchromatic, highresolution (0.5 m) aerial images.
With the support of a debris flow movement
numerical simulation, flood analysis, remote sensing (RS), and GIS techniques, we developed a
systematic and quantitative method of risk
assessment for mountain towns. Finally, this
method was applied to the case study of the

Progress in Physical Geography 37(4)

8-13 debris flow disaster in Qingping Town


in 2010, whereby the calculated risk assessment
was consistent with the actual effects of the
disaster. The successful validation suggests that
this method can be applied in risk analysis and
risk management of mountain townships
exposed to multiple debris flows.
Our research has some limitations. In the
hazard analysis process, some empirical equations such as equation (9) and equations (22)
(24) were applied to calculate debris flow
hazard parameters. Although these equations
were often used in previous studies, more accurate tests are necessary in order to further
improve the precision of those empirical equations or to develop theoretical models. For
instance, equation (22) of the debris flow density calculation was established based on over
100 tests and investigated data sets; however,
there exists an opportunity for improvement
by replacing the current polynomial equation
with a theoretical model based on dimensional
analysis. Compared to objects of construction
and properties in mountain townships, human
beings represent a mobile object whose vulnerability it is difficult to calculate accurately.
Accordingly, we did not consider the loss of life
in our appraisal of vulnerability. The mobility of
human beings as a hazard-affected object produces new challenges for vulnerability analysis.
Therefore, it is vital to develop a solution to
this issue in future vulnerability analysis of
debris flows in mountain townships.
Acknowledgements
The authors thank George Malanson and the other
two anonymous reviewers of this paper for their
detailed remarks and helpful discussions.

Funding
This research was supported by the National Basic
Research Program of China (973 Program) (Grant
No. 2011CB409902; 2008CB425802) and the National Nature Science Foundation of China (Grant
No. 41030742).

Cui et al.

References
Archetti R and Lamberti L (2003) Assessment of risk due
to debris flow events. Natural Hazards Review 4:
115125.
Alcantara-Ayala I and Goudie AS (2010) Geomorphological Hazards and Disaster Prevention. Cambridge:
Cambridge University Press.
Bisson M, Favalli M, Fornaciai A, et al. (2005) A rapid
method to assess fire-related debris flow hazard in the
Mediterranean region: An example from Sicily (southern Italy). International Journal of Applied Earth
Observation and Geoinformation 7: 217231.
Bohle HG and Glade T (2007) Vulnerabilitatskonzepte in
Sozial- und Naturwissenschaften. In: Felgentreff C and
Glade T (eds) Naturrisiken und Sozialkatastrophen.
Berlin: Springer, 99119.
Calvo B and Savi F (2009) A real-world application of
Monte Carlo procedure for debris flow risk assessment.
Computers and Geosciences 35: 967977.
Cannon SH, Gartner JE, Rupert MG, et al. (2010) Predicting the probability and volume of postwildfire
debris flows in the intermountain western United
States. Geological Society of America Bulletin 122:
127144.
Carrasco RM, Pedraza J, Martin-Duque JF, et al. (2003)
Hazard zoning for landslides connected to torrential
floods in the Jerte Valley (Spain) by using GIS techniques. Natural Hazards 30: 361381.
Chen GW (2002) Study on quasi-period of breaking out
debris flow under storm. Journal of Natural Disasters
4: 4954 (in Chinese).
Chen NS, Cui P, Liu ZG, et al. (2003) Calculation of the
debris flow concentration based on clay content. Science in China (Series E), Technological Sciences 46
(Supp.): 164174.
Cui P, Chen XQ, Zhu YY, et al. (2011a) The Wenchuan Earthquake (May 12, 2008), Sichuan Province,
China, and resulting geohazards. Natural Hazards
56: 1936.
Cui P, He YP, and Chen J (2006) Debris flow sediment
transportation and its effect on rivers in mountain areas.
Journal of Mountain Science 24: 539549 (in Chinese).
Cui P, He SM, Yao LK, et al. (2011b) The Formation
Mechanism and Risk Management for Geo-hazards in
Wenchuan Earthquake Affected Area. Beijing: Science Publications (in Chinese).
Cui P, Hu KH, Zhang JQ, et al. (2011c) Prediction of the
debris flow danger area by combing hydrological and

