Flash Flood Risk Assessment Based on FFIA in China
Changzhi Li
*
, Dongya Sun, Liang Guo, Hong Wang, Xiaolei Zhang, Miao Zhang
China Institute of Water Resources and Hydropower Research, Beijing, 100038
Keywords: Flash flood, risk assessment, FFIA, China
Abstract: Reliable flood risk level information is significant to take appropriate strategies and measures for flash flood
management from area to area in China that suffers heavily from flash flood disasters. A nation-wide
project, called Flash Flood Investigation and Assessment (FFIA), was performed during the period of 2013-
2016 for the purpose of a great improvement in flash flood management. Based on the data from FFIA on
hazard, exposure and vulnerability for each watershed in mountainous area, this study performed flash flood
risk assessment by steps of risk index system development, risk assessment model construct, risk
component computation and flash flood risk analysis; the risk index system is consisted of three layers of
general risk layer, component layer and factor layer (mainly from FFIA); and the model for flash flood risk
indicates the overlying effect of hazard, exposure and vulnerability. The main conclusions include: 1) the
outcomes of flash flood risk assessment agree well with the places where flash flood events occurred, 2) the
protected objects at different risk levels are identified on different administrative jurisdiction levels, and 3)
areas with high flash flood risk are highlighted as the Qin-Ba Mountains area, the Wuling-Xuefeng
Mountains area, the Wuyi Mountains area, the Nanling Mountains, the Sichuan Basin and its surrounding
area, the Yun-Gui plateau, the Yanshan-Taihang Moutains, the Loess Plateau, and the Changbai Mountains;
and suggestions were presented for flash flood risk management in these areas according to local conditions
of climate, geography, population and urbanization.
1 INTRODUCTION
Flash floods are highlighted by deep, fast flowing
water which – combined with the short time
available to respond - increases the risk to local
people and property
(Sene, 2013). China suffers
heavily from flash floods due to much covering of
mountain and hilly area, frequent high-intensity and
short-duration storms, and increasing human actions.
The mountainous area covers roughly two thirds
of the land area of China, and the topography is high
in the west and low in the east, taking three level
ladder-like steps from west to east. The first one is
the Tibet Plateau with average elevation over 4,500
meters and bounded by the line of Kunlun-Qilian-
Hengduan mountain ranges. The area in the east of
the line along the Greater Khingan-Taihang-
Wushan-Xuefeng mountain ranges, is the third step,
consisting of vast plains, hills and low mountains
with elevation less than 500 meters. The remaining
is the second step with large basins and plateaus,
and average elevations ranging from 1000 to 2000
meters (See Figure 1).
This topography in China leads much warm
moist air of the Pacific to flow into the south-east
areas but pretty less in the north-west inland areas.
This causes great regional differences in average
annual rainfall, generally, over 1,000 mm in the
south-east areas and less 200 mm in the north-west
inland areas. It is easy for hills or mountains to
obstruct the movement of hot and wet air flow
which makes a great local difference in rainfall
amount. Therefore, as far flash flood event is
concerned, local topography plays significant role in
the formation of abrupt orographic rain with heavy
rainfall on the windward side and little even no
rainfall on the leeward side. Figure 2 presents the
spatial distribution of rainstorm depth with 6-hour-
duration and indicates a significant consistence with
the land framework, which is constituted by the long
and high mountain ranges (See Figure 2).
566
Li, C., Sun, D., Guo, L., Wang, H., Zhang, X. and Zhang, M.
Flash Flood Risk Assessment Based on FFIA in China.
In Proceedings of the International Workshop on Environment and Geoscience (IWEG 2018), pages 566-575
ISBN: 978-989-758-342-1
Copyright © 2018 by SCITEPRESS – Science and Technology Publications, Lda. All rights reserved

