Nat Hazards (2013) 68:657674

DOI 10.1007/s11069-013-0642-x

ORIGINAL PAPER

Application of a triangular fuzzy AHP approach for flood

risk evaluation and response measures analysis

Xiao-ling Yang Jie-hua Ding Hui Hou

Received: 5 June 2012 / Accepted: 11 March 2013 / Published online: 27 March 2013
! Springer Science+Business Media Dordrecht 2013

Abstract Flood risk evaluation and prediction represents an essential analytic step to
coherently link flood control and disaster mitigation. The paper established a hybrid
evaluation model based on fuzzy analytic hierarchy process (AHP) and triangular fuzzy
number. It comprises flood risk evaluation and prediction to obtain risk factors ranking and
comprehensive flood risk prediction, and then analyzed flood risk response measures. A
case study is proposed entailing a flood risk evaluation and prediction in the Lower
Yangtze River region. The evaluation results showed that the proposed evaluation and
prediction model was capable of adequately representing the actual setting. In addition, a
comparison with the previously described AHP and trapezoidal fuzzy AHP, and experi-
mental results are encouraging, which fully demonstrates the effectiveness and superiority
of the proposed model.

Keywords Flood risk evaluation and prediction ! Risk index ! Triangular fuzzy AHP !
Flood risk response measures

1 Introduction

Floods are the most common type of disaster globally, responsible for almost 53,000 deaths
in the last decade alone (23:1 low- vs. high-income countries) (Alderman et al. 2012).
Flood disaster risk analysis is based on variable fuzzy sets theory. Flood risk evaluation
and prediction approaches can be divided into two broad categories: (1) methods
employing fuzzy logic set theory (Dagoberto et al. 2012) and (2) methods rooted in

X. Yang (&)
College of Science, Hubei University of Technology, Wuhan 430068, China
e-mail: yxling316@yahoo.com.cn

J. Ding
Central Southern Electric Power Design Institute, Wuhan 430071, China

H. Hou
College of Automation, Wuhan University of Technology, Wuhan 430070, China

123
658 Nat Hazards (2013) 68:657674

probability theory, which will encounter some difficulties in practice (Chen and Tseng
2012). This paper aims to find an effective method to establish a flood risk evaluation and
prediction system for flood-prone locations. Based on the safety evaluation and early
warning rating results, proper precautions against the risks and hazards can be made to
prevent and reduce the flood damage.
Several methods have been developed for flood risk evaluation problems, such as
analytic hierarchy process (AHP) (Saaty 1996), fuzzy AHP (Saaty 2004), analytic network
process (ANP) (Metin et al. 2008), and gray method. Fuzzy AHP is a useful tool to deal
with imprecise, uncertain, or ambiguous data and the high nonlinearity and complexity of
hazard systems. Decision makers usually feel more confident to give linguistic variables
rather than expressing their judgments in the form of exact numeric values. Flood risk
evaluation is an intrinsically complex multidimensional process including both quantitative
and qualitative factors which may be uncertain (Li et al. 2012). Therefore, fuzzy AHP is
deemed to be particularly appropriate for flood risk evaluation and prediction. According to
the characteristics of flood risk analysis, this paper used triangular fuzzy AHPbased
model for the importance ranking of risk indexes, comprehensive flood risk prediction, and
risk response measures analysis.

2 Methodology

A hybrid evaluation model is established based on fuzzy AHP and triangular fuzzy number
(TFN). AHP has excellent performance in dealing with interdependent criteria and the
local problems involving both quantitative and qualitative. And fuzzy set theory deals with
sets or categories whose boundaries are blurry or, in other words, fuzzy. The combination
of these two methods can deal with multiple attribute decision-making problems effec-
tively. The following are proposed methodologies for evaluation model.

2.1 Analytic hierarchy process (AHP)

Analytic hierarchy process established by Saaty is a method to solve multiple criteria

decision problems by setting their priorities (Erden and Karaman 2012). As underlined by
Saaty (1996, 2004) and exposed by Fumey, AHP aims to settle the conflict between
practical demand and scientific decision making, and it also aims to find a way to blend
qualitative analysis and quantitative analysis, which makes it an efficient and effective
method under complex contexts, as synthesized in Fig. 1.

