You are on page 1of 13

bs_bs_banner

A comparative evaluation of flood mitigation alternatives using


GIS-based river hydraulics modelling and multicriteria
decision analysis
M.R. Ghanbarpour1, S. Salimi2 and K.W. Hipel3
1 Global Institute for Water Security, University of Saskatchewan, Saskatoon, SK, Canada
2 Faculty of Natural Resources, University of Mazandaran, Sari, Iran
3 Department of Systems Design Engineering, University of Waterloo, Waterloo, ON, Canada

Correspondence Abstract
M. Reza Ghanbarpour, Global Institute for
Water Security, University of A multicriteria framework is developed for the selection of optimal flood miti-
Saskatchewan, 11 Innovation Boulevard, gation and river training measures in a selected reach of Zaremroud River in
Saskatoon, SK, Canada S7H 1A4 Northern Iran. A river model, Hydrologic Engineering Center River Analysis
Email: ghanbarpour.m.reza@gmail.com System, combined with geographic information system analysis is used to simu-
late water levels for steady, gradually varied flow and mapping inundated flood
DOI: 10.1111/jfr3.12017 extents. The modelling is performed for four different alternatives, considering
various channel modifications with different dimensions and levee construc-
Key words
tion. Flood inundation area, flood level, flow velocity and stream power on the
Decision analysis; flood mitigation; flood
plain; HEC-RAS; TOPSIS.
downstream and outside of the river bend are used as decision criteria for each
alternative. Economic analysis is conducted to evaluate the cost-effectiveness of
each alternative. The decision analysis method, technique for order of prefer-
ence by similarity to ideal solution, is used to compare different flood hazard
mitigation measures based on risk, and environmental and economic impacts
criteria. The findings of the analysis are that a levee construction at the right side
of the river bank adjacent to the residential area is superior to the other three
alternatives, which is confirmed using a scenario analysis of different flood
mitigation measures.

Introduction the negative consequences. Structural measures, including


levees, high flow diversion, channel modification and dams,
The hazard of flooding is receiving increasing attention could be implemented to mitigate flood risk by reducing the
in both the public and professional arenas. Flooding is a volume of run-off, water level or extent of the area of flood-
disastrous natural phenomenon, producing many socio- ing. However, non-structural methods, such as flood insur-
economic and environmental consequences within the ance, land use regulation and flood forecasting, serve as
affected flood plain. Flood risks are rising in many parts of preventive measures for reducing flood hazards.
the world owing to continually increasing populations and Structural measures are favoured in situations in which
rapidly escalating land developments (Walker and Maid- portions of adjacent residential areas are situated about the
ment, 2006; Chang et al., 2009). Studies have shown that maximum flood level, or when it is important to protect land
more than one third of the world’s land area is flood prone, adjoining a river from inundations due to an existing flood
and about 196 million people in more than 90 countries are risk. By implementing structural measures, flood hazards for
exposed to catastrophic flooding (United Nation Develop- these kinds of areas are reduced. However, structural flood
ment Programme (UNDP), 2004; Dilley et al., 2005). As a control methods like channelisation have some predictable
result, the frequency, extent and subsequent hazards associ- negative effects. Studies have shown that sediment loads may
ated with flooding are global concerns. increase below channelisation works and other river training
Flood hazard mitigation plans could be implemented as measures (Hill, 1976). Moreover, changes may occur in dis-
either structural or non-structural measures, depending on charge characteristics below channelisation works; in par-
the particular case (Correia et al., 1999). These measures ticular, flood levels may be increased downstream (Brookes,
involve managing the effects of flooding and preventing 1987).

J Flood Risk Management •• (2013) ••–•• © 2012 Blackwell Publishing Ltd and The Chartered Institution of Water and Environmental Management (CIWEM)
2 Ghanbarpour et al.

