Flood Risk Assessment in Urban Areas of Southern Taiwan: Sustainability

sustainability
Article
Flood Risk Assessment in Urban Areas of Southern Taiwan
Wen-Cheng Liu * , Tien-Hsiang Hsieh and Hong-Ming Liu
Department of Civil and Disaster Prevention Engineering, National United University, Miao-Li 36063, Taiwan;
leo2601673@gmail.com (T.-H.H.); dslhmd@gmail.com (H.-M.L.)
* Correspondence: wcliu@nuu.edu.tw; Tel.: +886-37-382357
Abstract: A flood risk assessment of urban areas in Kaohsiung city along the Dianbao River was
performed based on flood hazards and social vulnerability. In terms of hazard analysis, a rainfall-
runoff model (HEC-HMS) was adopted to simulate discharges in the watershed, and the simulated
discharges were utilized as inputs for the inundation model (FLO-2D). Comparisons between the
observed and simulated discharges at the Wulilin Bridge flow station during Typhoon Kongrey
(2013) and Typhoon Megi (2016) were used for the HEC-HMS model calibration and validation,
respectively. The observed water levels at the Changrun Bridge station during Typhoon Kongrey
and Typhoon Megi were utilized for the FLO-2D model calibration and validation, respectively.
The results indicated that the simulated discharges and water levels reasonably reproduced the
observations. The validated model was then applied to predict the inundation depths and extents
under 50-, 100-, and 200-year rainfall return periods to form hazard maps. For social vulnerability,
the fuzzy Delphi method and the analytic hierarchy process were employed to select the main factors
affecting social vulnerability and to yield the weight of each social vulnerability factor. Subsequently,
a social vulnerability map was built. A risk map was developed that compiled both flood hazards

and social vulnerability levels. Based on the risk map, flood mitigation strategies with structural and
nonstructural measures were proposed for consideration by decision-makers.
Citation: Liu, W.-C.; Hsieh, T.-H.;
Liu, H.-M. Flood Risk Assessment in Keywords: flood risk; hazard; social vulnerability; hydrological and hydrodynamic model; analytic
Urban Areas of Southern Taiwan. hierarchy process; fuzzy Delphi
Sustainability 2021, 13, 3180.
https://doi.org/10.3390/su13063180
Academic Editors: Marc A. Rosen 1. Introduction

and George D. Bathrellos
In recent years, the economy has developed rapidly in various parts of the world, and
countries have used many fuels, such as oil and coal, that cause holes in the ozone layer
Received: 31 January 2021
and produce a large amount of greenhouse gases such as carbon dioxide and methane,
Accepted: 12 March 2021
resulting in a continuous rise in temperature. Droughts, floods, and other extreme weather
Published: 14 March 2021
events occur continuously, and extreme weather causes natural disasters that continue to
threaten human lives and cause severe property losses [1–5]. According to statistics from
Publisher’s Note: MDPI stays neutral
the United Nations Disaster Database (www.emdat.be, (accessed on 30 January 2021)),
with regard to jurisdictional claims in
published maps and institutional affil-
since 1950, the number of major natural disasters has increased year by year, and after 1990,
iations.
it has exceeded 200 annually. The total number of disasters that occurred in 2018 was 282.
In the seven continents in the world, the number of disasters that occurred in Asia in 2018
was 129, accounting for 46% of the total. Among the types of natural disasters that occurred
in 2018, flood-related disasters accounted for 74% of all disasters; floods made up 39% of
the total number of disasters, storms made up 30%, and slope disasters made up 5%. The
Copyright: © 2021 by the authors.
rest of the disaster types account for approximately 2% to 9% each. Therefore, quantifying
Licensee MDPI, Basel, Switzerland.
flood risk is an important way to develop adaptation and management strategies for
This article is an open access article
distributed under the terms and
decision-making [6,7]. In general, flood risk can be defined as the product of flood hazard
conditions of the Creative Commons
and vulnerability [8–10].
Attribution (CC BY) license (https://
Based on a literature review, several approaches have been utilized to perform flood
creativecommons.org/licenses/by/ risk assessments. These methods include historical disaster flood risk assessments [11–17],
4.0/). index systems for flood risk assessments [18–23], scenario simulation approaches [24–30],
Sustainability 2021, 13, 3180. https://doi.org/10.3390/su13063180 https://www.mdpi.com/journal/sustainability

Sustainability 2021, 13, 3180 2 of 22
and remote sensing and geographic information systems [31,32]. Different approaches
display corresponding advantages and disadvantages [26]. Cai et al. [26] employed a
one-dimensional pipe network and a two-dimensional surface-coupled model and a multi-
index fuzzy comprehensive evaluation model to build flood disaster risk. Wang et al. [23]
utilized a fuzzy synthetic evaluation method based on combined weight to generate flood
risk maps in the Beijing-Tianjin-Hebi urban area. Geng et al. [33] proposed a coupled one-
and two-dimensional hydrodynamic model to simulate inundation as a hazard indicator,
and social and economic characteristics were analyzed as vulnerability indicators. The
analytic hierarchy process (AHP) was used to compute the weights of the assessment
indices. Afterwards, a fuzzy comprehensive evaluation approach was adopted to assess
urban flood risk. However, these studies did not clearly describe that how the social
vulnerability was selected for further analysis.
Taiwan is located on the western edge of the Pacific Ocean, surrounded by the sea; it
is an island country. The river slopes are steep and winding. It is often exposed to extreme
rains and typhoons during the rainy season and in summer and autumn. In the wet season,
southwesterly winds and typhoons bring frequent heavy rains, resulting in failures of
regional drainage systems in the middle and lower reaches of rivers; these failures cause
levees to break or overflow, resulting in flooding and threatening the lives of the people in
the middle and lower reaches of the rivers. The property and economic losses associated
with these floods are serious. In the future, rainfall patterns will change as a result of
climate change. There may be more rainfall during the wet season, and flooding situations
in urban regions and low-lying areas will become more serious. Therefore, the government
must take preventive measures with flood control projects such as constructing urban
drainage systems, riverbanks, and rainwater sewers as early as possible to ensure the safety
of people and property.
In addition to establishing flood warning systems and improving existing drainage
systems, governments are now gradually moving toward flood prevention and disaster
mitigation planning; these goals are realized through the formulation of laws and regula-
tions, the implementation of disaster preventative measures, and mitigation education and
training aimed at achieving effective risk reduction. Flood risk assessments are composed
of hazards and vulnerability. Among these, hazards are associated with occurrences of
natural disasters that humans cannot eliminate or change through existing technology. To
effectively reduce risk, we must decrease the level of vulnerability.
The objective of the present study is to utilize the hydrological model HEC-HMS and
the inundation model FLO-2D to explore the areas that are prone to flooding in the urban
region of the Dianbao River watershed in southern Taiwan and draw a flood potential
map using the degree of hazard. The fuzzy Delphi method and the hierarchical analysis
approach were utilized to analyze and calculate the degree of vulnerability. By combining
the degree of hazard and the degree of vulnerability, an urban area flood disaster risk map
was produced. Based on the risk map, adaptation strategies were proposed for disaster
prevention and mitigation planning.
The hypotheses of this study include that (1) the minimum unit for flood risk assess-
ment is village. It means that flood risk level in a village is same; (2) no engineering works
is executed to reduce the flood hazard; and (3) the social vulnerability for different rainfall
return periods is also same.
2. Data Collection
This study collected rainfall data at eight rain gauge stations, flow data at the Wulilin
Bridge flow station, and water level data at the Changruu water level station (see Figure 1a).
The collection period of the rainfall, flow, and water level data was from August 2012 to
March 2018.
of this pond is approximately 425,000 m3, while the area of the flood detention pond in
Area B is 42 hectares and its flood detention volume is approximately 1.05 million m3. In
this study, the data of rainwater sewers and flood detention ponds in the middle and
lower reaches of the Dianbao River were taken from the Taiwan Experimental Watershed
Sustainability 2021, 13, 3180 Information Platform (Figure 1d). 3 of 22
Apart from the data collection used for the rainfall-runoff model and inundation
model, data related to social vulnerability were also collected for further analysis.
Figure 1. (a) The study area in the Dianbao River watershed and the urban region, (b) land use classification, (c) elevation,
and (d) rainwater
Figure sewerarea
1. (a) The study system andDianbao
in the detention ponds.
River watershed and the urban region, (b) land use classification, (c) elevation,
and (d) rainwater sewer system and detention ponds.
The land use data were obtained from the Taiwan Experimental Watershed Infor-
mation Platform established by the Taiwan Typhoon Flood Research Institute and the
Water Resources Agency. The land use types can be classified as agriculture, forest, traffic,
hydraulic, construction, public, recreation, mineral, or others (Figure 1b).
For the elevation data of the Dianbao River watershed, 20-m DTM data released from
the information disclosure platform of the Ministry of the Interior were adopted (Figure 1c).
A model was utilized to generate grids using DTM data to obtain the numerical elevation
of the terrain, which was further used to determine the accuracy of the model simulations.
The cross-sectional river data were collected from the Taiwan Experimental Watershed
Information Platform.
In urban areas, cities are frequently flooded due to heavy rainfall events with short
durations and the rapid collection of water resulting from urban development. Therefore,
the construction of rainwater sewer systems and flood detention ponds would help urban
areas drain excessive rainwater quickly. In the Dianbao River watershed, there are flood
detention ponds in zones A and B, located on both banks of the Daliao drainage area. The
area of the flood detention pond in zone A is 17 hectares, and the flood detention volume
of this pond is approximately 425,000 m3 , while the area of the flood detention pond in
Area B is 42 hectares and its flood detention volume is approximately 1.05 million m3 .
In this study, the data of rainwater sewers and flood detention ponds in the middle and
lower reaches of the Dianbao River were taken from the Taiwan Experimental Watershed
Information Platform (Figure 1d).
Apart from the data collection used for the rainfall-runoff model and inundation
model, data related to social vulnerability were also collected for further analysis.
3. Materials and Methods