Complexity system approach

Compromise multiple objectives

Interdependencies between the systems

A
H Synthesis through the evaluation of a unique criterion
P Hierarchical structure of several levels
Coherence and consistency of judgments
Measure of non-tangible characteristics

Fig. 1 Using the AHP under complex contexts

123
Nat Hazards (2013) 68:657674 659

Fig. 2 AHP pairwise positive 1 v12 v1i v1 j v1n

reciprocal comparison matrices 1 / v 1 v2i v2 j v2 n
12


1 / v1i 1 / v2i 1 vij vin
V( nn ) =


1 / v1 j 1 / v2 j 1 / vij 1 v jn


1 / v1n 1 / v2 n 1 / vin 1 / v jn 1

Decisions made using the AHP occur in two sequential phases (Saaty 1996): the first is
hierarchy design, which involves decomposing the decision problem into a hierarchy of
interrelated decision elements (i.e., goal and evaluation criteria); the second is hierarchy
evaluation, which involves eliciting weights of the criteria and synthesizing these weights
and preferences to determine alternative priorities.
Pairwise comparisons are classically carried out by asking how more valuable an
alternative A is to criterion c than another alternative B. As shown in Fig. 2, pairwise
comparisons constitute the end square matrices, the values of which are between 1/9 and 9
(Hasekiogullar and Ercanoglu 2012).
According to the analysis above, one of the main advantages of the AHP method is the
simple structure. The AHP is designed in a way that represents human mind and nature.
Therefore, AHP provides the possibility of searching and evaluating the cause and effect
relationship between goal, factor, sub-factor, and alternatives using breaking down the
structure of the problem (Pourghasemi et al. 2012). Moreover, the use of AHP does not
involve cumbersome mathematics, and thus, it is easy to understand and it can effectively
handle both qualitative and quantitative data.

2.2 Fuzzy theory

In the traditional AHP method, the scale of pair comparisons among criteria is restricted to
crisp numbers. Therefore, AHP is criticized for its unbalanced scale of judgment and
failure to precisely handle the inherent uncertainty and vagueness in carrying out pairwise
comparisons (Pourghasemi et al. 2012).
Fuzzy set theory, introduced by Zadeh (1965), resembles human reasoning in its use of
approximate information and uncertainty to generate decisions. Good decision-making
models should be able to tolerate vagueness or ambiguity (Lious and Wang 1992). Thus, if
the uncertainty of human decision making is not taken into account, the results from the
models can be misleading. A fuzzy set is a class of objects with a continuum of grades of
membership. Such a set is characterized by a membership function, which assigns to each
object a grade of membership ranging between zero and one. There are two most com-
monly used fuzzy numbers: trapezoidal fuzzy number and TFN. A TFN, which is denoted
simply as (l, m, u), is shown in Fig. 3.
Theorem M1 = (l1, m1, u1) and M2 = (l2, m2, u2) are two TFNs; let V(M2 C M1) = l(d),
and d is the abscissas of intersection points of M1 and M2, while Eq. (1) is tenable.
! l1 &u2
; l1 ' u2
V M2 " M1 ld m2 &u2 &m1 &u1 1
0; others

123
660 Nat Hazards (2013) 68:657674

Fig. 3 Triangular fuzzy number u(x)

M2 M1
1

0 l2 m2 l1 d u2 m1 u1 x

3 Flood risk evaluation model

In this section, the computational steps of the proposed integrated framework of flood risk
evaluation model based on a triangular fuzzy AHP approach are introduced.

3.1 Flood risk indexes

Flood disaster is that the hazard factor acts on the hazard affected body in certain hazard
including environment (Yang et al. 2009). Therefore, the recognition of the risk related
to the occurrence of flood disasters is the first step within a comprehensive risk evalu-
ation process, to find all relevant risk factors of the cause of flooding and loss, to classify
them according to occurrence and characteristics, and to appropriate choices according to
their impact on disaster bodies (Power et al. 2001). According to the water dynamics
numerical simulation (Song et al. 2011), hydrology information data, and GIS statistical
results (Zou et al. 2012), the identified results of the general flood risk indexes are shown
in Table 1.