Interaction between river flow and stream channel is economic cost. An MCDA method, technique for order of
dynamic, constantly responding to natural and human- preference by similarity to ideal solution (TOPSIS), is
induced river and flood plain changes. Finding sustainable employed to find the best flood control measure, one which
mitigation measures for flood hazards is made difficult by protects residential areas from flood hazards while mini-
the complex and dynamic interrelationships of environmen- mising the economic costs and downstream effects. This
tal, technical, economic and managerial factors in river decision analysis method is used to integrate economic
systems. Therefore, accurate simulation of water surface pro- analysis with the results of the river model, HEC-RAS,
files and reliable delineation of flood extents and depths combined with geographic information system (GIS)
within the flood plain (Pappenberger et al., 2005), as well as analysis.
an understanding of the complex interaction between flood
mitigation measures and river systems, combined with
socio-economic impacts (Simonovic and Carson, 2003; Yi Materials and methods
et al., 2010), are necessary for flood mitigation planning and
flood plain management. Study area
Although some studies have been conducted to model
Zaremroud River is a major tributary river in Tajan Basin,
flood hazards and flood plain mapping (e.g. Correia et al.,
which lies about 15 km southeast of Sari, Mazandaran,
1999; Noman et al., 2003; Pappenberger et al., 2005;
Iran, between 36° 26′ 15″ and 36° 26′ 44″ N latitude, and
Ghanbarpour et al., 2011), less effort has been done on the
53° 8′ 23″ and 53° 9′ 52″ E longitude (Figure 1). Tajan River
topic of modelling best river engineering and training meas-
flows south of Sari City and drains into the Caspian Sea. The
ures. Bana e Costa et al. (2004) have used a multicriteria
length of the river is approximately 95 km, flowing westward
process to evaluate flood control options for the catchment
to join Tajan River, and ultimately draining into the Caspian
of the Livramento Creek in the peninsula of Setúbal, in
Sea. The section of the Zaremroud River flood plain chosen
Portugal. They have demonstrated a quantitative evaluation
for this research lies adjacent to the village and residential
model based on qualitative value judgments formulated by a
area called Garm Roud (Figure 1). The altitude of the
group of experts to integrate socio-economic criteria into
selected reach ranges from 145 to 162 m above mean sea
technical and environmental objectives. Williams (2006)
level. The climate is extremely humid, and the maximum
carried out a study on Santee River to evaluate the impact of
precipitation occurs in autumn and minimum precipitation
the Santee River rediversion. The flooding pattern within the
in summer. The maximum and minimum temperatures in
Santee flood plain has been altered by rediversion and opera-
Zaremroud Watershed are 20.4 °C and 8.95 °C, respectively,
tion of the hydroelectric plant on the rediversion canal.
and the mean annual temperature is about 14.5 C. The
Hydrologic Engineering Center River Analysis System
length of river reach selected in this study is about 3 km of
(HEC-RAS) and the HEC-GeoRAS extension for ArcView
Zaremroud River. Figure 1 shows the Tajan River Watershed
were used to examine the flooding regime prior to and sub-
and the location of the Garm Roud reach, which was selected
sequent to rediversion operations.
in this research. The dangerous flood events of this river
However, performance analysis of any flood mitigation
caused considerable losses, such as damages to two bridges,
measure should be considered before its implementation
the destruction of the Garm Roud hydrometric station and
using simulation models and decision analysis methods.
other similar cases during recent years, according to a report
Moreover, economic analysis along with river hydraulic
from the Mazandaran Agricultural Organisation (Salimi
simulation play important roles in formulating measures for
et al., 2008). Garm Roud residential areas have seriously have
flood hazard mitigation (Simonovic and Carson, 2003).
been affected by flood hazards recently, as the river stream
Multicriteria decision analysis (MCDA) methods (Hwang
causes bank undercutting, which has resulted in landslides in
and Lin, 1987; Ghanbarpour and Hipel, 2011) could be
some parts of the village in the long term.
employed to integrate technical, environmental and socio-
economic objectives to reach an optimal decision.
Data
In this research, the 3-km end of Zaremroud River,
upstream of Tajan River in northern Iran, was selected. The In this research, three types of information, namely hydro-
main objective of this study is to compare different river logical data, GIS-based spatial layers, and cost and quantity
training measures, such as channel modifications and levee estimation of engineering practices are used. Hydrological
construction, to find the optimal approach for meeting data include annual maximum river flow obtained from
technical, environmental and economic criteria among the the Regional Water Company of Mazandaran. Annual
various alternatives. The core concern is to minimise the maximum river flow data recorded at the Garm Roud hydro-
effects of the floods on the residential areas and environ- metric station, located upstream of the selected river reach, is
ment, through a technically sound measure, at a reasonable used for this study. The length of the selected data set is 24