3. Materials and Methods
3.1. Study Area
3.1. Study Area
The boundaries of the Dianbao River watershed in Kaohsiung city border the
The boundaries of the Dianbao River watershed in Kaohsiung city border the Agong-
Agongdian River watershed in the north, the Houjin River drainage basin in the south,
dian River watershed in the north, the Houjin River drainage basin in the south, the hilly
the hilly area of the Central Mountain Range in the east, and Mituo District in the west.
area of the Central Mountain Range in the east, and Mituo District in the west. The basin
The basin originates from the top of Wushan Mountain in Yanchao District, Kaohsiung
originates from the top of Wushan Mountain in Yanchao District, Kaohsiung city. The
city. The elevation is approximately 320 m, and the river flows westward through
elevation is approximately 320 m, and the river flows westward through Yanchao District,
Yanchao District, Dashe District, Nanzi District, Qiaotou District and Gangshan District
Dashe District, Nanzi District, Qiaotou District and Gangshan District and finally into the
and finally into the Taiwan Strait on the south side of the Oziliao fishing port in Ziguan
Taiwan Strait on the south side of the Oziliao fishing port in Ziguan District [34]. Most
ofDistrict [34]. Most
the Dianbao Riverofwatershed
the Dianbao Riverofwatershed
consists plains. Theconsists of plains.
main roads Thethrough
passing main roadsthe
area are high-speed highways, longitudinal railways, other highways and roads,highways
passing through the area are high-speed highways, longitudinal railways, other such as
and roads,
county roads.such as county
The road networkroads. Thedense,
is quite road and
network is quite
the traffic dense, and
is convenient forthe traffic
users. Thisis
convenient
study uses thefor users.High
Taiwan This Speed
study Rail
usesasthe Taiwan High
a boundary Speedrainfall-runoff
to simulate Rail as a boundary
with theto
simulate rainfall-runoff with the HEC-HMS model in the upstream watershed
HEC-HMS model in the upstream watershed area in the east and to simulate the flooding area in the
east and to simulate the flooding area
area with the FLO-2D in the west (Figure 1a). with the FLO-2D in the west (Figure 1a).
Theaverage
The averageannual
annualrainfall
rainfallininthe
thewatershed
watershedofofthe
theDianbao
DianbaoRiver
Riverisisapproximately
approximately
1825 mm [35]. The wet season is mainly from June to August, and
1825 mm [35]. The wet season is mainly from June to August, and the dry season the dry seasonisisfrom
from
Octoberofofaagiven
October givenyear
yeartotoMarch
Marchofofthe thefollowing
followingyear.
year.
3.2.Methodological
3.2. MethodologicalOutline
Outline
Theprocedure
The procedure followed
followed in thisinresearch
this research includes
includes two parts: two parts:
hazards hazards and
and vulnerability.
Before starting Before
vulnerability. the research,
startingfundamental
the research,data, includingdata,
fundamental terrain and land
including classification,
terrain and land
urban drainage networks,
classification, basin geography,
urban drainage networks, meteorological and hydrological
basin geography, observations,
meteorological and
population,
hydrological social economy, population,
observations, rescue equipment, and special
social economy, agency
rescue data, wereand
equipment, gathered.
special
Inagency
terms data,
of hazards, the research
were gathered. steps were
In terms to establish
of hazards, the hydrological
the research steps wereand inundation
to establish the
models, perform model calibration and validation, predict the flooding
hydrological and inundation models, perform model calibration and validation, predict extents and depths
under differentextents
the flooding rainfall and
return periods,
depths and different
under finally complete
rainfallthe hazard
return map. In
periods, terms
and of
finally
social vulnerability,
complete the hazardthe map.research
In termssteps were vulnerability,
of social to select socialthevulnerability
research stepsfactors,
were design
to select
questionnaires, applyfactors,
social vulnerability the fuzzy Delphi
design method to determine
questionnaires, apply thethe mainDelphi
fuzzy factorsmethod
affectingto
social vulnerability, use the AHP to determine the weights of the main social
determine the main factors affecting social vulnerability, use the AHP to determine the vulnerability
factors,
weights calculate the social
of the main socialvulnerability
vulnerability index (SVI),
factors, and finally
calculate the establish a social vulnera-
social vulnerability index
bility
(SVI), and finally establish a social vulnerability map. According to these twooutline
map. According to these two maps, the risk map was built. The research maps, theis
illustrated in Figure 2.
risk map was built. The research outline is illustrated in Figure 2.
Figure2.2.Outline
Figure Outlineofofthe
theflood
floodrisk
riskassessment
assessmentmethodology.
methodology.
3.3. Hazard Assessment with Hydrological and Hydrodynamic Models

In recent years, the impacts of climate change have caused extreme rainfall events,
which have resulted in urban areas being unable to discharge internal waters, leading to
flooding. This study uses the hydrological model HEC-HMS and the inundation model
FLO-2D to simulate an inundation map of urban areas, serving as the hazard map.
3.3.1. Hydrological Model (HEC-HMS)

The rainfall-runoff model used in this study, HEC-HMS software, was developed by
the Hydrological Engineering Center of US Army Corps Engineers. The HEC-HMS model,
which was modified from HEC-1, is a new version used for hydrological simulations [36].
In this study, HEC-HMS version 4.3 was employed to calculate the rainfall-runoff discharge
in the watershed [36]. The model was designed to be applied to different geographic areas,
such as in studies of river basin flood hydrology and large and small watershed/urban
runoff [37–43]. The HEC-HMS includes three modules: the basin model, meteorological
model, and control specification. In the basin model, the US Soil Conservation Service
(SCS) curve number (CN) was used to calculate precipitation loss. The input data include
the initial loss, CN, and impermeability. The HEC-GeoHMS toolbox can be adopted to
calculate the average CN for each subwatershed. The SCS synthetic unit hydrograph
method was applied to convert rainfall to runoff discharge. Two parameters, the time of
concentration (TC ) and storage coefficient (R), were used in the Clark unit hydrograph
method. The time of concentration (TC ) and the regional value (RV), which is related to the
time of concentration and the storage coefficient, are expressed as follows [44,45]:
0.7
L0.8 1000
TC = 2.27 × 10−4 ( − 9 ) (1)
S0.5 CN
R
RV = (2)
R + TC
where L denotes the length of the main channel (m) and S expresses the slope of the
watershed (%).
The time lag (Tlag ) then can be determined as follows:
Tlag = 0.6 ∗ TC (3)
The recession base flow method, which uses an exponentially decayed algorithm, was
employed to separate the base flow. In a rainfall event, the recession base flow strongly
affects the volume and timing of the base flow. In this study, the recession constant and
ratio to the peak were set in the ranges of 0.2–0.35 and 0.01–0.1, respectively.
The rainfall method was used to establish a meteorological model in this study. In the
module, the time-series rainfall data of each subcatchment and the total rainfall of each
subcatchment are given. The main parameters to be set in the control specification are the
start time and end time of the simulation and the time step. In this study, the computational
time step was set to one hour.
3.3.2. Inundation Model (FLO-2D)

FLO-2D model is generally adopted for studies on flood routing. The model is
currently certified by the Federal Emergency Management Agency. It is mainly used in
urban flood simulations, flood plain management, engineering risk designs, and debris flow
simulations. FLO-2D is a combination of a one-dimensional river channel unsteady flow
model and a two-dimensional horizontal overland flow model [46]. It is used to simulate
the depth and velocity of a two-dimensional fluid flow and to estimate the reasonable
disaster range of a fluid. The detailed governing equations can be found in Dimitriadis
et al. [47] and Hu et al. [48].
The FLO-2D model divides the DEM (digital elevation model) into fixed-size square
grids. Each grid can be input with a surface elevation value, Manning roughness coefficient,
The FLO-2D model divides the DEM (digital elevation model) into fixed-size square
grids. Each grid can be input with a surface elevation value, Manning roughness
coefficient,
infiltration infiltration
coefficient, coefficient,
area reduction factor,area reduction
outflow, inflow, factor, outflow,
and other inflow,and
hydrological and other
hydrological and geological data to simulate the real flood flow in
geological data to simulate the real flood flow in the study area. This model starts with the study area. This
model starts with basic two-dimensional surface overland
basic two-dimensional surface overland flow, and many data parameters can be added; flow, and many data
parameters can be added; users can freely choose to turn on or off the additional
users can freely choose to turn on or off the additional functions, including the presence of functions,
including
river channels, theroads,
dikes, presence of urban
streets, river channels, dikes,rainfall,
area reduction, roads, debris
streets,flow,
urban area reduction,
infiltration,
rainfall, debris
and rainwater sewers. flow, infiltration, and rainwater sewers.
3.3.3. Establishment
3.3.3. Establishment of HEC-HMS
of HEC-HMS and FLO-2D
and FLO-2D Models Models
To save To save computational
computational time, thetime, the simulation
simulation domain domain of the FLO-2D
of the FLO-2D model ismodel
smalleris smaller
than
than that thatHEC-HMS
of the of the HEC-HMS model; therefore,
model; therefore, connectingconnecting
these two these two modeling
modeling domainsdomains
is is
necessary. Figure 3a
necessary. illustrates
Figure the connection
3a illustrates of the HEC-HMS
the connection and FLO-2D
of the HEC-HMS models. models.
and FLO-2D It It
shows that the whole domain of the Dianbao River watershed is used
shows that the whole domain of the Dianbao River watershed is used for HEC-HMS, for HEC-HMS, while
the urban areathe
while is adopted
urban area for is
FLO-2D.
adopted for FLO-2D.
Using geomorphological
Using geomorphological analysis with the input
analysis with function
the inputoffunction
the HEC-HMSof the model
HEC-HMSand model
the output
andfunction
the output of HEC-GeoHMS can achieve thecan
function of HEC-GeoHMS purpose
achieveof automatically
the purpose of building
automatically
watersheds. The watersheds.
building Dianbao River Thewatershed
Dianbao Riverwas built for thewas
watershed HEC-HMS
built formodel as shownmodel as
the HEC-HMS
in Figure 3b and includes 7 subcatchment units, 16 channel units,
shown in Figure 3b and includes 7 subcatchment units, 16 channel units, and 16 and 16 confluence
point units.
confluence point units.
To simulate the inundation
To simulate extent and
the inundation depth
extent andofdepth
urbanofareas,
urbana areas,
numerical grid wasgrid was
a numerical
established in the FLO-2D
established model. The
in the FLO-2D DTM
model. Thedata
DTMwere
dataimported into the into
were imported FLO-2D model, model,
the FLO-2D
and using
and the grid the
using generation module,module,
grid generation a grid with a size
a grid with m×
of a20size of20
20m min themhorizontal
× 20 in the horizontal
plane was generated.
plane The newThe
was generated. version
new of FLO-2D
version has beenhas
of FLO-2D enhanced and can and
been enhanced be combined
can be combined
with thewith
EPA-SWMM model. Therefore, the grid system
the EPA-SWMM model. Therefore, the grid system calculations calculations can be used
can beto used to
simulatesimulate
the drainage of rainwater sewer nodes and the overflow of full pipes.
the drainage of rainwater sewer nodes and the overflow of full pipes. Figure 3c Figure 3c
displaysdisplays
the gridthe generation of the FLO-2D
grid generation model, which
of the FLO-2D model,contains a total ofa 142,608
which contains grids. grids.
total of 142,608
Figure 3. Cont.
FigureFigure 3. (a) Modeling