Figure 1: The ladder-like pattern of China
topography.
Figure 2: Rainstorm depth with 6-hour-duration in
China.
In addition to this, human actions have been
increasingly activated in mountain and hill areas in
recent years, such as farming, leisure, entertainment,
mining, tourist, and so on, which have put more and
more people and properties to the threat of flash
flood.
There were many records on flash flood events
in China since 1950s, and these events are
highlighted by unexpected occurrence, sporadic and
isolated distribution in large mountain area, and
huge destructive power. Flash flood hazard mostly
occur in isolated or remote communities. Therefore,
flash flood management has become one of the most
challenges in flood management in China.
According to international experiences, one of the
effective strategies on flash flood mitigation is to
practice risk management that can reduce flash flood
risk levels and represent a guidance on strategies and
countermeasures from area to area.
Many current researches regarding flood
disasters consist of pregnant environment, disastrous
factors, exposures and disaster prevention capacity.
Flood risk is the possible consequence among
interactions of hazard, exposure and vulnerability
(
Cheng, 2009) while the earlier concept of risk is
usually the product of losses and possibility
(I
nternational Union of Geological Sciences (IUGS),
1997). WMO/GWP (WMO/GWP, 2007) regarded the
regional flood risk should be determined by
quantizing the hazard, exposure and vulnerability
while Merz and Thieken(
Merz et al., 2004) thought
that the aim of flood hazard appraisal is to estimate
the possible inundated area and intensity of various
scenarios. It is quite difficult to obtain entire data
about total components of the flash flood risk and
many studies focused on respective component, such
as hazards estimation(
Zhang et al., 2000; Zhao, 1996;
Azmeri et al., 2016
), exposure and vulnerability
appraisal, especially, in recent years, more and more
attentions were drawn to vulnerability or resilience
and uncertainty at community level (Papathoma et
al.,2012; Birkmann et al.,2013; Totschnig and
Fuchs, 2013; Jakob et al., 2012; Sanyal and Lu,
2005; Shi et al., 2004). As for methods for risk
analysis, historical approaches (Copien et al., 2008;
D’Agostino, 2013; Greardo et al., 2004) were
frequently used while some studies on flood hazard
assessment focused mainly on small-scale region
with comparative complete methods and techniques,
such as hydrological and hydraulic methods and
tools (Capello et al., 2016; Leticia et al., 2008;
Fuchs et al., 2013). Apel H, et al (
Apel et al., 2009)
discussed how to choice methods to how detailed do
we need to be in risk analysis. At the same time,
more and more information technologies have been
used to make flash flood analysis, such as RS
(Remote Sensing) and GIS (Geographic Information
System) (Solaimani et al., 2005; Sanyal and Lu,
2006; Lepuschitz, 2015).
The impacts of flash floods are so heavy for the
socioeconomic developments and the achievement
of the sustainable development goals that much
attention has been paid to flash flood management.
During the period of 2013-2016, a nation-wide
project named Flash Flood Investigation and
Assessment (FFIA) was performed to improve flash
flood management (the Project). Based on the
fundamental data from FFIA on hazard, exposure
and vulnerability, this study aimed at supporting
decision making on countermeasure for flash flood
management from area to area in China. The risk
conception of WMO/GWP (
WMO/GWP, 2007) was
adopted in this study because the authors regard this
conception presents not only the expression on
components of flash flood risk, but also on macro-
thought of flood risk computation and guidance on
flash flood management. At the same time, the
literature review indicated that most studies
combined exposure and vulnerability as one entity
for risk analysis, but the authors found that
vulnerability is, to some extent, independent on
exposure in the process of data analysis. Therefore,
risk assessment in this study was performed
Flash Flood Risk Assessment Based on FFIA in China
567