3.2 Setting up the triangular fuzzy AHP hierarchical structure

In flood risk evaluation and prediction, issues to be addressed include three aspects: the
importance ranking of risk indexes, comprehensive flood risk prediction, and risk
response measures analysis. The ultimate goal of evaluating the ideal comprehensive
flood risk can be achieved, using four evaluation criteria and twelve sub-criteria as
shown in Fig. 4.

Table 1 Flood risk indexes

Flood risk classification Flood risk indexes

Flood hazard
Hazard factors Rainstorm, dam break, typhoon, tsunami
Hazard including Vegetation coverage, drainage density, terrain elevation, proportion of easily
environment flooded farmland
Vulnerability of hazard affected body
Property Population density, residential property, industrial production, agriculture,
characteristics forestry, animal husbandry, and fishing production
Societys bearing Flood control standard, the accuracy of flood dispatching, early warning
capacity mechanism, disaster relief agencies

123
Nat Hazards (2013) 68:657674 661

Ranking of risk factors

(Goal) GOAL

Hazard factors Disaster Property Society bearing

Criteria
(C1) environment (C2) characteristics (C3) capacity (C4)

Rainstorm (C11) Vegetation coverage (C21) Population density (C31) Flood control standard (C41)

Dam break (C12) Drainage density (C22)
Tsunami (C13) Terrain elevation (C23)
Typhoon (C14) Proportion of easily
flooded farmland (C24)
Residential property (C32)
Industrial production (C33)
Agriculture, forestry, animal
husbandry and fishing
production (C34)
Accuracy of flood dispatching (C42)
Sub-Criteria
Early warning mechanism (C43)
Disaster relief agencies (C44)

Fig. 4 Triangular fuzzy AHP hierarchical structure

3.2.1 Flood risk indexes importance ranking

In the triangular fuzzy AHP hierarchical structure of the importance ranking of risk
indexes, the first level is the ultimate goal of evaluating the ideal comprehensive flood risk.
The second level is flood risk classification that is generally divided into four aspects:
hazard factors, disaster environment, property characteristics, and societys bearing
capacity. In practice, the classification should be adjusted according to actual situation
during the flood, such as the geographical position, flood characteristics, climate condition.
The third level is the lowest level of the structure, which includes concrete risk factors of
the flood risk indexes classification. A graphical representation of the above outlined
hierarchical structure is shown in Fig. 4.

3.2.2 Flood risk response measures analysis

Based on the previously outlined analytic steps, it is necessary to analyze risk response
according to significant flood risk factors. In the hierarchical structure as shown in Fig. 5,
the first level is the reduction in a risk factor. The second level includes four aspects:
reduction in the flood risk factors occurrence probability, reduction in the flood risk loss,
uncontrollability, and measure cost. In the evaluation of relief measures, the cost of relief
measures must be considered. Three levels of risk response measures classification of flood
prevention and disaster reduction, the general should be protocoled based on the actual
situation of flood prevention and disaster reduction.

Goal Factors Measure

Strengthen flood embankment (SFE)

Reduction of risk probability (B1) Enhance rainfall prediction accuracy

(ERPA)
Reduction of a Reduction of risk loss (B2)
certain flood Strengthen scheme demonstration of
risk factor (A) Uncontrollability (B3) weather conditions analysis (SSD)

Measure cost (B4) Establish specialized agency of risk

management (ESA)
Strengthen site monitoring (SSM)

Fig. 5 Flood risk response measures

123
662 Nat Hazards (2013) 68:657674

3.3 Comparison matrix and relative weights

Based on the established hierarchy, comparison matrix of each level can be obtained by
multiple comparisons between several elements, and the relative weight of various factors
can be calculated, and finally, the total importance ranking of risk factors can be calculated.
The specific calculation procedures are as follows (Xie and Tang 2010):
Step 1 Construct triangular fuzzy judgment matrix.
To a certain factor (criteria) of k - 1 layer, the paired comparison in all nk factors of the
related k layer, the TFN quantitative said that element aij = (lij, mij, hij) in the fuzzy
judgment matrix A aij nk (nk is a closed interval for mij to median. mij is an integer
between 1 and 9 of comparison and judgment in AHP method. Fuzzy matrix is also
positive and negative reverse matrix as follows:
" #
&1 1 1 1
aij aij ; ; 2
uij mij lij
Step 2$ Calculate%comprehensive degree function.
atij ltij ; mtij ; utij ; i; j 1; 2; . . .; nk ; t 1; 2; . . .; T represent the TFN obtained by
comparison result of factor i and j in layer k for decision maker t.
Comprehensive TFN of layer k is calculated by Eq. (3).
1 $ %
Mijk ) a1ij a2ij ! ! ! aTij 3
T
Thus, comprehensive judgment matrix of factor h of layer k 1 for all the factors of
layer k can be obtained. Then, comprehensive degree function Ski can be calculated by
Eq. (4).
!&1
Xn nk X
X nk
k k k
Si Mij ! Mij ; i 1; 2; . . .; nk 4
j1 i1 j1

Step 3 Hierarchical ranking.