3. (a) Modeling connection
connection for HEC-HMS
for HEC-HMS and FLO-2D
and FLO-2D in thearea,
in the study study(b)
area, (b) setup
model modelfor
setup
for HEC-HMS, and (c) grid established for the FLO-2D model simulations.
HEC-HMS, and (c) grid established for the FLO-2D model simulations.
3.4. Fuzzy
3.4. Fuzzy DelphiDelphi
(FD) (FD)
The Delphi
The Delphi method method
proposedproposed by Dalkey
by Dalkey and Helmer
and Helmer [49] is[49] is a procedural
a procedural method method
used toused to express
express expertexpert
opinionsopinions in a systematic
in a systematic manner. manner. However,
However, the Delphi
the Delphi method method
has has
some some shortcomings;
shortcomings; it is time-consuming,
it is time-consuming, and the and theofcost
cost of collecting
collecting expertexpert
opinions opinions
can can
be high.
be high. The so-called
The so-called consistency
consistency of expert
of expert opinions opinions onlywithin
only falls falls within
a certaina certain
range range
of of
expertexpert opinions
opinions but doesbut not
does not determine
determine the point
the point of impact.
of impact. Therefore,
Therefore, this law thisimplies
law implies
ambiguity,
ambiguity, so the so theresults
final final results
cannotcannot truly reflect
truly reflect the opinions
the opinions of experts.
of experts. To resolve
To resolve these these
shortcomings,
shortcomings,the fuzzy DelphiDelphi
the fuzzy method was developed.
method was developed. Murray et al. [50]
Murray et al.first
[50]combined
first combined
the fuzzy set with
the fuzzy setDelphi, and then
with Delphi, andIshikawa [51] proposed
then Ishikawa the max-min
[51] proposed the max-minmethod and the
method and the
fuzzyfuzzy
integration method to integrate the opinions of experts into fuzzy
integration method to integrate the opinions of experts into fuzzy numbers, which numbers, which
is called the fuzzy
is called Delphi
the fuzzy (FD) method.
Delphi (FD) method.In theIn first
thestage
first of the of
stage expert questionnaire,
the expert the the
questionnaire,
method proposed
method by Jeng
proposed by[52]
Jengwas
[52]adopted
was adoptedto select important
to select socialsocial
important vulnerability factors.
vulnerability factors.
The expert questionnaire
The expert is mainly
questionnaire filled out
is mainly filledbased on theon
out based judgment
the judgment of each ofexpert’s
each expert’s
professional literacy
professional and the
literacy andimportance
the importance of each evaluation
of each factor.
evaluation The The
factor. evaluation
evaluation scale
scale is
is divided into levels from 0–10. The higher the evaluation score is,
divided into levels from 0-10. The higher the evaluation score is, the more representative the more represen-
tativethe
the factor
factor is is of
ofthe theresearch
researchtrend.
trend.AnAn option
optioncolumn
column waswas added at the
added at end
the ofendtheof the
questionnaire. If the experts felt that there were deficiencies, their
questionnaire. If the experts felt that there were deficiencies, their opinion can be opinion can be supple-
mented, making the questionnaire
supplemented, more completemore
making the questionnaire and in line with
complete thein
and meaning
line with of the
the meaning
fuzzy of
Delphi questionnaire. The consensus value can be yielded by triangular
the fuzzy Delphi questionnaire. The consensus value can be yielded by triangular fuzzy fuzzy number
method [50]. The
number results
method using
[50]. The FD to obtain
results usingimportant
FD to obtain social vulnerability
important factors can befactors
social vulnerability
foundcan in Section 4.4.1.
be found in Section 4.4.1.
3.5. Analytic Hierarchy Process (AHP)
3.5. Analytic Hierarchy Process (AHP)
Saaty [53] proposed the analytic hierarchical process (AHP) method. Through the
Saaty [53] proposed the analytic hierarchical process (AHP) method. Through the
establishment of a hierarchical structure with mutual influence, the AHP can be used
establishment of a hierarchical structure with mutual influence, the AHP can be used to
to solve multiple evaluation criteria and decision-making problems associated with un-
solve multiple evaluation criteria and decision-making problems associated with
certainty; through quantitative judgments and comprehensive evaluation, the AHP can
uncertainty; through quantitative judgments and comprehensive evaluation, the AHP can
provide decision-makers with sufficient information to choose appropriate options and
provide decision-makers with sufficient information to choose appropriate options and
reduce the risk of making mistakes. The purpose of the development of the hierarchical
reduce
analysis the risk
method is toofsystematize
making mistakes.
complexThe purpose
issues, of thehierarchical
provide development of the hierarchical
decomposition
analysis method is to systematize complex issues, provide hierarchical
from different levels, organize the levels in a quantitative manner, and comprehensively decomposition
from
evaluate different
them levels,decision-makers
to provide organize the levels
withininformation
a quantitative
withmanner, and comprehensively
which they can choose
evaluate them
appropriate solutions. to provide decision-makers with information with which they can choose
appropriate solutions.
The AHP builds complex multi-objective decision-making problems into hierarchical
systems withThe dendritic
AHP builds complex multi-objective
structures; decision-making
each layer is composed problems
of different into hierarchical
elements. The
systems with dendritic structures; each layer is composed of different elements. The
complex problems are changed from high-level problems to low-level problems as they
are gradually decomposed, and each level in the framework only affects one other level
and, at the same time, is only affected by one other level. Before establishing the AHP
framework, one thing must be confirmed: each factor must be independent of the other
factors. The hierarchical structure method can be used to confirm a hierarchical relationship,
but there is no specific construction method in the actual construction process. The AHP
uses the characteristic vector method to calculate the weights of the elements and obtains
the priority value of each scheme. The larger the value is, the higher the priority of the
adopted scheme is.
Therefore, this study adopts the AHP method to design expert questionnaires, which
are completed by experts in related fields, and the weights of each vulnerability factor are
determined based on the results of the questionnaire. The results using AHP to obtain the
weight of social vulnerability factors can be found in Section 4.4.2.
3.6. Social Vulnerability Index (SVI) and Normalization

This study uses the FD method to select and screen out the vulnerability factors and
then uses the AHP method to determine the weight of each vulnerability factor. In terms of
calculating vulnerability, this research refers to the social vulnerability index proposed by
Shaw et al. [54] to identify the vulnerability of a certain area. Then, the vulnerability data
are standardized based on this index. Since the values of the variables are different and
the units are also different, to compare different variables with each other, it is necessary
to standardize all the statistical data obtained before the vulnerability index is calculated.
The index is calculated in an average manner. The standardized formula is expressed
as follows:
n
∑ ( xi − M )
z − score = Z = α i=1 (4)
SD
v
( x − M )2
u n
SD = t ∑ i
u
(5)
i =1
n−1
n
where x denotes the statistics of different variables, M represents the mean value (= n1 ∑ xi ),
i =1
n expresses the total number of variables, SD indicates the standard deviation, and α
is the correction factor of each vulnerability factor (=+1 or −1). The correction factor
(α = −1) decreases the level of social vulnerability, representing a positive effect on
social vulnerability, whereas the corrector factor (α = +1) increases the level of social
vulnerability, denoting a negative effect on social vulnerability. The social vulnerability
index (SVI) can be expressed as follows:
N
SV I = ∑ k j Zj (6)
j =1
where k is the weight value, Z denotes the standardized value, and N indicates the total
number of vulnerability factors.
The next step is to normalize the SVI value and convert it to a value between 0 and
1 (=ISVI ). When the ISVI value is closer to 1, the vulnerability is higher; in contrast, when
the ISVI value is closer to 0, the vulnerability is lower. The normalization formula can be
represented as follows:
SV I − SV Im
ISV I = (7)
D
where SVIm represents the minimum SVI value and D denotes the full range of SVI values,
which is the maximum SVI value minus the minimum SVI value.
After calculating the ISVI value and dividing the social vulnerability into 5 grades,
including extremely low, low, medium, high, and extremely high (see Table 1) based on
every 0.2 increment [55], ArcGIS software was utilized to draw the social vulnerability map.
Table 1. Grades and levels of the social vulnerability index (SVI) normalization (ISVI ) values, inundation depths, and
risk values.
Degree of Social Vulnerability