according to three risk components: hazard,
exposure and vulnerability.
2 FLASH FLOOD HAZARDS
INVESTIGATION AND
ASSESSMENT (FFIA)
2.1 About the Project
As mentioned above, the nation-wide project FFIA
focuses on flash flood risk reduction. The Project
was implemented in 2,058 counties in China through
the following 3 periods: 1) early preparedness
period, during which many technical documents
were developed, basic data and map prepared,
special software kit developed for data collection
and process for field work, watershed information
extraction from digital elevation model (DEM), and
determination on what data and information to be
further acquired during the next period; 2)
investigation and assessment period, during which
all of the tasks were done at county level, the tasks
during investigation include identification on local
flash flood prone areas and communities threatened
by flash floods, data collection and process on local
hydrology and flash flood events, and field
measurements on the local river transverse and
longitudinal sections; while the tasks during
assessment include computation on design storm-
flood in watersheds, and estimation on the flood
control capacity and rainfall thresholds for flash
flood early warning for riverside communities; and
3) result summarizing period, during which data
recheck and review were conducted at county,
provincial and national levels, respectively; and a
national fundamental database has developed for
flash flood management.
2.2 Outcomes of the Project
Great progresses were made through the Project in
the fundamental data for flash flood management.
All of these information were summarized according
to watershed scales and different administrative
jurisdiction levels (county, province, and nation) for
the purpose of both administrative and technical
high-efficiency. In summary, fundamental
information of 255,382 watersheds and 2,058
counties were included in the national database for
flash flood management. For each watershed and
administrative jurisdiction unit, the following data
were collected: 1) the basic attributes of the
watershed, such as catchment area, channel system,
length and slope of each channel, landuse cover; 2)
flash flood prone area; 3) the number and
distribution of population, houses, household asset,
monitoring and warning devices, and current flood
control capacity of communities threatened by flash
flood; 4) typical water-related structures potentially
causing disaster, such as bridges, culverts, and
weirs; 5) survey data on longitudinal and cross
sections of river channel near riverside communities;
and 6) historical flash flood events. Therefore, a
good foundation has been laid by the Project, and
more and deep understandings on flash flood
disasters were obtained, such as properties of flash
flood environment, hazard, exposure, vulnerability.
3 FLASH FLOOD RISK
ASSESSMENT MODEL
CONSTRUCTION
According to the aim of flash flood risk assessment,
it is feasible to develop a simple and operable
method to compute flash flood risk. The key factors
for risk should be considered in the method that are
of abundant flash flood information, liable to be
obtained, and to be quantified. Obviously, the
outcomes of the Project meet the requirements very
well for choice of key factors. In this study, risk
was regarded as the overlaying effect of hazard (H),
exposure (E), and vulnerability (V). Hazard is
mainly from physical factors, such as short-duration
storm, and steep landform within a watershed;
exposure depends from socioeconomic factors and,
for instance, populations and houses in mountainous
area; the vulnerability depends chiefly on
susceptibility to flash flood, for example, the
material and structure of houses, the capacity on
flash flood monitoring and warning of a community,
and the awareness of local people on flash floods. It
should be pointed out that watershed is the basic
geomorphic entity for flash flood risk assessment in
this study. For this reason, the original values of
each factor were acquired and processed according
to each watershed in mountainous and hilly areas.
IWEG 2018 - International Workshop on Environment and Geoscience
568

Figure 3: Flash flood risk index system.
3.1 Index System Construction
The index system for risk assessment was developed
from three aspects: hazard, exposure, and
vulnerability. The indexes are satisfied with the
following conditions as much as possible: (1) utmost
use of the data from the Project; (2) easy to be
quantified; (3) the independence between factors,
and (4) directly serving flash flood management.
Figure 3 presents the index system that consists
of three layers of general risk layer, component layer
and factor layer. Layer 1 is the general risk (R) that
stands for the overlaying effect of all components of
risk; layer 2 includes three components of risk:
hazard (R
h
), exposure (R
e
) and vulnerability (R
v
), all
of which result from factors of risk; and layer 3
includes the factors corresponding to three
components of risk, respectively.
In this study, great attention was paid to the
characteristics of flash floods, such as short duration
and high intensity rainstorms, high slope of channels
in watersheds with small drainage area, and
population and properties of local people. Moreover,
the main considerations on the choice of factors at
the third layer were as follows.
Hazard (R
h
) refers to the hazardous degree of
flash flood events, chiefly decided by the features of
rainfall and landform. Larger scale and higher
frequency of flash flood events are, possible heavier
loss in the events. The hazard is determined by the
integrated effects of pregnant environment, the
disastrous factors, and disaster prevention capacity.
In this study, the rainstorms with durations of 6
hours (H
r6
) and 3 hours (H
r3
) were selected as
rainfall feature, while flood peak modus (H
lm
) and
time of concentration (H
lt
) as landform feature,
which considered the characteristics of runoff
generation and surface volume in a watershed, from
the point view of hydrology and hydraulics.
Exposure (R
e
) means the population and houses
and household assets threatened by flash flood in a
watershed. Obviously, more population, houses, and
household assets threatened by flash flood are,
higher flash flood risk. The features of spatial and
temporal distribution of population and assets are
the focuses of exposure study. In this study, the
population (E
p
), houses (E
hse
) and household assets
(E
asset
) were chosen to as three indexes to represent
exposure. The household assets were simply
estimated as the magnification of the number of
households in mountain and hill area in the process
of FFIA to estimate the possible losses due to flash
flood.
Vulnerability (R
v
) is the inner attribute of
exposure and represents the fragility of exposures in
same flash flood hazard. Generally, more
vulnerability of exposures is, and higher flash flood
risk. Vulnerability is closely related to the capacity
of exposure of response to flash flood. In this study,
both the ratio of weak houses (V
r
) and covering
scope of single auto- or manual- monitoring station
(V
astn
and V
mstn
) are on half of vulnerability (R
v
). In
Flash Flood risk (R)
Exposure
Vulnerability
Hazard (R
h
)
Storm with duration of 6hr (H
r6
)
Rainfall
Storm with duration of 3hr (H
r3
)
Flood peak modus (H
lm
)
Landform
Time of concentration (H
lt
)
Population in watershed (E
p
)
Population
House in watershed (E
hse
)
House
Household asset type (E
asset
)
Household
asset
Auto-monitoring station
(V
astn
)
Manual-monitoring (V
mstn
)
Monitoring
station
Ratio of weak houses (V
r
)
Ratio of
weak house
Flash Flood Risk Assessment Based on FFIA in China
569