According to Sect. 2.1, Eq. (5) is calculated.
$ % & ' $ %
V Ski " Skj ; i 1; 2; . . .; nk : i 6 j and Pkih Aki min V Ski " Skj ;
5
i; j 1; 2; . . .; nk : i 6 j
Equation (5) represents single ranking of factors in layer k to factor h in layer k - 1, and
Aki represents factor i in layer k.
& '
Pkih Aki is normalized and then Eq. (6) represents single ranking of factors in layer k to
factor h in layer k - 1.
& 'T
Pkh Pk1h ; Pk2h ; . . .; Pknh 6
Step 4 Comprehensive general ranking.
According to Pkh , when h = 1, 2, , nk-1, we can obtain nk 9 nk-1 matrix:
2 k 3
P11 Pk12 ! ! ! Pk1nk&1
& 'T 6 Pk Pk22 ! ! ! Pk2nk&1 7
Pkh Pk1h ; Pk2h ; . . .; Pknh 6 21
4 !!!
7 7
!!! !!! !!! 5
Pknk 1 Pknk 1 ! ! ! Pknk nk&1

123
Nat Hazards (2013) 68:657674 663

& 'T
if W k&1 W1k&1 ; W2k&1 ; . . .; Wnk&1 is ranking weight vector of layer k - 1 to goal, and
k
synthesis general ranking W can be calculated by Eq. (8).
& 'T X
nk&1
W k W1k ; W2k ; . . .; Wnk Pk W k&1 or W k Pkij Wjk&1 ; i 1; 2; . . .; nk 8
j1

4 Case study

4.1 Study area

The aim of the study is to calculate flood risk in a region (shown in Fig. 6) along the
Yangtze River. The river twists and turns within the region, and the actual river length is
about three times the straight-line distance. The water level is higher than the floodplains,
and water is completely contracted by the water levee beam. This section is the most
dangerous in the 6,300-km Yangtze River section. According to a recent survey (year
2000), the population in the region is 30.5 million and the overall value of the private
property is 2.7 billion (Zou et al. 2012). Failure to take appropriate measures will result in
great property losses of local people and will also bring enormous pressure to society after
the flood.
Hydrological and geological conditions in the region are very complex (Xie et al. 2011).
Floods occurrence will face numerous risk factors (uncertainty) effect (Song et al. 2011).
Therefore, comprehensive and accurate evaluation of these risk factors is very important in
flood risk management and disaster prevention and reduction.

4.2 Triangular fuzzy judgment matrix construction

According to the proposed evaluation model discussed in Sect. 3.3 and the hierarchy
structure shown in Fig. 4, pairwise comparisons of the elements in clusters are conducted