ISVI Value Inundation Depth (m) Risk Value
Vulnerability/Hazard/Risk Level/Hazard Level/Risk Level
Extreme low 1 0.0–0.2 <0.3 1, 2, 3, 4
Low 2 0.2–0.4 0.3–0.5 5, 6, 8, 9
Medium 3 0.4–0.6 0.5–1.0 10, 12
High 4 0.6–0.8 1.0–3.0 15, 16
Extreme high 5 0.8–1.0 >3.0 20, 25
3.7. Hazard Assessment

Urban flooding disasters are mainly caused by heavy rainfall. When rainfall exceeds
the rainfall volume that the drainage system of an area can load, flooding disasters occur,
causing casualties and property losses and indirectly leading to the suspension of commer-
cial activities and classes in schools. In this study, the simulated inundation depth is used
as the hazard assessment index, and the inundation depth is divided into 5 grades: below
0.3 m, 0.3 m to 0.5 m, 0.5 m to 1.0 m, 1.0 m to 3.0 m and above 3.0 m (see Table 1). When the
inundation depth is 0.3 m, the water depth is approximately the height of a child’s knees,
so it is difficult for children to escape. When the inundation depth is 0.5 m, the water depth
is approximately the height of an adult’s knees, which causes adults to have difficulty
walking. When the inundation depth is 1.0 m, the water depth is approximately the height
of the waist of an adult, so it indeed causes difficulties for adults walking. When the
inundation depth is 3.0 m, the water depth is as high as a building is tall, which threatens
human lives.
3.8. Risk Assessment

Different definitions of risk have been documented by various researchers [56,57].
Generally, risk can be defined as the product of hazard (H) and social vulnerability (V) in
Taiwan [58]. The simple equation to describe the risk is expressed as follows:
Risk = H × V (8)
Because hazards and social vulnerability are each graded in five levels, the approach
of a semiquantitative risk assessment can be defined as the risk (Water Resources Agency,
2015). The risk value does not represent what actually happened but only stands for the
relative relationships between districts. Figure 4 depicts the 5 × 5 semiquantitative matrix
used for the risk assessment. Based on the matrix, the classifications of the risk assessment
can also be graded, as shown in Table 1. Different risk values can be attributed to different
degrees of risk.
Sustainability2021,
Sustainability 13,3180
2021,13, 3180 1010ofof24
22
Figure
Figure4.4.55××5 5matrix
matrixfor
forthe
thesemiquantative
semiquantativemap
mapused
usedfor
forthe
therisk
riskassessment.
assessment.
3.9.Indices
3.9. IndicesofofModel
ModelPerformance
Performance
To ascertain
To ascertain the
the simulated
simulated discharge
discharge values values using
using the
the HEC-HMS
HEC-HMS model
model and
and the
the
simulatedwater
simulated waterlevel
levelusing
usingthetheFLO-2D
FLO-2Dmodel, model,threethreeindices,
indices,flood
floodpeak
peakerror
errorpercentage
percentage
(EYpp),),flood
(EY floodpeak
peakarrival
arrivaltime
timeerror
error(ET (ETp),p ),and
andNash-Sutcliffe
Nash-Sutcliffeefficiency
efficiencycoefficient
coefficient(NSE),
(NSE),
were employed for model performance. These indices can
were employed for model performance. These indices can be calculated as follows:be calculated as follows:

(Yp ) sim (−Y(pY)sim
p ) obs
− (Yp )obs (9)
EY p =EYp =
× 100 % × 100% (9)
(Y ) (Yp )obs
p obs

ETp = ( Tp ) − ( Tp ) (10)
sim obs
ET p = (T p ) sim − (T p ) obs (10)
N
2
∑ [Yobs (ti ) − Ysim (ti )]
N

NSE = 1[Y− i= 1
) − Ysim (ti )]2
obs (t iN 2
(11)
NSE = 1 − i=1 ∑ [Yobs (ti ) − Y obs ]
N i =1 (11)
where N denotes the total number
2
[Y (t ) − Y ] obs i obs
of observational/simulation data, (Yp )
i =1 obs
and (Yp )sim
represent the observed and simulated peak discharge/water level, respectively, ( Tp )obs and
where N denotes the total number of observational/simulation data, ( Y p ) obs and ( Y p ) sim
( Tp )sim indicate the observed and simulated peak arrival time for the discharge/water level,
represent theYobserved
respectively, and simulated peak discharge/water level, respectively, ( T plevel
obs ( ti ) and Ysim ( ti ) denote the observed and simulated discharge/water ) obs
N
and (Ttpi), sim
at time respectively, theY obs
indicate and is the average
observed and value of the observations
simulated
1
N ∑ Yfor
peak arrival(=time obs ( ti the
)).
i =1
The NSE value
discharge/water ranges
level, from negative
respectively, Y obs (infinity
t i ) and toY1, and when
sim ( t i )
thethe
denote NSEobserved
is close toand1, it
simulated discharge/water level at time ti, respectively, and Y obs is the average valuethe
indicates that the model is of good quality and that the model has high reliability. If of
NSE value is close to N0, the simulation result is close to the average level of the observed
1
the observations
values; Yobsresult
(= overall
that is, the
N
(ti ) ). is credible, but the error in the simulation process is large.
If the NSE value is fari=1less than 0, the model is not credible.
The NSE value ranges from negative infinity to 1, and when the NSE is close to 1, it
indicates
4. Results that the model is of good quality and that the model has high reliability. If the
NSE value is closethe
To ascertain to 0, the simulation
availability of theresult
models, is close
twoto the average
typhoon level
events, of the observed
Typhoon Kongrey,
values;
which occurred from August 27 to 2 September 2013, and Typhoon Megi, which hitiscentral
that is, the overall result is credible, but the error in the simulation process large.
IfTaiwan
the NSE value is far less than 0, the model is not credible.
in the period of 25–28 September 2016, were adopted for the calibration and
validation of the models, respectively. The validated models, HEC-HMS and FLO-2D,
4.were
Results
then applied to predict inundation extents and depths, and these values were used to
To ascertain
evaluate theindices
the hazard availability of the models,
for different two typhoon events, Typhoon Kongrey,
return periods.
which occurred from August 27 to September 2, 2013, and Typhoon Megi, which hit
central Taiwan in the period of September 25–28, 2016, were adopted for the calibration
and validation of the models, respectively. The validated models, HEC-HMS and FLO-
2D, were then applied to predict inundation extents and depths, and these values were
used to evaluate the hazard indices for different return periods.
4.1. HEC-HMS Model Calibration and Validation