the process of FFIA, the houses in mountain and hill
area were classified as four types and the house’s
capacity against flash flood increases from type IV
to type III, to type II and type I, both type IV and
type III belongs to weak house.
3.2 Model Descriptions
3.2.1 Risk Model
As mentioned above, flash flood risk is the
overlying effect of hazard, exposure and
vulnerability, as expressed by the following
equation:
Risk = H ∩ E ∩ V
(1)
where, R is regional flood risk; H, E and V the
elements of flood risk, hazard, exposure and
vulnerability, respectively.
The risk components of hazard, exposure and
vulnerability are computed as follow:
H=
∑




=


∑




(
∑





′

)
(2)
E=
∑




=


∑




(
∑





′

)
(3)
V=
∑




=


∑




(
∑


′


′

′

′

)
(4)
Where,
H, E, V —components of layer 2: hazard,
exposure and vulnerability;


, 

, 

—factors of layer 3 corresponding to
components of layer 2;
 ,  ,  — numbers of factors of layer 3
corresponding to components of layer 2;

′
, 
′
, 
′
— numbers of factors of layer 3;
 ,  ,  , 
′
— intermediate variables to
summarize;
 — weights of components of layer 2 and
factors of layer 3.
3.2.2 Considerations on Weights
The following three considerations were taken into
account:
Components of layer 2: weights for hazard,
exposure and vulnerability were set as equal, 1/3, for
they are all the components of risk triangle.
Factors of layer 3: as for hazard, more weight set
for rainstorm with short duration that trigger flash
flood, and for exposures, more weight set for
population, and for vulnerability, more weight set
for monitoring station, which is important for
emergency evacuation.
Weight value calibration with flash flood event:
trial-and-error method was used for obtaining
appropriate weight values for each factor. Initial
values were set to each factors for typical areas and
comparison was made between the calculated results
with the places where flash flood events occurred to
reset the weights until a good agreement reached.
3.2.3 Considerations on Thresholds
Considerations on thresholds were performed for
components of layer 2 and risk level of layer 1 as
follows:
Thresholds for components of layer 2: sort
descending all values of the samples, the values at
1/3, and 2/3 of samples were determined as
thresholds for the corresponding to levels of high,
medium, low for hazard (H), exposure (E) and
vulnerability (V) (see Figure 4).
Figure 4: threshold for hazard, exposure and
vulnerability.
Thresholds for risk level of layer 1: thresholds of
3 levels (high, medium, low) were taken in this
study. Hazard, exposure and vulnerability levels
were classified as 3 levels (high, medium, and low)
and developed an overlaying effect of H-E-V Cube
with 27 sub-cubes (see Figure 5). The thresholds
were made according to the overlaying effect of H-
E-V for the corresponding to levels of high,
medium, low for sub-cubes (see table 1).
IWEG 2018 - International Workshop on Environment and Geoscience
570