Puhe Town

Yangtze River
Douhudi Town
Jiazhuyuan TownYangjiachang Town

Zhakou Town
Mahaokou Town

OuChi Town

Huangshantou Town

0 2.5 5 10 15
Kilometers

Fig. 6 Location of study area

123
664 Nat Hazards (2013) 68:657674

with respect to their relative importance toward their control criterion and the positive
reciprocal matrices are constructed. The relative importance of two elements is rated using
the nine-scale method. A committee of experts with E members is formed to determine the
alternatives and
& 'sets of criteria for the evaluation and to establish the fuzzy judgment
matrix A aij nk (nk as shown in Table 2. Applying Eqs. (15), we can obtain the
& '
intermediate results Mijk ; Ski ; Pkih Aki ; i; j 1; 2; . . .; nk : i 6 j of the model as shown in
Tables 2, 3, 4, 5, and 6.
Applying Eqs. (6) and (7), we can obtain the parameter P21 , the single ranking of factors
in layer 2 to factor h in layer 1.
2 3
1 0:211 1 0:398
6 0:711 1 0:207 0:347 7
Pkh 6 4 0:281 0:445 0:373
7
1 5
0:167 0:625 0:365 0:647
And W1 = (1, 0.225, 0.585, 0123)T is the ranking weight of layer 1 to goal, and then,
synthesis general ranking W2+ can be calculated by Eq. (8) as follows.
2 3 2 0:5173 3
1 0:211 1 0:398
6 0:711 6 0:1164 7
1 0:207 0:347 7 7!6 7
W2+ P2 W1 64 0:281 0:445 0:373 5 6 7
1 4 0:3026 5
0:167 0:625 0:365 0:647 0:0636
2 3
0:5173 0:0246 0:3026 0:0253
6 0:0864 0:1164 0:0626 0:0221 7
6 4 0:1454 0:0518 0:1129 0:0636 5
7

0:3678 0:0728 0:1104 0:0411

At last, after unitarily dealing, we can obtain the normalized synthesis general ranking
W2.
W2 w1; w2; . . .; w16T
0:244; 0:173; 0:068; 0:041; 0:012; 0:055; 0:024; 0:034; 0:143; 0:03; 0:053; 0:052;
0:012; 0:01; 0:03; 0:019T

The result of W2 shows that the flood risk importance degree of some risk indexes is
greater than others, including rainstorm risk and dam break risk in hazard factors. Rain-
storm is the greatest risk, and the next is dam break risk. It is obviously the accurate
reflection of actual situation in the region; rainfall along the Yangtze River coastal is very
high, and the flood is mainly formed by rainstorm; floods occur from May to October in
this region and are most concentrated in August. Therefore, flood control projects should
strengthen the rainstorm forecast and the research on tackling scheme. In addition, the
Yangtze River basin mainly relies on the dam block floods; therefore, dam break is one of
the major flood hazard factors, and it should strengthen the consolidation and maintenance
of flood control dike beam to reduce dam break risk in multiple storm seasons. The detail
of risk ranking is shown in Fig. 7.
There are six risk factors in the middle risk level: risk of drainage density, industrial
production, agriculture, forestry, and animal husbandry, and fishing production is relatively
higher than others. The region which shares the Yangtze River tributary faces greater dam
break risk, and some industrial bases and region of agriculture, forestry, animal husbandry,

123
 Nat Hazards (2013) 68:657674

Table 2 AC judgment matrix

A C1 C2 C3 C4 S P W

C1 (1, 1, 1) (0.57, 1.23, 1.94) (2.45, 3.3, 3.78) (0.37, 1.05, 1.76) (0.167, 0.3, 0.571) 1 0.5173
C2 (0.85, 1.15, 1.93) (1, 1, 1) (0.27, 0.35, 0.44) (0.43, 0.57, 0.68) (0.097, 0.147, 0.273) 0.225 0.1164
C3 (0.54, 1, 1.25) (0.65, 1.89, 1.56) (1, 1, 1) (0.35, 0.42, 0.53) (0.097, 0.206, 0.293) 0.585 0.3026
C4 (2.34, 2.86, 3.27) (0.46, 0.91, 2.26) (1.56, 2.45, 2.86) (1, 1, 1) (0.204, 0.347, 0.633) 0.123 0.0636
665

123
 666

123
Table 3 C1D judgment matrix
C1 C11 C12 C13 C14 S P W

C11 (1, 1, 1) (0.86, 1.17, 1.83) (2.5, 3, 3.5) (0.49, 1.33, 1.83) (0.193, 0.335, 0.543) 1 0.4632
C12 (0.86, 1.17, 1.83) (1, 1, 1) (0.29, 0.33, 0.4) (0.4, 0.5, 0.67) (0.094, 0.146, 0.238) 0.711 0.3293
C13 (0.67, 1, 1.5) (0.58, 1.17, 1.56) (1, 1, 1) (0.32, 0.39, 0.49) (0.102, 0.184, 0.302) 0.281 0.1302
C14 (2.17, 2.67, 3.16) (0.58, 0.83, 2.17) (1.5, 2, 2.5) (1, 1, 1) (0.209, 0.333, 0.588) 0.167 0.0774
Nat Hazards (2013) 68:657674
 Nat Hazards (2013) 68:657674
Table 4 C2D judgment matrix
C2 C21 C22 C23 C24 S P W