541
inundation simulation methods. Journal of Mountain
Science 8: 19.
Cui P, Zhuang JQ, Chen XC, et al. (2010) Characteristics
and countermeasures of debris flow in Wenchuan area
after the earthquake. Journal of Sichuan University
(Engineering Science Edition) 42: 1019 (in Chinese).
David P (2001) MapScript: A map algebra programming
language incorporating neighborhood analysis.
GeoInformatica 5: 145163.
De Pippo T, Donadio C, Pennetta M, et al. (2008) Coastal
hazard assessment and mapping in Northern Campania,
Italy. Geomorphology 97: 451466.
Fuchs S, Heiss K, and Hubl J (2007). Towards an empirical
vulnerability function for use in debris flow risk
assessment. Natural Hazards and Earth System
Sciences 7: 495506.
Fuchs S, Ornetsmuller C, and Totschnig R (2012). Spatial
scan statistics in vulnerability assessment an application to mountain hazards. Natural Hazards 64:
21292151.
Gentile F, Bisantino T, and Liuzzi GT (2008) Debris flow
risk analysis in South Gargano watersheds (SouthernItaly). Natural Hazards 44: 117.
Granger K, Jones T, Leiba M, et al. (1999) Community Risk
in Cairns: A Multi-Hazards Risk Assessment. Canberra:
Australian Geological Survey Organisation (AGSO).
Gret-Regamey A and Straub D (2006) Spatially explicit
avalanche risk assessment linking Bayesian networks
to a GIS. Natural Hazards Earth System Science 6:
911926.
Hu KH, Cui P, Wang CC, et al. (2010) Characteristic
rainfall for warning of debris flows. Journal of Mountain Science 7: 207214.
Hu KH and Wei FQ (2005) Numerical-simulation-based
debris flow risk zoning. Journal of Natural Disasters
14: 1014 (in Chinese).
Hufschmidt G (2011) A comparative analysis of several
vulnerability concepts. Natural Hazards 58: 621643.
Huggel C, Kaab A, Haeberli W, et al. (2003) Regionalscale GIS-models for assessment of hazards from glacier lake outbursts: Evaluation and application in the
Swiss Alps. Natural Hazards and Earth System
Sciences 3: 647662.
Hurlimann M, Copons R, and Altimir J (2006) Detailed
debris flow hazard assessment in Andorra: a multidisciplinary approach. Geomorphology 78: 359372.
Jakob M and Hungr O (2005) Debris-Flow Hazards and
Related Phenomena. Berlin: Springer.