© 2012 Blackwell Publishing Ltd and The Chartered Institution of Water and Environmental Management (CIWEM) J Flood Risk Management •• (2013) ••–••
Comparative evaluation of flood mitigation alternatives 3

Figure 1 Location of study area and selected reach of Zaremroud River in Northern Iran.

years, from 1986–1987 to 2000–2001, based on the consist- profiles available in AutoCAD format, which is converted
ency and availability of data during that time period. Flood into standard shapefiles layers that can be used in ArcGIS. A
frequency analysis is conducted using various statistical dis- topographical map of the area with a scale of 1:25000 and a
tributions by selecting annual maximum discharge at the river plan with a scale of 1:1000 was applied for triangulated
Garm Roud hydrometric station. The most commonly used irregular network generation, using the 3D analyst capability
statistical distributions for flood frequency analysis used in of ArcGIS. Field and lab analysis were conducted to obtain
this research include Log-Normal, Log-Normal III, Pearson other spatial information for modelling purposes, which
III, Log-Pearson III and Gumbel. Peak river flow in 2-, 5-, include the Manning’s roughness coefficient for river and
10-, 20-, 50- and 100-year return periods is estimated based flood plains, land use, vegetation and flow conditions, and
on the selected most accurate statistical distribution. channel obstruction. Whereas the surveyed points of the
The GIS data include cross section profiles of Garm Roud Zaremroud River plan have relative elevation, global posi-
River, elevation points and a topographical map collected tioning system has been used for determining absolute and
from the Regional Water Company, resulting from a survey correct positioning of all river cross sections and elevation
conducted in May 2000. There are 32 river cross section points.

J Flood Risk Management •• (2013) ••–•• © 2012 Blackwell Publishing Ltd and The Chartered Institution of Water and Environmental Management (CIWEM)
4 Ghanbarpour et al.

An economic analysis is conducted to determine the cost satellite-derived data in the Zaremroud River flood plain.
of different river training measures. The cost of implemen- Few points were recognised based on silt lines on the outcrop
tation and maintenance of flood mitigation alternatives, rocks and trees that were believed to be the stage indicators
including engineering, design and supervision cost, are cal- of the 1999 flood in the study reach (274 m3/s). We were not
culated based on an economic analysis and a review of tech- able to prepare continuous-observed flood level because of
nical reports (Deputy of Technical Affairs (DTA), 2009), and very few reliable pieces of evidence. Therefore, only three
previous related projects in the study area. acceptable points were used as observed control points for
calibration of the model. It is noticeable that many opera-
Steady flow water surface simulation tional flood inundation predictions are made in situations
when the observation data to calibrate and evaluate the
A simulation of water surface profiles for steady, gradually
model performance are limited (Beven, 2002; Pappenberger
varied flow is conducted using the HEC-RAS model. The
et al., 2005). Calibration of the model was started from the
HEC-RAS computational procedure is based on the solution
well-estimated values of the Manning’s n based on filed
of the one-dimensional steady flow calculations using an
works and well-known literatures (Hicks and Mason, 1991).
energy equation. The HEC-RAS model was developed at the
The initial n-value is changed until an acceptable simulation
HEC, which is a division of the Institute for Water Resources,
is achieved based on a trial and error method. The percent-