C21 (1, 1, 1) (0.48, 1.35, 1.76) (0.67, 1, 1.5) (0.41, 0.56, 0.98) (0.11, 0.198, 0.254) 0.211 0.0925
C22 (0.78, 1.09, 1.78) (1, 1, 1) (2.5, 3, 3.5) (0.52, 1.38, 1.65) (0.189, 0.275, 0.437) 1 0.4384
C23 (0.54, 1, 1.25) (0.33, 0.41, 0.5) (1, 1, 1) (0.51, 0.62, 0.85) (0.078, 0.176, 0.298) 0.445 0.1951
C24 (2.23, 2.98, 3.56) (0.65, 1.27, 2.72) (1.25, 209, 2.75) (1, 1, 1) (0.234, 0.317, 0.504) 0.625 0.274
123

667
 668

123
Table 5 C3D judgment matrix
C3 C31 C32 C33 C34 S P W

C31 (1, 1, 1) (0.59, 1.18, 1.57) (0.67, 1, 1.5) (0.31, 0.38, 0.48) (0.204, 0.354, 0.578) 1 0.5141
C32 (0.75, 1.16, 1.85) (1, 1, 1) (2.5, 3, 3.5) (0.48, 1.31, 1.82) (0.176, 0.256, 0.324) 0.207 0.1064
C33 (0.78, 1.1, 1.7) (0.28, 0.32, 0.41) (1, 1, 1) (0.4, 0.5, 0.67) (0.124, 0.207, 0.335) 0.373 0.1918
C34 (2.16, 2.68, 3.17) (0.59, 0.89, 2.27) (1.35, 2.13, 2.5) (1, 1, 1) (0.187, 0.279, 0.478) 0.365 0.1877
Nat Hazards (2013) 68:657674
 Nat Hazards (2013) 68:657674

Table 6 C4D judgment matrix

C4 C41 C42 C43 C44 S P W

C41 (1, 1, 1) (0.52, 1.12, 1.52) (0.62, 1, 1.52) (0.32, 0.39, 0.49) (0.214, 0.315, 0.554) 0.398 0.1664
C42 (0.87, 1.18, 1.84) (1, 1, 1) (2.5, 3, 3.5) (0.53, 1.35, 1.85) (0.087, 0.136, 0.218) 0.347 0.1451
C43 (0.68, 1, 1.55) (0.27, 0.35, 0.45) (1, 1, 1) (0.45, 0.55, 0.65) (0.184, 0.325, 0.567) 1 0.4181
C44 (2.18, 2.68, 3.18) (0.59, 0.88, 2.19) (1.56, 2, 2.56) (1, 1, 1) (0.102, 0.184, 0.302) 0.647 0.2705
669

123
670 Nat Hazards (2013) 68:657674

Fig. 7 Indicator weights of C11C44

and fishing area will face relatively large losses after the flood. It is necessary to pay
attention to the construction and maintenance of flood control projects in the distribution of
dense river network, industry, and animal husbandry and fishery areas, and pay attention to
typhoons, tsunamis, and other hazard factors. Rainstorm weather system of the Yangtze
River is mainly formed by typhoon storm, avoiding the flood disaster at the same time, and
the proportion of easily flooded farmland may be appropriate to reduce disaster losses.
The low risk levels have eight risk factors: vegetation coverage, terrain elevation, soil
structure, residential property, flood control standard, the accuracy of flood dispatching,
early warning mechanism, and disaster relief agencies. Because these risk factors can
usually be reduced by response measures of sharing and insurance and together with
controllability of the risk factors brought into the evaluation criteria, their relative risk
important degree becomes smaller. This is in accord with the actual situation of flood
control and disaster mitigation.