Table 1: Overlaying effect of H-E-V and flash flood risk classification.
Risk level Number Code of sub-cube
High 7 H1E3V3, H2E3V3, H3E1V3, H3E2V3, H3E3V1, H3E3V2, H3E3V3
Medium 13
H1E2V2, H1E2V3, H1E3V2, H2E1V2, H2E1V3, H2E2V1, H2E2V2, H2E2V3,
H2E3V1, H2E3V2, H3E1V2, H3E2V1, H3E2V2
Low 7 H1E1V1, H1E1V2, H1E1V3, H1E2V1, H1E3V1, H2E1V1, H3E1V1
Table 2: demo data of flood risk index for watershed.
Watershed code
H
r6
/mm
H
r3
/mm
H
lm
/
m
3
/(s·k
m
2
)
H
lt
/h
r
E
p
E
hse
E
asset
/10
3
Yuan
Vr
V
astn
/k
m
2
V
mst
n
/k
m
2
WJB3410F00000000 161 130 0.21 1.33 822 64 5,120 0.28 11 1
WJB32006L0000000 142 115 0.26 1.00 3075 260 20,800 0.51 5 *
WJB3400121Q00000 172 137 0.23 1.33 684 66 5,280 0.58 13 *
W
JB3400123UM0000 166 134 0.17 1.67 506 130 10,400 0.48 24 *
WJB3400127kE0000 165 133 0.16 1.67 821 129 10,320 0.55 31 *
WJB000010111vA00 156 126 0.20 1.33 359 93 7,440 0.89 * 1
WJB31101CA000000 133 109 0.17 1.67 2000 260 20,800 0.50 * 5
WJB3110700000000 133 109 0.24 1.17 2911 78 6,240 0.50 * 9
WJB3400121h00000 169 136 0.23 1.17 1120 71 5,680 0.66 * 5
WJB3400121kED000 170 136 0.16 1.67 1301 126 10,080 0.56 * 3
(*stands for no stations in the watershed)
Table 3: weights of component and factors in the risk index system.
Component Hazard Exposure Vulnerability
Weight 1/3 1/3 1/3
Factor H
r6
H
r3
H
lm
H
lt
E
p
E
hse
E
asset
Vr V
astn
V
mstn
Weight 0.45 0.15 0.25 0.15 0.55 0.35 0.10 0.30 0.35 0.35
Figure 5: Overlaying effect of H-E-V Cube and risk level
threshold.
4 COMPUTATION AND
ANALYSIS ON FLASH FLOOD
RISK
4.1 Data Acquiring and Process
The analysis on flash flood risk level in China was
done based on the model described in section 3.2.
And each computed entity is a watershed-level
element with area equal or less than 200 km
2
. There
were 255,382 watersheds or entity included in the
assessment. Table 2 presents some original values of
sample data of flash flood risk index for watersheds.
4.2 Method and Steps
The risk analysis was performed according to the
following four steps.
Flash Flood Risk Assessment Based on FFIA in China
571

Step 1, index normalization. Table 2 presents
that the 10 indexes are quite different in magnitude
and dimensions, it is necessary to make
normalization before performing flash flood risk
assessment. After normalization, the absolute values
of data of different indexes can be change into
relative values in same magnitude and
dimensionless. The following expression presents
the algorithm of normalization:


∗
=








(5)
where, 

is the value of original data, 

∗
the
normalized value of original data, and 

and


the maximum and minimum of a same index,
respectively.
Step 2, weights set. The initial values of weights
were set referring to expert’s experiences, that is, the
hazard factors of rainstorms with durations of 6
hours (H
r6
) and 3 hours (H
r3
), flood peak modus
(H
lm
) and time of concentration (H
lt
) were set values
of 0.5, 0.1, 0.2, and 0.2; the exposure factors of
population, numbers of houses and household assets
were set of 0.5, 0.4 and 0.1; and the vulnerability
factors of ratio of weak houses (type III- and IV) to
the total houses, covering areas of single auto- or
manual monitoring station were set of 0.3, 0.35 and
0.35. Then, the initial values were modified by trial-
and-error method, using the flash flood events
records in three typical watersheds, the Jinghe River,
the Longhe River and the Yihe River (see Figure 1),
which stands for south area, north area and Loess
Plateau area in China. Table 3 presents the
calibrated weight values of components and factors
in the risk index system.
Step 3, the values of risk components
computation. The contributions of H, E and V were
computed according to the model developed in
section 3.2. The value of flash flood risk can be
computed based on formula (2), (3) and (4) as
follows: first, obtaining the weighted values of each
factor through values of each factor multiplying its
weight; second, summarizing the values of
components of layer 2 (hazard, exposure and
vulnerability); third, multiplying the values of
components of layer 2 and getting the values of flash
flood risk in each computed entity.
Step 4, perform flash flood risk assessment and
risk level classification, namely, the contributions of
H, E, and V were classified as three levels of high,
medium, and low, then made a risk assessment using
the H-E-V Overlaying Cube to obtain the general
risk levels for each watershed (refer to table 1).
5 CONCLUSIONS
The main understandings from this flash flood risk
assessment are as follows:
(1) The consideration on the computed entity and
weight set for risk factors was special and made the
results more creditable in this study. On one hand,
the basic entity for flash flood computation is
watershed that the relationships among various
hazard factors were taken into consideration. Flood
peak modus and time of concentration were selected
as factors for watershed geographic delineation for
hazard component. In fact, the calculation processes
of the two parameters involve the longest distance
from the mouth to the origin of a river, the mean
slope, landuse situation, soil type, vegetation cover,
and average surface slope in the watershed, the
shape of cross section of river channel. Generally,
the hazard component was considered in terms of
hydrology and hydraulics. On the other hand, weight
set was performed by trial-and-error method using
the flash flood events records in three typical
watersheds, the Jinghe River, the Longhe River and
the Yihe River, that made the weights in this
analysis more reasonable. These consideration on
entity and weight set made the results more
creditable.
(2) The third layer factors in risk index system
are highly representative and the approach on risk
analysis are rational in this study. The outcomes of
flash flood risk assessment agree well with the
places where flash flood events occurred. Generally,
there are about 49,000 flash flood events records
since 1950 in China, about 91% of them located
within the high and medium flash flood risk area in
this study. The statistical results in this study
indicated that the densities of flash flood events are
about 19, 12 and 10 per thousand square kilometer
in high, medium and low risk level area,
respectively. In other words, the density in high risk
level area is about twice of that in low risk level area.
Therefore, the results are credible and worth of
reference.
(3) The protected objects at different risk levels
are identified at different scales that is significantly
important for flash flood management from area to
area. In general, the nation-wide areas in high,
medium and low risk level reach 0.46, 1.22, and
2.17 million square kilometers, respectively; and the
populations are 99 million, 184 million and 302
million, severally. These outcomes can be further
refined to each watershed, then to county level, and
to provincial level, which are quite helpful for
IWEG 2018 - International Workshop on Environment and Geoscience
572