US Army Corps of Engineers (Hydrologic Engineering
age of correct overlap between predicted and observed flood
Center (HEC), 2010). The steady flow component is capable
level data was used as a criterion to find an optimal n-value.
of modelling subcritical, supercritical and mixed flow
regimes water surface profiles. The basic computational pro-
cedure for water surface profiles computation is based on Flood mitigation alternatives
solving the energy equation from one cross section to the The effects of various river training measures can be ana-
next using an interactive procedure called the standard step lysed based on their environmental and socio-economic
method ( HEC, 2010). The energy equation is written as impacts. Regarding the urgent requirement of resolving the
follows: flood hazard to residential areas, the structural measures
a2v 22 a1v12 were proposed as the main strategy and the non-structural
y2 + z2 + = y1 + z1 + + he (1) measures, such as flood warning system, retarding basin,
2g 2g
land use planning and integrated watershed management,
where y1 and y2 are depth of water at two adjacent cross should be adopted as supplements to strengthen flood
sections, z1 and z2 are elevation of the main channel inverts, hazard mitigation. For this study, four flood mitigation alter-
v1 and v2 are average velocities (total discharge divided by natives, including levee construction and different scale of
total flow area), a1 and a2 are velocity-weighting coefficients, channel modification along with dredging river channel of
g is gravitational acceleration and he is energy head loss debris and sediment, are proposed in order to reduce flood
between two cross sections. hazards in a residential area and its surrounding flood plain,
Cross sectional geometry of the channel, river system as described later.
schematics, ineffective flow areas, reach lengths, energy loss, Alternative 1: A levee is an earthen embankment sup-
Manning’s coefficient and steady flow input are the basic ported by Gabion-type revetments (riprap) and is con-
data requirements for water surface simulations. The rough- structed on the right side of the river bank adjacent to the
ness coefficients, which represent the surface’s resistance to residential area, with the capability of withstanding a 100-
flow, were estimated using reliable literature showing a set of year flood. Levees are chosen for this reach because they
roughness values for various surface materials, including seem to be an effective solution for reducing flood damage
descriptions and photographs (Chow, 1959; Sellin et al., and are practical because of the availability of materials and
2003). Boundary conditions are specified to establish the cost-effectiveness. Dimensions of the levee considered have
starting water surface in a subcritical, critical and mixed flow an average elevation of 2 m and length of 250 m. Riprap
regime. After completing all essential data inputs, the model revetment with an additional 1-m foundation form a steep
could be run. slope in the riverside to prevent any limitation in the cross
The HEC-GeoRAS extension is used as an interface to sectional area of the river channel.
visualise flood extents and water depth, and conduct pre- Alternative 2: In this alternative, river bed elevation
and post-processing of flood plain mapping (HEC, 2009; should be set to 157.5 m with a slope of 0.0065. River width
Ghanbarpour and Hipel, 2011). is widened an average of 25 m in the main channel using
A fieldwork was conducted to collect some pieces of evi- channel excavation and dredging.
dence on the historical flood extent in the area because of the Alternative 3: Channel modification is conducted in two
lack of observed flood level data and no available concurrent stages to construct a compound river channel. First, river