4.3 Risk response measures analysis

From the risk ranking results, the rainstorm risk in hazard factors has greatest relative
weight of risk factors. It is shown as follows to take rainstorm risk as an example to choice
risk response measures using fuzzy AHP-TFN model.
According to the actual situation, 5 rainstorm risk response measures include
strengthening flood embankment (SFE), strengthening scheme demonstration of weather
conditions analysis (SSD), enhancing rainfall prediction accuracy (ERPA), establishing
specialized agency of risk management (ESA), and strengthening site monitoring (SSM).
From the conclusion as shown in Fig. 8, enhancing the accuracy of rainfall forecasts and
strengthening the program argument of weather conditions analysis is the effective rain-
storm risk reduction measure. In the course of flood occurrence, reinforcing levees,
establishing risk management special agencies, and strengthening on-site monitoring can
be auxiliary measures.
It is certain that the risk reduction measures will be different at different flood levels and
different stages of flood occurrence. It is necessary to study risk response measures

123
Nat Hazards (2013) 68:657674 671

Fig. 8 Indicator weights of risk response measures

combined with the actual situation of the flood control and disaster mitigation in the
specific analysis.

4.4 Flood risk map of the study area

The study area located in the southern part of the Yangtze River is the main part of flood
control system, and the resolution of DEM data in the study area range is 9 m. The DEM
data are mainly used in the high-precision terrain modeling and numerical calculation of
flood evolution as shown in Fig. 9a. The DOM data for 5-m resolution provided a refresh
rate and true reality effect for flood evolution simulation background scene in the case of
the effect assurance as shown in Fig. 9b. At last, in GIS software support, the study
combined data and statistical yearbook provided by the local government, collected rainfall
data, elevation data, land use data, and the social and economic data, etc., and provided
data support for flood risk analysis. And according to the proposed analysis and conclusion
above, the flood risk map illustrated the flood risk value relative size distribution of eight
villages and towns within the territory of the study area as shown in Fig. 9c.

4.5 Comparison of three evaluation models

Taking the same evaluation indexes in the same region along the Yangtze River mentioned
above as the example, as shown in Fig. 10, the flood risks evaluated by AHP (Lin et al.
2009; Yuan et al. 2008), trapezoidal fuzzy AHP (TF-AHP) (Zheng et al. 2012), and fuzzy
AHPTFN model are identified, respectively. It can be analyzed from Fig. 10 that the
complex and interdependent relations of the evaluative indexes are simplified by AHP and
TF-AHP models. However, in fuzzy AHPTFN model, the dependent information and
feedback among the indexes are fused into the super matrix by pair comparison success-
fully and expressed in the comprehensive weights exactly. The comparison results verify
the validity and practicability of the model proposed in the paper.

123
672 Nat Hazards (2013) 68:657674

Fig. 9 Flood risk map of study area. a DEM data. b DOM data. c Flood risk distribution

Fig. 10 The evaluation result comparative analysis of indexes values

5 Discussions and conclusions

In this work, the application of a triangular fuzzy AHP approach based on TFN is
developed to evaluate flood risk and analyze response measures. A hierarchy evaluation
index system is established. Four factors and 16 sub-factors are included in the index
system. The TFNs are adopted to determine the weights of the indexes and evaluate the

123
Nat Hazards (2013) 68:657674 673

work safety performance of the indexes. The fuzzy evaluating vectors are calculated. Then,
flood risk factors ranking, comprehensive flood risk prediction, and flood risk response
measures analysis are determined. The Lower Yangtze River region is then fed into the
proposed model to evaluate the flood risk and analyze the risk response measures.
This methodology, combining the fuzzy AHP and TFN, provides a new scientific
method for flood risk evaluation and makes the evaluation results more reasonable and
comprehensive. As the evaluation index system has been developed, it is believed that the
proposed method provides a more reliable reference and evaluation method for flood risk
management.
Through the application of a triangular fuzzy AHP approach, we analyze the fuzzy risk
of flood disasters and calculate the flood risk quantitatively in the study area by considering
the rainfall data, elevation data, land use data, and the social and economic data, and so on.
It is hoped that the results would provide the government, engineers, analysts, decision
makers, and local authorities with a more suitable and invaluable guidance and overview
on flooding, which is helpful for them to outline the policy and practice of managing flood
risk clearly. Moreover, according to the risk response measures analysis, the questions on
how to respond, treat, and defend when we encounter flooding can be solved, such as the
use of flood warning system, implementation of temporary flood defenses, measures to
protect the valuables, and mobilization of emergency services. Nonetheless, if more
samples can be collected, the results will be more objective and accurate. And the proposed
hybrid evaluation model based on fuzzy AHP and TFN approach will ensure that we can
identify, evaluate quantitatively, control, and mitigate risks associated with human activ-
ities, thus managing future flood risk more effectively. It also can be generalized to other
natural disaster risk analysis.