Sustainability 2021, 13, 3180 To ensure the correctness of the rainfall runoff simulated by the HEC-HMS11model, of 22
calibrating and validating the HEC-HMS model are important procedures. The observed
discharges at the Wulilin Bridge flow station were compared with the simulated results.
Figure 5a displays the comparison between the observed and simulated discharges
4.1.
at HEC-HMS
the WulilinModelBridge Calibration
flow stationand during
Validation
Typhoon Kongrey (2013); this comparison was
usedTofor the model
ensure calibration.
the correctness Therainfall
of the figure shows
runoff that the simulated
simulated results reproduced
by the HEC-HMS model,
temporal and
calibrating variations in thethe
validating discharge.
HEC-HMS Twin observed
model and simulated
are important peak discharges
procedures. The observedwere
selected for
discharges at the analysis of statistical
Wulilin Bridge errors.were
flow station The EY p values for
compared with peak
the discharge
simulated 1results.
and peak
discharge
Figure 5a2 were
displays0.6%theand 19.3%, respectively,
comparison between the while the ET
observed p values
and weredischarges
simulated 1 hr and 0athr,
the Wulilin Bridge
respectively. Theflow
NSEstation
value wasduring Typhoon Kongrey (2013); this comparison was used
0.89.
for theAmodel calibration.
comparison The simulated
of the figure shows andthat the simulated
observed resultsduring
discharges reproduced temporal
Typhoon Megi
variations in the
(2016), used fordischarge.
the modelTwin observed
validation, is and
shownsimulated peak
in Figure 5b.discharges
The figurewere selected
indicates thatfor
the
the analysis of
computed statistical
water levelserrors.
faithfully EYp values for
The reproduced thepeak discharge Based
observations. 1 and peak
on thedischarge
analysis2of
were 0.6% and
statistical 19.3%,
errors, therespectively,
EYp and ETpwhilevaluesthe ETp4.6%
were valuesandwere 1 hrespectively.
2 hrs, and 0 h, respectively.
The NSE The
value
NSE
wasvalue
0.91. was 0.89.
Figure 5. Comparison of the simulated and observed discharges at the Wulilin Bridge (a) during
Figure 5. Comparison of the simulated and observed discharges at the Wulilin Bridge (a) during
Typhoon Kongrey (2013), used for the model calibration, and (b) during Typhoon Megi (2016),
Typhoon Kongrey (2013), used for the model calibration, and (b) during Typhoon Megi (2016), used
used for the model validation.
for the model validation.
4.2.AFLO-2D Model Calibration and Validation
comparison of the simulated and observed discharges during Typhoon Megi (2016),
used forThe
theobserved water levels
model validation, at the
is shown Changrun
in Figure Bridge
5b. The figureflow station
indicates during
that Typhoon
the computed
Kongrey (2013) and Typhoon Megi (2016) were used to calibrate and validate the
water levels faithfully reproduced the observations. Based on the analysis of statistical FLO-2D
model,
errors, therespectively.
EYp and ETpAvalues
comparison of the
were 4.6% andsimulated and observed
2 h, respectively. The NSE water
valuelevels during
was 0.91.
Typhoon Kongrey, used for the model calibration, is presented in Figure 6a. The figure
4.2. FLO-2D Model Calibration and Validation
The observed water levels at the Changrun Bridge flow station during Typhoon
Kongrey (2013) and Typhoon Megi (2016) were used to calibrate and validate the FLO-2D
model, respectively. A comparison of the simulated and observed water levels during
Typhoon Kongrey, used for the model calibration, is presented in Figure 6a. The figure
shows twin peaks in the water level as a result of runoff discharge. The statistical errors
represented by the EYp values at peak water level 1 and peak water level 2 were 13.5% and
shows twin peaks in the water level as a result of runoff discharge. The statistical errors
represented by the EYp values at peak water level 1 and peak water level 2 were 13.5%
and 23.1%, respectively, while the ETp values were 1 hr and 4 hrs, respectively. The
Sustainability 2021, 13, 3180
simulated water level indicated an underestimation of the observed water level at peak 12 of 22
water level 1, while the simulation overestimated the observation at peak water level 2.
The NSE value was 0.69.
Figure 6b illustrates a comparison between the computed and observed water levels
23.1%,
during respectively,
Typhoon Megi,while the ET
as used values
forp the were
model 1 h and 4 The
validation. h, respectively. The simulated
computed water levels
water level
matched indicated an
the tendencies underestimation
of the observations.ofThe
theEY
observed
p and ETwater level
p values at peak
were 4.7%water
and 0level
hr, 1,
while the simulation overestimated the observation at peak water level 2.
respectively, and the NSE value was 0.79. The model performance observed during theThe NSE value
was validation
model 0.69. was better than that observed during the model calibration.
Figure 6. Comparison of the simulated and observed water levels at the Changrun Bridge (a)
Figure 6. Comparison of the simulated and observed water levels at the Changrun Bridge (a) during
during Typhoon Kongrey (2013), used for the model calibration, and (b) during Typhoon Megi
Typhoon Kongrey (2013), used for the model calibration, and (b) during Typhoon Megi (2016), used
(2016), used for the model validation.
for the model validation.
4.3. Inundation Disaster and Hazard Map
Figure 6b illustrates a comparison between the computed and observed water levels
4.3.1. Inundation
during Typhoon Depths
Megi,andas Extents
used forunder Different
the model Return Periods
validation. The computed water levels
matched the tendencies
To obtain of the observations.
rainfall information The EYpreturn
under different and ET p values this
periods, werestudy
4.7% and
used0 h,
historical rainfall data from eight rainfall stations in the catchment area from 2012 to 2017the
respectively, and the NSE value was 0.79. The model performance observed during
formodel validation
frequency wasapplied
analysis, better than that observed
the Horner equationduring the model
to establish thecalibration.
intensity-duration-
frequency curve [43], and finally employed the alternating group method to obtain the
4.3. Inundation Disaster and Hazard Map
designated rainfall of each rainfall station under different return periods. The designated
4.3.1.was
rainfall Inundation
then input Depths andvalidated
into the Extents under
HEC-HMS Different
model Return Periods
to obtain the runoff discharges
To obtain rainfall information under different return
in the upstream catchment area; then, the runoff discharges served asperiods, this study usedinto
inputs historical
the
rainfall data from eight rainfall stations in the catchment area from
validated FLO-2D model, and inundation disasters under different return periods 2012 to 2017 for fre-
quency as
regarded analysis,
hazards applied
were the Horner equation
obtained. For example,to establish the intensity-duration-frequency
the analysis result for the rainfall
curve [43],
hyetograph and finally
under employed
a 200-year return the alternating
period group
is illustrated inmethod
Figure 7.to obtain the designated
rainfall of each rainfall station under different return periods. The designated rainfall
was then input into the validated HEC-HMS model to obtain the runoff discharges in the
upstream catchment area; then, the runoff discharges served as inputs into the validated
FLO-2D model, and inundation disasters under different return periods regarded as haz-
ards were obtained. For example, the analysis result for the rainfall hyetograph under a
200-year return period is illustrated in Figure 7.
6.37 m, respectively. The maximum flooding site is located in the Yanchao District,
resulting from the low-lying area around the riverside in the middle reach. There are 9
villages in the 50-year rainfall return period, 11 villages in the 100-year rainfall return
period, and 12 villages in the 200-year rainfall return period with inundation depths
Sustainability 2021, 13, 3180 greater than 3 m. Because most flooding zones are located downstream of the areas 13 ofofthe
22
river where the flooding depth exceeds 0.3 m, the possible impacts on hazards are not
negligible.
Figure 7. Rainfall hyetograph under a 200-year return period at stations (A–F) used as inputs for the HEC-HMS model.
The simulated results of the inundation depths and extents for 50-, 100-, and 200-year
rainfall return periods are displayed in Figure 8. It can be seen from the figure that the
flooding extent gradually expands from the 50-year rainfall return period to the 200-year
rainfall return period, and the flooding depth gradually deepens. The maximum flooding
depths under the 50-, 100-, and 200-year rainfall return periods are 5.50 m, 5.97 m, and
6.37 m, respectively. The maximum flooding site is located in the Yanchao District, resulting
from the low-lying area around the riverside in the middle reach. There are 9 villages in
the 50-year rainfall return period, 11 villages in the 100-year rainfall return period, and
12 villages in the 200-year rainfall return period with inundation depths greater than 3 m.
Because most flooding zones are located downstream of the areas of the river where the
Figure 7. Rainfall hyetograph under a 200-year
flooding depthreturn period
exceeds 0.3 at
m,stations (A–F) used
the possible as inputs
impacts for the HEC-HMS
on hazards model.
are not negligible.
Figure
Figure 8. Inundation
8. Inundation predictions
predictions under
under (a) (a) 50-year,
50-year, (b) 100-year,
(b) 100-year, and (c) 200-year
and (c) 200-year rainfall return
rainfall return
periods.
periods. Note Note thatmaximum
that the the maximum flooding
flooding site issite is indicated
indicated with awith a circle.
circle.
4.3.2. Inundation Depths and Extents under Different Return Periods

The inundation depths and extents were converted to hazard maps according to
The inundation depths and extents were converted to hazard maps according to 5
5 grades: below 0.3 m, 0.3 m to 0.5 m, 0.5 m to 1.0 m, 1.0 m to 3.0 m, and above 3.0 m
grades: below 0.3 m, 0.3 m to 0.5 m, 0.5 m to 1.0 m, 1.0 m to 3.0 m, and above 3.0 m (see
Table 1). Figure 9 presents hazard maps under the 50-, 100-, and 200-year rainfall return
periods. The figure shows that there are 7 villages in Qiaotou District, 2 villages in
Gangshan District, and one village each in Dashe, Nanzi, and Yanchao Districts belonging
to the high and extremely high hazard levels under different rainfall return periods.
Figure 8. Inundation predictions under (a) 50-year, (b) 100-year, and (c) 200-year rainfall return
periods. Note that the maximum flooding site is indicated with a circle.
The inundation depths and extents were converted to hazard maps according to 5
grades:
(seebelow
Table 0.3
1). m, 0.3 m9 to
Figure 0.5 m, 0.5
presents m tomaps
hazard 1.0 m,under
1.0 m the
to 3.0
50-,m,100-,
and and
above 3.0 m (see
200-year rainfall
Tablereturn
1). Figure
periods. The figure shows that there are 7 villages in Qiaotou District, 2 return
9 presents hazard maps under the 50-, 100-, and 200-year rainfall villages in
periods. The figure
Gangshan shows
District, thatvillage
and one there each
are 7invillages in Qiaotou
Dashe, Nanzi, District,Districts
and Yanchao 2 villages in
belonging
Gangshan District, and one village each in Dashe, Nanzi, and Yanchao Districts belonging
to the high and extremely high hazard levels under different rainfall return periods. Under
to the
thehigh and rainfall
200-year extremely high
return hazard
period, the levels
hazardunder
gradesdifferent rainfall
of 12 villages return
increase periods.
to the extremely
Under the 200-year
high level. rainfall return period, the hazard grades of 12 villages increase to the
extremely high level.
Figure
Figure 9. Hazard
9. Hazard map under
map under (a) 50-year,
(a) 50-year, (b) 100-year,
(b) 100-year, and200-year
and (c) (c) 200-year rainfall
rainfall return
return periods.
periods.
4.4. Analysis of Social Vulnerability and the Social Vulnerability Map
4.4. Analysis of Social Vulnerability and the Social Vulnerability Map
This study used two questionnaires to calculate social vulnerability. The first was
This study used two questionnaires to calculate social vulnerability. The first was the
the fuzzy Delphi questionnaire, which was adopted to screen the vulnerability factors.
fuzzyThrough
Delphi this
questionnaire,
questionnaire,which was
experts adopted
were asked totoidentify
screen the
themain
vulnerability factors.social
factors affecting
Through this questionnaire, experts were asked to identify the main factors affecting
vulnerability in urban areas. The second was the AHP questionnaire. The main purpose of
socialthis
vulnerability in urban
questionnaire was to areas.
set theThe second
weight valuewas the AHP
of each questionnaire.
selected The main
social vulnerability factor.
purpose of this questionnaire was to set the weight value of each selected social
vulnerability factor. Because some factors had a greater degree of influence than others, it
was necessary to determine the weight value of each factor through the AHP method.
4.4.1. Process with Fuzzy Delphi (FD)