appropriate human interventions in flash flood
management at different levels.
(4) The areas with high flash flood risk are
highlighted in main mountainous regions. Figure 6
presents the results of flash flood risk assessment.
Generally, flash flood prone areas with high and
medium risk level concentrate mainly on the
following nine:① the Qin-Ba Mountains area, ②
the Wuling-Xuefeng Mountains area, ③ the Wuyi
Mountains area, ④ the Nanling Mountains area, ⑤
the Sichuan Basin and its surrounding area, ⑥ the
Yun-Gui plateau area, ⑦ the Yanshan-Taihang
Moutains area, ⑧ the Loess Plateau area, and⑨
the Changbai Mountains area (see Figure 6).
Therefore, more attention should be paid to these
areas in flash flood management. Suggestions are
presented for flash flood risk management in these
areas as follow, and Table 4 demonstrates the
characteristics of flash flood and general suggestions
in these areas.
Figure 6: flash flood risk in China.
① the Qin-Ba Mountains area, ② the Wuling-
Xuefeng Mountains area, ③ the Wuyi Mountains
area, ④the Nanling Mountains area, ⑤the Sichuan
Basin and its surrounding area, ⑥ the Yun-Gui
plateau area, ⑦the Yanshan-Taihang Moutains area,
⑧ the Loess Plateau area, and ⑨ the Changbai
Mountains area
Table 4. Characteristics and general suggestions in flash flood prone areas.
No. Area Characteristics Province involved
Suggestions
①
Qin-Ba
Mountains
abundant rainfall, frequent storms, good
vegetation, highly populated, low
urbanization
Shanxi, Gansu, Henan,
Sichuan, Chongqing, and
Hubei
I, II, III, and
IV
②
Wuling-Xuefeng
Mountains
abundant rainfall, frequent storms, good
vegetation, highly populated, low
urbanization
Hunan, Hubei, Chongqing
I, II, III, and
IV
③
Wuyi Mountains
abundant rainfall, frequent storms, good
vegetation, highly populated, high
urbanization
Fujian, Jiangxi
I, II, III, and
IV
④
Nanling
Mountains
abundant rainfall, frequent storms, good
vegetation, highly populated, low
urbanization
Hunan, Jiangxi, Guangdong
and Guangxi
I, II, III, and
IV
⑤
Sichuan Basin
abundant rainfall, frequent storms, highly
populated
Sichuan, Chongqing
I, II, III, and
IV
⑥ Yun-Gui Plateau
abundant rainfall, frequent storms, common
vegetation, highly populated
Yunnan, Guizhou
I, II, III, IV
and V
⑦
Yanshan-Taihang
Moutains
common rainfall, frequent storms, common
vegetation, common populated
Beijing, Shaanxi,Hebei
I, II, III, IV
and V
⑧
Loess Plateau
poor rainfall, frequent storms, common
populated, sediment problem
Shanxi, Shaaxi, Gansu
I, II, III, IV
and V
⑨
Changbai
Mountains
common rainfall, good vegetation, common
populated
Jilin, Liaoning
I, II, III, and
IV
These areas can be classified as the following
five categories according to local conditions of
climate, geography, population and urbanization.
Category I is characterized by abundant rainfall,
frequent storms, good vegetation cover, highly
populated but low urbanized. This category include
the Qin-Ba Mountains area, the Yun-Gui plateau
Flash Flood Risk Assessment Based on FFIA in China
573

area, the Wuling-Xuefeng Mountains area, and the
Nanling Mountains area. The suggestions on
intervention to flash flood hazard mitigation include
macro-scale rainfall monitoring, local rainfall and
water stage monitoring and warning, appropriate
local structural measures, and community-based
awareness and drill.
Category II is particular for abundant rainfall,
frequent storms, good vegetation cover, highly
populated, but highly local urbanized. This category
cover the Wuyi Mountains area, and the Sichuan
Basin and its surrounding area. In these areas, more
attention should be paid to appropriate local
structural measure arrangement in highly urbanized
area besides suggestions on intervention to flash
flood in the areas of Category I.
Category III is highlighted by common rainfall,
frequent storms, and good vegetation cover, highly
populated, but low urbanized. The Changbai
Mountains area belongs to this category area.
Suggestions include macro-scale rainfall monitoring,
local rainfall and water stage monitoring and
warning, appropriate local structural measures, and
community-based awareness and drill.
Category IV is made outstanding by common
rainfall, frequent storms, and common vegetation
cover, highly populated and highly local urbanized.
This area cover the Yanshan-Taihang Moutains area.
In this area, more attention should be paid to
appropriate local structural measure arrangement in
those locally urbanized area, vegetation protection,
and community-based awareness and drill, in
addition to those common suggestions on
intervention to flash flood.
Category V is made special by common rainfall,
frequent storms, very poor vegetation cover, highly
populated but very low urbanized. This area covers
the Loess Plateau area. Strong suggestions for flash
flood risk management in this area include long-
term vegetation restoration, river harnessing, and
appropriate landuse arrangement, community-based
awareness and drill, beyond that common
suggestions in other areas.
ACKNOWLEDGEMENTS
This study is financially supported by “Mechanism
and model on mixing runoff generation from spatial-
temporal changing sources” (No. KY1793-IWHR),
and “China National flash flood prevention and
control (2013-2015, 2016-2018), MWR”.
REFERENCES
Apel H, Aronica G T, Kreibich H and Thieken A H 2009
Flood risk analyses-how detailed do we need to be?
Nat Hazards April 49(1) 79-98
Azmeri, Hadihardaja I K and Vadiya R 2016
Identification of flash flood hazard zones in
mountainous small watershed of Aceh Besar Regency,
Aceh Province, Indonesia The Egyptian Journal of
Remote Sensing and Space Science 19(1) 143-160
Birkmann J, Cardona O D and Carreo M L, et al. 2013
Framing vulnerability, risk and societal responses: the
MOVE framework Natural hazards 67(2) 193-211
Capello M, Cutroneo L and Ferretti G, et al. 2016
Changes in the physical characteristics of the water
column at the mouth of a torrent during an extreme
rainfall event Journal of Hydrology 541 146-157
Cheng X T 2009 Flood and drought risk management
evolving with progress of sciences and technologies
Water Resources and Hydropower Engineering 40(8)
122-125
Copien C, Frank C and Becht M 2008 Natural hazards in
the Bavarian Alps: a historical approach to risk
assessment Natural Hazards 45(2) 173-181
D’Agostino V 2013 Assessment of past torrential events
through historical sources (M)//Dating torrential
processes on fans and cones. Springer Netherlands
131-146
Fuchs S,Keiler M and Sokratov S, et al. 2013
Spatiotemporal dynamics: the need for an innovative
approach in mountain hazard risk management
Natural hazards 68(3) 1217-1241
Greardo B, Michel L and Mariano B,et al. 2004 Use of
Systematic, Palaeoflood and Historical Data for the
Improvement of Flood Risk Estimation. Review of
Scientific Methods. Natural Hazards 31 623-643
International Union of Geological Sciences (IUGS), 1997
Working Group on Landslides, Committee on Risk
Assessment. Quantitative risk assessment for slopes
and landslides - The state of the art. Proceedings of the
Landslide Risk Workshop, IUGS Working Group on
Landslides, Honolulu 3-12
Jakob M, Stein D and Ulmi M 2012 Vulnerability of
buildings to debris flow impact Natural hazards 60(2)
241-261
Lepuschitz E 2015 Geographic information systems in
mountain risk and disaster management Applied
Geography 63 212-219
Leticia B Rodriguez, Pablo A Cello and Carlos A Vionnet,
et al. 2008 Fully conservative coupling of HEC-RAS
with MODFLOW to simulate stream–aquifer
interactions in a drainage basin Journal of Hydrology
353(1) 129-142
Merz B, Kreibich H, Thieken A and Schmidtke R 2004
Estimation uncertainty of direct monetary flood
damage to buildings Nat Hazards Earth Syst Sci 4(1)
153-163
Papathoma Khle M, Totschnig R and Keiler M, et al.
2012 A new vulnerability function for debris flow -
The importance of physical vulnerability assessment
IWEG 2018 - International Workshop on Environment and Geoscience
574