Acknowledgments The authors would like to thank the National Natural Science Foundation of China
(No. 51107047) and Research project of Hubei Provincial Department of Education under Grant (No.
B20121403).

References

Alderman K, Turner LR, Tong SL (2012) Floods and human health: a systematic review. Environ Int
47:3747
Alvarado-Aguilar D, Jimenez JA, Nicholls RJ (2012) Flood hazard and damage assessment in the Ebro
Delta (NW Mediterranean) to relative sea level rise. Nat Hazards 62:13011321
Chen CW, Tseng CP (2012) Default risk-based probabilistic decision model for risk management and
control. Nat Hazards. doi:10.1007/s11069-012-0183-8
Erden T, Karaman H (2012) Analysis of earthquake parameters to generate hazard maps by integrating AHP
and GIS for Kucukcekmece region. Nat Hazards Earth Syst Sci 12:475483
Hasekiogullar GD, Ercanoglu M (2012) A new approach to use AHP in landslide susceptibility mapping: a
case study at Yenice (Karabuk, NW Turkey). Nat Hazards. doi:10.1007/s11069-012-0218-1
Li KZ, Wu SH, Dai EF, Xu ZC (2012) Flood loss analysis and quantitative risk assessment in China. Nat
Hazards. doi:10.1007/s11069-012-0180-y
Lin HF, Lee HS, Wang DW (2009) Evaluation of factors influencing knowledge sharing based on a fuzzy
AHP approach. J Inf Sci 35(1):2544
Lious TS, Wang MJJ (1992) Ranking fuzzy numbers with integral value. Fuzzy Sets Syst 50:247255
Metin D, Ihsan Y, Mustafa K (2008) A fuzzy analytic network process (ANP) model to identify faulty
behavior risk (FBR) in work system. Saf Sci 46(5):771783
Pourghasemi HR, Pradhan B, Gokceoglu C (2012) Application of fuzzy logic and analytical hierarchy
process (AHP) to landslide susceptibility mapping at Haraz watershed, Iran. Nat Hazards. doi:
10.1007/s11069-012-0217-2

123
674 Nat Hazards (2013) 68:657674

Power DJ, Sohal AS, Rahman S (2001) Critical success factors in agile natural disaster risk management: an
empirical study. Int J Phys Distrib Logist 31(4):247265
Saaty TL (1996) Decision making with dependence and feedback: the analytic network process. RWS
Publication, Pittsburgh
Saaty TL (2004) Decision making-the analytic hierarchy and network processes (AHP/ANP). J Sci Syst Eng
13(3):135
Song LX, Zhou JZ, Li QQ, Yang XL, Zhang YC (2011) An unstructured finite volume model for dam-break
floods with wet/dry fronts over complex topography. Int J Numer Methods Fluids 67(8):960980
Xie F, Tang DS (2010) The application of AHP and fuzzy comprehensive evaluation in harmonious level
measurement between human and water in city. J Comput Inf Syst 6(14):46474655
Xie T, Zhou JZ, Song LX (2011) Dynamic evaluation and implementation of flood loss based on GIS grid
data. Commun Comput Inf Sci 228:558565
Yang XL, Zhou JZ, Ding JH, Deng WP, Zhang YC (2009) Study on evaluation methods of flood disaster
grade. 6th Int Conf Fuzzy Syst Knowl Discov 4:386390
Yuan SH, Liu X, Tu YL, Xue DY (2008) Evaluating supplier performance using DEA and piecewise
triangular fuzzy AHP. J Comput Inf Sci Eng 8(3):03100410310047
Zadeh L (1965) Fuzzy sets. Inf Control 8:338353
Zheng GZ, Zhu N, Tian Z, Chen Y, Sun BH (2012) Application of a trapezoidal fuzzy AHP method for work
safety evaluation and early warning rating of hot and humid environments. Saf Sci 50(2):228239
Zou Q, Zhou JZ, Zhou C, Chen SS, Song LX, Guo J, Liu Y (2012) Flood disaster risk analysis based on
principle of maximum entropy and attribute interval recognition theory. Adv Water Sci 23(3):323334
(in Chinese with English abstract)

123