Four assessment elements, which included twenty social vulnerability factors, were
Because some factors had a greater degree of influence than others, it was necessary to
determine the weight value of each factor through the AHP method.
4.4.1. Process with Fuzzy Delphi (FD)

Four assessment elements, which included twenty social vulnerability factors, were
selected for the expert questionnaire, as shown in Table 2. Through the 41 collected
expert questionnaires and the FD screen, the social vulnerability factors with consensus
values higher than 6.09 were selected. Therefore, ten main social vulnerability factors,
including individuals over 65 years old, individuals under 14 years old, elderly individuals
living alone, individuals with disabilities, rubber boats, mobile pumps, firefighters, refuge
shelters, schools (kindergarten, elementary, and high schools), and nursing homes for
elderly individuals (see Table 2), were selected for further APH analysis.
Table 2. Social vulnerability factors and consensus values according to the expert questionnaire.
Assessment Element Social Vulnerability Factor Consensus Value

Over 65 years old 6.89
Under 14 years old 7.6
Low-income household 5.47
Elderly living alone 6.89
Disability 6.89
Demographic characteristic
Aboriginal 5.5
New immigrants (foreign spouse) 4.5
Agricultural population 5.69
Illiterate 3.89
University graduate 4.8
Household income 5.09
Social economy
Total value of building 4.5
Rubber boat 7.6
Rescue equipment Mobile pump 7
Firefighter 7.6
Refuge shelter 7.6
School (kindergarten, elementary,
6.8
Special agency and high school)
Nursing home for the elderly 6.32
Dialysis center 6
Homeless shelter 5.19
Mean consensus value 6.09
4.4.2. Process with the AHP

After completing the fuzzy Delphi questionnaire and selecting the main vulnerability
factors, an AHP questionnaire was conducted to determine the weight of each vulnerability
factor. A total of 40 questionnaires were completed and returned by experts. Table 3 shows
the weights of the vulnerability factors and the cascading weights of the hierarchy. The
sum of the cascading weights of the hierarchy is 1.0. The table also indicates that experts
believe that being a firefighter is the most important social vulnerability factor. From the
AHP method, the weight of each factor can be determined, and the SVIs of urban areas can
be evaluated.
Table 3. Weights of the social vulnerability factors.
Assessment Element (A) Social Vulnerability Factor Weighting (B) Cascade Weighting of Hierarchy (C = A × B)
Over 65 years old 0.1812 0.0717
Demographic
Under 14 years old 0.2170 0.0859
characteristic
Elderly living alone 0.2675 0.1058
(0.3957)
Disability 0.3343 0.1323
Rubber boat 0.2276 0.0679
Rescue equipment
Mobile pump 0.2780 0.0830
(0.2084)
Firefighter 0.4944 0.1476
Refuge shelter 0.3316 0.1014
Special agency School (kindergarten,
0.3168 0.0969
(0.3059) elementary, and high school)
Nursing home for the elderly 0.3516 0.1076
4.4.3. Social Vulnerability Map

According to Equations (4)–(6), the social vulnerability index (SVI) can be obtained.
To normalize the factors according to Equation (7), the SVI value was converted to the ISVI
value. Since five social vulnerability levels were graded based on their ISVI values (see
Table 1), the social vulnerability map was built, as illustrated in Figure 10. The map shows
that the social vulnerability in the Nanzi District is the highest, at level 5 (i.e., extremely
high vulnerability). The main reason for this result is that there are more nursing homes
for elderly individuals, individuals over 65 years old, and individuals under 14 years
old in this region than in other regions, resulting in the highest social vulnerability. The
Sustainability 2021, 13, 3180 social vulnerability values of the two villages in Qiaotou District are both level183.ofThis23
is due to there being more nursing homes for elderly individuals in this district than in
other districts.
Figure 10. Social vulneriability map.

Social vulnerability values express relative, comparative relationships and do not
4.5. Risk Map
represent actual vulnerability values. A village with a higher degree of vulnerability means
thatAfter
the village comprises
completing the ahazard
disadvantaged
map undergroup with a rainfall
different denser population or fewer
return periods andsocial
the
resources. Therefore, when a disaster occurs, more serious damage and losses
social vulnerability map, a semiquantitative risk assessment approach was employed would occur
to
in a village
construct with
a risk a higher
map. degree
The map of vulnerability.
of risks under different rainfall return periods is displayed
in Figure 11. It shows that six villages in Qiaotou District and one village in Nanzi District
4.5. Risk Map
belong to medium risk and extremely high risk levels, respectively, under the 50-year
rainfallAfter completing
return the are
period. There hazard
sevenmap under
villages anddifferent rainfall
one village return District
in Qiaotou periodsthat
andare
the
at medium risk and high risk levels, respectively, under the 200-year rainfall returnto
social vulnerability map, a semiquantitative risk assessment approach was employed
construct
period. a risk
It can be map.
notedThe
thatmap
the of risks under
hazard different
risk under rainfall return
the 200-year periods
rainfall returnis period
displayed
is
more serious than that under the 50-year rainfall return period, resulting in increased risks
under the 200-year rainfall return period. However, the risk levels of most villages in the
districts consist of extremely low and low levels.
Sustainability 2021, 13, 3180 4.5. Risk Map 18 of 22

After completing the hazard map under different rainfall return periods and the
social vulnerability map, a semiquantitative risk assessment approach was employed to
construct a risk map. The map of risks under different rainfall return periods is displayed
rainfall return period. There are seven villages and one village in Qiaotou District that
rainfall return period. There are seven villages and one village in Qiaotou District that are
are at medium risk and high risk levels, respectively, under the 200-year rainfall return
at medium risk and high risk levels, respectively, under the 200-year rainfall return
period. It can be noted that the hazard risk under the 200-year rainfall return period is
period. It can be noted that the hazard risk under the 200-year rainfall return period is
Figure
Figure 11. Risk
11. Risk mapmap under
under (a) 50-year,
(a) 50-year, (b) (b) 100-year,
100-year, andand (c) 200-year
(c) 200-year rainfall
rainfall return
return periods.
periods.
5. Discussion
5. Discussion
Flood-related disasters in urban areas are subject to several factors, such as urban
Flood-related disasters in urban areas are subject to several factors, such as urban
development, construction of transportation facilities, extreme weather and climate change,
development, construction of transportation facilities, extreme weather and climate
change, and anthropogenic changes that alter vulnerability, hazards, and risks [59]. To
control the impacts of disasters caused by extreme rainfall events, an adaptation strategy
must be adopted to reduce these disasters. To cope with the impacts of extreme rainfall
events on urban areas, this study formulated adaptation strategies for villages with high
and anthropogenic changes that alter vulnerability, hazards, and risks [59]. To control the
impacts of disasters caused by extreme rainfall events, an adaptation strategy must be
adopted to reduce these disasters. To cope with the impacts of extreme rainfall events on
urban areas, this study formulated adaptation strategies for villages with high risks levels.
Generally, two measures, including nonstructural measures and structural measures, are
recommended to mitigate disasters [25,60].
According to the flood risk assessment, the social vulnerability and hazards under
different rainfall return periods reach extremely high levels, resulting in extremely high
risk levels yielded for a village in Nanzi District. This village possesses a large population
over 65 years old and under 14 years old as well as many nursing home facilities for elderly
individuals. This study suggests that when planning escape routes with nonstructural
measures, complex routes should be avoided, and the routes should be as simple and
clear as possible to avoid dispersion in the escape process. In addition, the inundation
depths are relatively high at this village, so it is recommended that a flood detention pond
is added with structural measures to relieve the flooding load of the channel and reduce
the flood disasters.
The social vulnerability of a village in Qiaotou District is calculated to be at level 3
(i.e., medium level), and the risk in this village for each return period is level 3 for the
50-year return period, level 3 for the 100-year return period, and level 4 for the 200-year
return period. The number of mobile pumps in this area is lower than those of others area;
therefore, this study proposes that mobile pumps are purchased in this area to reduce the
damage caused by flooding.
The method used to reduce the social vulnerability index is a nonstructural measure,
while the method employed to reduce hazards is a structural measure. To reduce the risk
of flooding, it is necessary to first examine the sources of high risk from social vulnerability
or hazards, as measures taken to reduce the risk must be adapted to the local conditions.
In any case, the social vulnerability map, the hazard map, and the risk map can provide
decision-makers with a reference that can be considered for risk management.
Most studies that have discussed hazard assessments applied a combination of hy-
drologic and hydrodynamic models for flood simulations [27,29,54,61]. Additionally, there
are different research methods used to discuss vulnerability [23,33,62]. Further, two expert
questionnaires were used to determine the main social vulnerability factors and the weights
of the social vulnerability factors in the study area. This kind of research method is rarely
proposed or discussed, but is specific and feasible, and the methodology can be applied to
other urban areas affected by floods.
The limitation of this study is that because 20-m DTM data are available, the regional
flooding simulation cannot display the flooded state of the street and because the data of
the social vulnerability factor are not detailed enough, so the flood risk map can only be
presented in the village.
6. Conclusions
Typhoon-induced extreme rainfall frequently results in flood damage that threatens
the lives and property of people and causes economic losses in urban areas in Taiwan.
Apart from structural measures that prevent flood damage, risk assessments play important
roles in disaster management. To explore the flood risk map for disaster mitigation, the
hypotheses of this study include the flood risk level in a village to be same, no engineering
works executed to reduce hazard level, and social vulnerability for different rainfall return
periods also to be same.
The flood risk assessment conducted in this study was established based on rainfall-
runoff (HEC-HMS) and flooding simulation (FLO-2D) models applied for the hazard
analysis and the fuzzy Delphi (FD) method and analytic hierarchy process (AHP) applied
for the social vulnerability analysis. The HEC-HMS and FLO-2D models were calibrated
and validated with observational data. The results indicated that the model performances
during both the model calibration and validation were acceptable. The validated HEC-HMS
and FLO-2D models were utilized to predict inundation depths and extents in urban areas
under different rainfall return periods, and these factors were used to build hazard maps.
Furthermore, to explore social vulnerability, first, the fuzzy Delphi method was used
to select the main social vulnerability factors, and then the AHP was adopted to determine
the weight of each social vulnerability factor and to calculate the SVI to build a social
vulnerability map.
By means of a semiquantitative risk matrix, the social vulnerability map and hazard
map were combined into a risk map, and the risk map was divided into 5 levels to
comprehend the risk status of urban areas in the Dianbao River watershed under different
rainfall return periods.
Most studies did not clearly state that how the social vulnerability was selected
for flood risk analysis. The main contribution of this study was that the hydrological-
hydrodynamic model, the fuzzy Delphi, and the analytic hierarchy process were proposed
to build the flood risk map in urban areas of southern Taiwan. Based on the flood risk
map, the nonstructural measures and structural measures were recommended to mitigate
disasters according to the regional disaster characteristics.
Due to the impacts of climate change, urban flooding caused by extreme rainfall is
worsening. In future research, we will explore a risk assessment of urban flood disasters
caused by climate change.
Author Contributions: Conceptualization, W.-C.L.; methodology, T.-H.H. and H.-M.L.; software,