© 2012 Blackwell Publishing Ltd and The Chartered Institution of Water and Environmental Management (CIWEM) J Flood Risk Management •• (2013) ••–••
Comparative evaluation of flood mitigation alternatives 5

width is widened an average of 50 m in the main channel $US30 per cubic meter, respectively, in the region, and
without considerable deepening but more dredging. Second, embankment for construction of the levee costs $US8 per
the main channel is deepened by an elevation of 157.5 m cubic meter (DTA, 2009). Channel excavation cost depends
with a slope of 0.0065 and a 20 m width using machinery- on such factors as the size of the channel, the type of equip-
based channel excavation. ment required, access problems and river flow diversion. In
Alternative 4: Channel modification is conducted as the order to develop an estimate of cost, volumes of excavation
combination of alternatives 1 and 3. In this case, a levee as an were calculated per meter of length for the various channel
earthen embankment supported by rock-type revetments sizes. An average machinery excavation and disposal cost of
(riprap) is constructed. $US12 per cubic meter (DTA, 2009) was used as an estimate
Channel modification for flood control is based upon the of excavation costs. Engineering, design and supervision
idea that widening and deepening of the channel by sedi- costs, such as preliminary surveys, formulation, documenta-
ment removal will provide a substantially larger channel tion, construction drawings and surveying, are assumed to be
capacity to allow a greater volume of water to flow down- equal for all alternatives. In the study area, dredging is often
stream without causing an impact on the adjacent flood a locally initiated operation where bulldozers, end loaders,
plain. In proposed alternatives, no channel straightening is backhoes or other excavators are used to remove deposited
considered in order to prevent any negative erosion effects sediment and rock materials, and reconfigure the stream into
downstream (Wyzga, 1993), as there is a natural bend in the a trapezoidal channel. Soil material removed from the river
river reach that is thought to be more effective in decreasing channel is often used to raise the level of the adjacent flood
the energy and velocity of flow, and thereby prevent any plain or used as construction materials. As a result, the reduc-
further erosion. Removing the natural bend of a river usually tion in flood levels, flow velocity and stream power at two
leads to stream bank erosion and often requires costly pro- selected control points, and the cost of implementation of the
tection measures. No water detention basins or flood diver- alternatives were adopted as the criteria for evaluating flood
sion structures are considered in this study area because of mitigation performance for each proposed measure.
limited land availability.
The selected flood mitigation alternatives could be posi-
tioned at appropriate locations along the Zaremroud River Decision analysis
reach in HEC-RAS model runs to extract model results for An MCDA method, TOPSIS (Hwang and Lin, 1987;
comparison. Flood level, flow velocity and stream power at Ghanbarpour and Hipel, 2011), was used to find an optimal
various locations along the river and flood plain are pre- flood mitigation alternative considering different criteria. In
dicted based on proper flood plain inundation analysis in the this method, the optimal alternative could be found based on
50- and 100-year flood associated with the four proposed the shortest distance from the ideal solution and the farthest
alternatives, including current condition (baseline). Imple- distance from the anti-ideal solution. As Chen and Hwang
mentation of river training practices would likely alter flow (1992) have shown, a decision matrix is a two-dimensional
characteristics along the river system including flood level, (m*n) matrix for which A1, A2, . . ., Am are m possible alter-
flow velocity and stream power. Therefore, changes in flood native among which decision makers have to choose, and C1,
level and flow characteristics, including velocity and total C2, . . . , Cn are n decision criteria which flood mitigation
stream power (total cross section shear stress multiplied by alternatives are compared based. xmnindicates the perform-
total cross section average velocity), are adopted as decision ance rating of alternative Am with respect to criterion Cn. The
criteria along the selected reach for selecting a reliable and relative importance of each criterion is usually given by a set
environmentally sound flood control alternative. Two spe- of weights that are normalised to sum to one. In the case of
cific control cross sections on the downstream and outside of n criteria, a weight set is:
the river bend are considered for comparison purposes. The
n
downstream of the residential area after the river bend is w = (w1, w2, … , wn ) when ∑ w j = 1 (2)
considered to control impacts of flood control measures on j =1
different hydraulics parameters in the downstream of the
river system. The outside of the river bend adjacent to the Then, the normalised value xij is calculated as:
residential area is also considered as the second control xij
point, which is important in terms of higher risk of erosion rij = , i = 1, 2, … , m; j = 1, 2, … , n
m
(3)
and flood hazard. ∑x 2
ij
Cost of implementation and maintenance of flood miti- i =1

gation alternatives is calculated based on economic analysis The weighted normalised value vij is calculated as:
and a review of previous related projects in the study area.
Rock-type and Gabion riprap construction cost $US10 and vij = w j xij , i = 1, 2, … , m; j = 1, 2, … , n. (4)

J Flood Risk Management •• (2013) ••–•• © 2012 Blackwell Publishing Ltd and The Chartered Institution of Water and Environmental Management (CIWEM)
6 Ghanbarpour et al.

Table 1 Weight of all decision-making criteria based on different scenarios


Decision criteria Decision-making scenarios for flood risk mitigation
Criteria Subcriteria Risk-oriented Environmental-oriented Economic-oriented Balance
Flood risk Inundation area (ha) 0.286 0.143 0.1666 0.2
Flood level (m) 0.286 0.143 0.1666 0.2
Environmental impact Flow velocity (m/s) 0.143 0.286 0.1666 0.2
Stream Power (N.m/s) 0.143 0.286 0.1666 0.2
Economic impact Cost (1000 $US) 0.143 0.143 0.333 0.2