542
Kappes M, Keiler M, and Glade T (2010) From single- to
multi-hazard risk analyses: A concept addressing emerging challenges. In: Malet JP, Glade T, and Casagli N
(eds) Mountain Risks: Bringing Science to Society.
Strasbourg: CERG Editions, 351356.
Kienholz H (1999) Anmerkungen zur Beurteilung von
Naturgefahren in den Alpen. In: Fischer K (Hrsg.)
Massenbewegungen und Massentransporte in den
Alpen als Gefahrenpotential. Relief, Boden, Palaoklima 14: 165184.
Li W (2006) Handbook of Hydraulic Calculations. Beijing: China Water Power Press (in Chinese).
Liu JF, You Y, Chen XZ, et al. (2010) Identification of
potential sites of debris flows in the upper Min River
drainage, following environmental changes caused by
the Wenchuan earthquake. Journal of Mountain Science 7: 255263.
Liu XL and Mo DW (2003) Debris Flow Risk Assessment.
Chengdu: Sichuan Science and Technology Publications (in Chinese).
Mark RK and Ellen SD (1995) Statistical and simulation
models for mapping debris-flow hazard. In: Carrara A and
Guzzetti F (eds) Geographical Information Systems in
Assessing Natural Hazards. Amsterdam: Kluwer, 93106.
Marzocchi W, Garcia-Aristizabal A, Gasparini P, et al.
(2012) Basic principles of multi-risk assessment: A
case study in Italy. Natural Hazards 62: 551573.
Marzocchi W, Mastellone ML, and Di Ruocco A (2009)
PrinciplesofMulti-RiskAssessment:InteractionsAmongst
Natural and Man-Induced Risks. Brussels: European
Commission, Directorate for Research, Environment.
OBrien JS, Julien PY, and Fullerton WT (1993) Twodimensional water flood and mudflow simulation.
Journal of Hydraulic Engineering 119: 244261.
Papathoma-Kohle M, Kappes M, Keiler M, et al. (2011)
Physical vulnerability assessment for Alpine hazards
state of the art and future needs. Natural Hazards 58:
645680.
Papathoma-Kohle M, Keiler M, Totschnig R, et al. (2012)
Improvement of vulnerability curves using data from
extreme events: Debris flow event in South Tyrol.
Natural Hazards 64: 20832105.
Perles-Roselloy MJ and Cantarero Prados F (2010) Problems and challenges in analyzing multiple territorial
risks: Methodological proposals for multi-hazard mapping. Boletn de la Asociacion de Geografos Espanoles
52: 399404.

Progress in Physical Geography 37(4)


Pradhan B (2010) Remote sensing and GIS-based landslide hazard analysis and cross-validation using multivariate logistic regression model on three test areas
in Malaysia. Advances in Space Research 45:
12441256.
Rickenmann D (1999) Empirical relationships for debrisflows. Natural Hazards 19: 4777.
Rickenmann D (2001) Methoden zur Gefahrenbeurteilung
von Murgangen. In: Projekt CADNAV, Etablissement
dune methodologie de mise en euvre des cartes de
dangers naturels du canton de Vaud; 2eme rapport
intermediare. Lausanne, Switzerland: Ecole Polytechnique Federal de Lausanne (in German).
Schoklitsch A (1948) Hydraulic Engineering. China Science Books and Equipment Corporation.
Totschnig R, Sedlacek W, and Fuchs S (2011) A quantitative vulnerability function for fluvial sediment
transport. Natural Hazards 58: 681703.
United Nations Disaster Relief Organization (UNDRO)
(1984) Disaster Prevention and Mitigation: A Compendium of Current Knowledge. Vol. 11, Preparedness
Aspects. New York: UNDRO.
United Nations Department of Humanitarian Affairs
(UNDHA) (1992) Internationally Agreed Glossary of
Basic Terms Related to Disaster Management, DNA/
93/36. Geneva: UNDHA.
Wang GQ, Shao SD, and Fei XJ (1997) Particle model for
alluvial fan formation. In: Chen CL (ed.) Debris Flows
Hazard Mitigation: Mechanics, Prediction, and Assessment. Proceeding of the First International DFHM Conference, San Francisco, 79 August. New York: ASCE,
143152.
Wei FQ, Hu KH, Lopez JL, et al. (2006) Model and
method of debris flow risk zoning based on momentum
analysis. Wuhan University Journal of Natural Sciences 11: 835839.
Wu JS, Tian LQ, Kang ZC, et al. (1993) Debris Flow and
its Comprehensive Control. Beijing: Science Publications (in Chinese).
Xu Q (2010) The 13 August 2010 catastrophic debris flows
in Sichuan province: Characteristics, genetic mechanism and suggestions. Journal of Engineering Geology
18: 596608 (in Chinese).
You Y, Liu JF, and Chen XC (2010) Debris flow and its
characteristics of Subao River in Beichuan County after
512 Wenchuan Earthquake. Journal of Mountain
Science 28: 358366 (in Chinese).

Reproduced with permission of the copyright owner. Further reproduction prohibited without
permission.

You might also like