T.-H.H.; validation, W.-C.L., T.-H.H., and H.-M.L.; formal analysis, W.-C.L.; investigation, W.-C.L.;
resources, W.-C.L.; data curation, T.-H.H. and H.-M.L.; writing—original draft preparation, W.-C.L.;
writing—review and editing, W.-C.L. and H.-M.L.; visualization, T.-H.H.; supervision, W.-C.L.;
project administration, W.-C.L.; funding acquisition, W.-C.L. All authors have read and agreed to the
published version of the manuscript.
Funding: This research was funded by Ministry of Science and Technology, Taiwan, grant nos.
107-2625-M-239-002 and 108-2625-M-239-002.
Institutional Review Board Statement: The study did not require ethical approval.
Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.
Acknowledgments: The authors would like to express their appreciation to the Water Resources
Agency for providing access to their recorded data.
Conflicts of Interest: The authors declare no conflict of interest.
References
1. Intergovernmental Panel on Climate Change. Managing the Risks of Extreme Events and Disasters to advance Climate Change Adaption;
Cambridge University Press: Cambridge, UK; New York, NY, USA, 2012.
2. Centre for Research on the Epidemiology of Disaster; United Nations Office for Disaster Risk Reduction. The Human Cost of
Weather-Related Disasters 1995–2015; CRED: Brussels, Belgium; UNISDR: Geneva, Switzerland, 2015.
3. United Nations Office for Disaster Risk Reduction. Sendai Framework for Disaster Risk Reduction 2015–2030; UNISDR: Geneva,
Switzerland, 2015.
4. Aerts, J.C.J.H.; Botzen, W.J.; Clarke, K.C.; Cutter, S.L.; Hall, J.W.; Merz, B.; Michel-Kerjan, E.; Mysiak, J.; Surminski, S.; Kunreuther,
H. Integrating human dynamics into flood disaster risk assessment. Nat. Clim. Chang. 2018, 8, 193–199. [CrossRef]
5. Yang, T.H.; Liu, W.C. A general overview of the risk-reduction strategies for floods and droughts. Sustainability 2020, 12, 2687.
[CrossRef]
6. UNISDR. Global Assessment Report on Disaster Risk Reduction: Revealing Risk, Redefining Development. Summary and Main Findings;
United Nations: New York, NY, USA, 2011.
7. Jenkins, K.; Surminski, S.; Hall, J.; Crick, F. Assessing surface water flood risk and management strategies under future climate
change. Sci. Total Environ. 2017, 595, 159–168. [CrossRef]
8. Kron, W. Flood risk = hazard·values·vulnerability. Water Int. 2005, 30, 58–68. [CrossRef]
9. European Commission. The EU Flood Directive; Directive 2007/60/EC of the European Parliament and of the Council on the
Assessment and Management of Flood Risk; European Comission: Brussels, Belgium, 2007.
10. Apel, H.; Aronica, G.T.; Kreibich, H.; Thieken, A.H. Flood risk analyses-how detailed do we need to be? Nat. Hazards 2009, 49,
79–98. [CrossRef]
11. De Moel, H.; Aerts, J.C.J.H.; Koomen, E. Development of flood exposure in the Netherlands during the 20th and 21th century.
Glob. Environ. Chang. 2011, 21, 620–627. [CrossRef]
12. Kalogrtopoulos, K.; Karalis, S.; Karymbalis, E.; Chalkias, C.; Chalkias, G.; Katsafados, P. Modeling flash floods in Vouraikos River
mouth, Greece. In Proceedings of the MEDCOAST Conference, Marmaris, Turkey, 30 October–3 November 2013; pp. 1135–1146.
13. Tsanakas, K.; Gaki-Papansdtassiou, K.; Kalogeropoulos, K.; Chalkias, C.; Katsafado, P.; Karymbalis, E. Investigation of flash flood
natural causes of Xirolaki Torrent, Northern Greece based on GIS modeling and geomorphological analysis. Nat. Hazards 2016,
84, 1015–1033. [CrossRef]
14. Bathrellos, G.D.; Karymbalis, E.; Skilodimou, H.D.; Gaki-Papanastassiou, K.; Baltas, E.A. Urban flood hazard assessment in the
basin of Athens Metropolitan city, Greece. Environ. Earth Sci. 2016, 75, 319. [CrossRef]
15. Garrote, J.; Diez-Herrero, A.; Escudero, C.; Garcia, I. A framework proposal for regional-scale flood-risk assessment of cultural
heritage sites and application to the Castile and Leon Region (Central Spain). Water 2020, 12, 329. [CrossRef]
16. Glas, H.; De Maeyer, P.; Merisier, S.; Deruyter, G. Development if a low-cost methodology for data acquisition and flood risk
assessment in the floodplain of the river Moustiques in Haiti. J. Flood Risk Manag. 2020, 13, e12608. [CrossRef]
17. Grezo, H.; Mocko, M.; Izsoff, M.; Vrbicanova, G.; Petrovic, F.; Stranak, J.; Muchova, Z.; Slamova, M.; Olah, B.; Machar, I. Flood risk
assessment for the long-term strategic planning considering the placement of industrial parks in Slovakia. Sustainability 2020, 12,
4144. [CrossRef]
18. Guo, E.; Zhang, J.; Ren, X.; Zhang, Q.; Sun, Z. Integrated risk assessment of flood disaster based on improved set pair analysis
and the variable fuzzy set theory in central Liaoning Province, China. Nat. Hazards 2014, 74, 947–965. [CrossRef]
19. Waghwala, R.K.; Agnihotri, P.G. Flood risk assessment and resilience strategies for flood risk management: A case study of Surat
City. Int. J. Disaster Risk Red. 2019, 40, 101155. [CrossRef]
20. Joo, H.; Choi, C.; Kim, J.; Kim, D.; Kim, S.; Kim, H.S. A Bayesian network-based integrated for flood risk assessment (InFRA).
Sustainability 2019, 11, 3733. [CrossRef]
21. Zhang, J.; Chen, Y. Risk assessment of flood disaster induced by typhoon rainstorms in Guangdong Province, China. Sustainability
2019, 11, 2738. [CrossRef]
22. Lai, C.; Chen, X.; Wang, Z.; Yu, H.; Bai, X. Flood risk assessment and regionalization from past and future perspectives at basin
scale. Risk Anal. 2020, 40, 1399–1417. [CrossRef]
23. Wang, G.; Liu, Y.; Hu, Z.; Lyu, Y.; Zhang, G.; Liu, J.; Liu, Y.; Gu, Y.; Huang, X.; Zheng, H.; et al. Flood risk assessment based fuzzy
synthetic evaluation method in the Beijing-Tianjin-Hebei metropolitan area, China. Sustainability 2020, 12, 1451. [CrossRef]
24. Mani, P.; Chatterjee, C.; Kumar, R. Flood hazard assessment with multiparameter approach derived from coupled 1D and 2D
hydrodynamic flow model. Nat. Hazards 2014, 70, 1553–1574. [CrossRef]
25. Van Berchum, E.C.; Mobley, W.; Jonkman, S.N.; Timmermans, J.S.; Kwakkel, J.H.; Brody, S.D. Evaluation of flood risk reduction
strategies through combination of interventions. J. Flood Risk Manag. 2018, 12 (Suppl. 2), e12506. [CrossRef]
26. Cai, T.; Li, X.; Ding, X.; Wang, J.; Zhan, J. Flood risk assessment based on hydrodynamic model and fuzzy comprehensive
evaluation with GIS technique. Int. J. Disaster Risk Red. 2019, 35, 101077. [CrossRef]
27. Criado, M.; Martinez-Grana, A.; Roman, J.S.S.; Santos-Frances, F. Flood risk evaluation in urban spaces: The study case of Tormes
River (Salamanca, Spain). Int. J. Environ. Res. Public Health 2019, 16, 5. [CrossRef] [PubMed]
28. Shen, Y.; Morsy, M.M.; Huxley, C.; Tahvildari, N.; Goodall, J.L. Flood risk assessment and increased resilience for coastal urban
watersheds under the combined impact of storm tide and heavy rainfall. J. Hydrol. 2019, 579, 124159. [CrossRef]
29. Apollonio, C.; Bruno, M.F.; Iemmolo, G.; Molfetta, M.G.; Pellicani, R. Flood risk evaluation in ungauged coastal areas: The case
study of Ippovampo (Southern Italy). Water 2020, 12, 1466. [CrossRef]
30. Lowe, R.; Arnbjerg-Nielsen, K. Urban pluvial flood risk assessment-data resolution and spatial scale when developing screening
approaches on the microscale. Nat. Hazards Earth Syst. Sci. 2020, 20, 981–997. [CrossRef]
31. Cabrera, J.S.; Lee, H.S. Flood risk assessment for Davo Oriental in the Philippines using geographic information system-based
multi-criteria analysis and maximum entropy model. J. Flood Risk Manag. 2020, 13, e12607. [CrossRef]
32. Zhang, D.; Shi, X.; Xu, H.; Jing, Q.; Pan, X.; Liu, T.; Wang, H.; Hou, H. A GIS-based spatial multi-index model for flood risk
assessment in the Yangtze River Basin, China. Environ. Impact Assess. Rev. 2020, 83, 106397. [CrossRef]
33. Geng, Y.; Zheng, X.; Wang, Z.; Wang, Z. Flood risk assessment in Quzhou City (China) using a coupled hydrodynamic model and
fuzzy comprehensive evaluation (FCE). Nat. Hazards 2020, 100, 133–149. [CrossRef]
34. Chang, C.Y. The Integration of HEC-RAS and FLO-2D Model for Inundation Simulation in the Dianbao River Watershed. Master’s
Thesis, National Cheng Kung University, Tainan, Taiwan, 2013.
35. Water Resources Agency. The Disaster Prevention Monitoring and Model Test Base Observation in the Dianbao River and Ilan River in
2016; Final Report; Water Resources Agency: Taichung, Taiwan, 2016.
36. US Army Corps Engineers. Hydrologic Modeling System (HEC-HMS) Application Guide: Version 4.3; Institute for Water Resources—
Hydrologic Engineering Center: Davis, CA, USA, 2018.
37. Halwatura, D.; Najim, M.M.M. Application of the HEC-HMS model for runoff simulation in a tropical catchment. Environ. Model.
Softw. 2013, 46, 155–162. [CrossRef]
38. Nourali, M.; Ghahraman, B.; Pourreza-Bilondi, M.; Davary, K. Effect of formal and informal likelihood functions on uncertainty
assessment in a single event rainfall-runoff model. J. Hydrol. 2016, 540, 549–564. [CrossRef]
39. Gumindoga, W.; Rwasoka, D.T.; Nhapi, I.; Dube, T. Ungauged runoff simulation in Upper Manyame Catchment, Zimbabwe:
Application of the HEC-HMS model. Phys. Chem. Earth Parts A/B/C 2017, 100, 371–382. [CrossRef]
40. Young, C.C.; Liu, W.C.; Wu, M.C. A physically based machine learning hybrid approach for accurate rainfall-runoff modeling
during extreme typhoon events. Appl. Soft Comput. 2017, 53, 205–216. [CrossRef]
41. de Moraes, T.C.; dos Santos, V.J.; Calijuri, M.L.; Torres, F.T.P. Effects on runoff caused by changes in land cover in Brazilian
southeast basin: Evaluation by HEC-HMS and HEC-GEOHMS. Environ. Earth Sci. 2018, 77, 250. [CrossRef]
42. Yuan, W.; Liu, M.; Wan, F. Calculation of critical rainfall for small-watershed flash flood based on the HEC-HMS hydrological
model. Water Resour. Manag. 2019, 33, 2555–2575. [CrossRef]
43. Ramly, S.; Tahir, W.; Abdullah, J.; Jani, J.; Ramli, S.; Asmat, A. Flood estimation for SMART control operation using integrated
radar rainfall input with the HEC-HMS model. Water Resour. Manag. 2020, 34, 3113–3127. [CrossRef]
44. Chow, V.T.; Maidment, D.R.; Mays, L.W. Applied Hydrology; McFraw-Hill Book Company: New York, NY, USA, 1988.
45. Straub, T.D.; Melching, C.S.; Kocher, K.E. Equations for Estimating Clark Unit Hydrograph Parameters for Small Rural Watersheds in
Illinois; Water-Resources Investigations Report 00-4184; Illinois Department of Natural Resources—Office of Water Resources U.S.
Geological Survey: Urbana, IL, USA, 2000; pp. 4–6.
46. Flo-2D Manual. Reference Manuals. 2012. Available online: http://www.flo-2d.com/download (accessed on 20 July 2019).
47. Dimitriadis, P.; Tegos, A.; Oikonomou, A.; Pagana, V.; Koukouvinos, A.; Mamassis, N.; Koutsoyiannis, D.; Efstratiadis, A.
Comparative evaluation of 1D and quasi-2D hydraulic models based on benchmark and real-world applications for uncertainty
assessment in flood mapping. J. Hydrol. 2016, 534, 478–492. [CrossRef]
48. Hu, M.; Sayama, T.; Zhang, X.; Tanaka, K.; Takara, K.; Yang, H. Evaluation of low impact development approach for mitigating
flood inundation at a watershed scale in China. J. Environ. Manag. 2017, 193, 430–438. [CrossRef] [PubMed]
49. Dalkey, N.C.; Helmer-Hirschberg, O. An experimental application of the Delphi method to the use of experts. Manag. Sci. 1963, 9,
458–467. [CrossRef]
50. Murray, T.J.; Pipion, L.L.; van Gigch, J.P. A pilot study of fuzzy set modification of Delphi. Hum. Syst. Manag. 1985, 5, 76–80.
[CrossRef]
51. Ishikawa, A.; Amagasa, M.; Shiga, T.; Tomizawa, G.; Tatsuta, R.; Mieno, H. The max-min Delphi method and fuzzy Delphi
method via fuzzy integration. Fuzzy Sets Syst. 1993, 55, 241–253. [CrossRef]
52. Jeng, S.J. Fuzzy Assessment Method for Maturity of Software Organization in Improving Its Staff’s Capability. Master’s Thesis,
National Taiwan University of Science and Technology, Taipei, Taiwan, 2001.
53. Saaty, T. The Analytic Hierarchy Process; McGraw-Hill: New York, NY, USA, 1980.
54. Shaw, D.G.; Li, H.C.; Yang, H.H. Disaster Loss and Social Vulnerability Assessment; Technical Report; National Science and
Technology Center for Disaster Reduction: New Taipei City, Taiwan, 2010.
55. Lin, T.H. Flood Hazard and Risk Analysis for Kaohsiung City. Master’s Thesis, National Taiwan University, Taipei, Taiwan, 2015.
56. Maskrey, A. Disaster Mitigation: A Community Based Approach; Oxfam: Oxford, UK, 1989.
57. Crichton, D.; Mounsey, C. How the insurance industry will use its flood research. In Proceedings of the 33rd MAFF Conference of
River and Coastal Engineers. Keele University, Staffordshire, UK, 1–3 July 1998; pp. 131–134.
58. Water Resources Agency. Research, Development and Value-Added Application of Flood Potential Risk Map Update; Technical Report;
Water Resources Agency: Taichung, Taiwan, 2015.
59. Adikari, Y.; Osti, R.; Noro, T. Flood-related disaster vulnerability: An impending crisis of megacities in Asia. J. Flood Risk Manag.
2010, 3, 185–191. [CrossRef]
60. Mai, T.; Mushtaq, S.; Reardon-Smith, K.; Webb, P.; Stone, R.; Kath, J.; An-Vo, D. Defining flood risk management strategies: A
systems approach. Int. J. Disaster Risk Reduct. 2020, 47, 101550. [CrossRef]
61. Sarchani, S.; Seiradakis, K.; Coulibaly, P.; Tsanis, I. Flood inundation mapping in an ungauged basin. Water 2020, 12, 1532.
[CrossRef]
62. Zou, Q.; Liao, L.; Qin, H. Fast comprehensive flood risk assessment based on game theory and cloud model under parallel
computation (P-GT-CM). Water Resour. Manag. 2020, 34, 1625–1648. [CrossRef]