in alpine areas//12
th
Congress Interpraevent
1033~1043
Sanyal Joy and Lu X X 2005 Remote sensing and GIS-
based flood vulnerability assessment of human
settlements: a case study of Gangetic West Bengal,
India. Hydrologial Procese 19 3699-3716
Sanyal Joy and Lu X X 2006 GIS-based flood hazard
mapping at different administrative Scales: A cale
study in Gangetic West Bengal, India. Singapore
Journal of Tropical Geography 27 207-220
Sene K 2013 Flash Floods: Forecasting and Warning DOI
10.1007/978-94-007-5164-4_1, @Springer
Science+Business Media Dordrecht
Shi P J, Wang J G and Zhou J H, et al. 2004 Integrated
risk management of flood disaster in China: To
balance flood disaster magnitude and vulnerability in
metropolitan regions Journal of Natural Disasters
13(4) 1-7 (In Chinese)
Solaimani K, Mohammadi H, Ahmadi M Z and
Habibnejad M 2005 Flood occurrence hazard
forecasting based on geographical information system
International Journal of Environmental Science and
Technology 2(3) 253-258
Totschnig R and Fuchs S 2013 Mountain torrents:
quantifying vulnerability and assessing uncertainties
Engineering Geology 155 31-44
WMO/GWP, 2007: Guidance on Flash Flood
Management-Recent Experiences from Central and
Eastern Europe. December 20-23
Zhang X N, Luo J and Chen L, et al. Zoning of Chinese
flood hazard risk Journal of Hydraulic Engineering
2000(3) 3-9(In Chinese)
Zhao S P 1996 An elementry study on whole
characteristics of mountain torrents disaster system in
China and its hazard regionalization Jounral of
Natural Disasters 3 95-101(In Chinese)
Flash Flood Risk Assessment Based on FFIA in China
575