Flood Risk Assessment in Urban Areas of Southern Taiwan: Sustainability

Uploaded by

Copyright:

Available Formats

Flood Risk Assessment in Urban Areas of Southern Taiwan: Sustainability

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Flood Risk Assessment in Urban Areas of Southern Taiwan: Sustainability

Uploaded by

Copyright:

Available Formats

sustainability

Academic Editors: Marc A. Rosen 1. Introduction

Sustainability 2021, 13, 3180. https://doi.org/10.3390/su13063180 https://www.mdpi.com/journal/sustainability

3. Materials and Methods

3.3. Hazard Assessment with Hydrological and Hydrodynamic Models

3.3.1. Hydrological Model (HEC-HMS)

Tlag = 0.6 ∗ TC (3)

3.3.2. Inundation Model (FLO-2D)

Sustainability 2021, 13, 3180 6 of 22

FigureFigure 3. (a) Modeling

3.6. Social Vulnerability Index (SVI) and Normalization

Degree of Social Vulnerability

3.7. Hazard Assessment

3.8. Risk Assessment

4.1. HEC-HMS Model Calibration and Validation

Sustainability 2021, 13, 3180 14 of 24

Sustainability 2021, 13, 3180 15 of 23

4.3.2. Inundation Depths and Extents under Different Return Periods

Sustainability 2021, 13, 3180 16 of 23

4.4.1. Process with Fuzzy Delphi (FD)

4.4.1. Process with Fuzzy Delphi (FD)

Assessment Element Social Vulnerability Factor Consensus Value

4.4.2. Process with the AHP

Table 3. Weights of the social vulnerability factors.

4.4.3. Social Vulnerability Map

Figure 10. Social vulneriability map.

Sustainability 2021, 13, 3180 4.5. Risk Map 18 of 22

Sustainability 2021, 13, 3180 19 of 23

Author Contributions: Conceptualization, W.-C.L.; methodology, T.-H.H. and H.-M.L.; software,

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.