n
where wj is the weight of the jth criterion and ∑w j = 1 . The For this purpose, four decision-making scenarios are con-
j =1 sidered for comparison, which are shown in Table 1. The first
-
ideal A* and anti-ideal A solution can be determined as:
scenario is flood risk-oriented decision making in which
A* = {(max i vij j ∈ J ) , (mini vij j ∈ J ′ ) i = 1, 2, … , m} higher weight are given to flood risk criteria (flood inunda-
(5) tion area and flood level). Therefore, the safety of residential
= {v1*, v 2*, … , v *j , … , v n* }
area is the main concern in this scenario. The second sce-
nario is defined as environmental-oriented in which higher
A − = {(mini vij j ∈ J ) , (max i vij j ∈ J ′ ) i = 1, 2, … , m}
(6) weight is given to environmental criteria (flow velocity and
= {v1−, v 2−, … , v −j , … , v n− }
stream power). Therefore, higher priorities are given to envi-
where J = {j = 1, 2, . . . , n|j} is associated with benefit criteria, ronmental issues such as erosion and bank cutting in the
and J- = {j = 1, 2, . . . , n|j} is associated with cost criteria. The river bend and downstream as a result of flood control
separation of each alternative from the ideal and anti-ideal measures implementation. The third scenario is defined as
solutions can be measured by the n-dimensional Euclidean economic-oriented decision-making option in which the
distance: construction cost of alternatives is given higher weight. The
last scenario is defined as a balance decision-making option,
n
which is considered as the balance between all decision cri-
Si* = ∑ (v ij − v *j )2 , i = 1, 2, … , m. (7)
teria with equal weight and priorities. Table 1 shows details
j =1
of criteria and subcriteria weighting for different scenarios
n considered in this research. It should be noted that the sum
∑ (v − v −j ) , i = 1, 2, … , m.
2
Si− = ij (8) of all criteria weights equals one. Therefore, the original
j =1
weight is divided by two if there are two subcriteria. For
At the final stage, the relative closeness of Ai with respect to example, in risk-oriented scenario, flood level criteria
A* is defined as: (0.286) have two subcriteria at two control points on the
downstream and outside of the river bend. Therefore, the
Ci* = Si− /(Si* + Si− ), 0〈Ci* 〈1, i = 1, 2, … , m. (9) weight of 0.143 is considered for each of the two control
points.
Three main decision criteria and their subcriteria including
flood risk (inundation area and flood level), environmental
impact (flow velocity and stream power) and economic
impacts (construction cost) are considered for comparison
Results and discussion
of flood mitigation measures (Table 1). For all decision cri-
Flood inundation analysis
teria and subcriteria, a set of weights should be assigned to
reflect the importance of that particular decision variable, The HEC-RAS model in combination with GIS analysis was
regarding Eqn (2) and Eqn (4). In this research, a scenario- used to simulate steady flow along the selected reach of
based decision analysis is conducted to show the importance Zaremroud River. The HEC-GeoRAS extension is used in
of each criterion and to integrate them into the decision conjunction with spatial and three-dimensional analyst for
process. The sensitivity of final decision to the criterion interpolation of digital terrain data, and proper display of
weightings could be assessed by varying the weight of each flood water depth and inundated areas using GIS. Flood
criterion. Scenario analysis could indicate how the variation frequency analysis was conducted to select the most accurate
of the weights of the criteria would affect the final decision. input for the hydraulic simulation of the river reach. It is
This is essential to guarantee that the proposed alternative is shown that Log-Normal III, Pearson III and Log-Normal
optimal in terms of different considerations and point of II are the best statistical distributions to estimate peak flow
views. in different return periods with 9.97, 11.30 and 15.25

© 2012 Blackwell Publishing Ltd and The Chartered Institution of Water and Environmental Management (CIWEM) J Flood Risk Management •• (2013) ••–••
Comparative evaluation of flood mitigation alternatives 7

mean relative deviation errors, respectively. Therefore, Log- together to reach an optimal and cost-effective alternative,
Normal III is used to estimate the flood peak flow design for which will be discussed in the next section. The evaluated
the Zaremroud River reach, as it demonstrated the least dif- results for all measures can be summarised as follows.
ferences between observed and estimated peak flows. Results Flood inundation area, flood level, flow velocity and
of flood frequency analysis showing a magnitude of 50- and stream power at two control points along the Zaremroud
100-year peak flows are 378.8 and 479.7 m3/s, respectively, River system are extracted based on flood plain inundation
which are used as a base to run the model and compare analysis for the 50- and 100-year floods associated with the
different flood control alternatives. The calibration of the four proposed alternatives using the HEC-RAS simulation
model was conducted to find the optimal values of Man- combined with GIS (Figures 2 and 3). Table 2 is determined
ning’s roughness by comparing simulated and observed based upon flood inundation layers in GIS depicted in Fig-
flood water depths at three selected points. The least differ- ures 2 and 3. Table 2 shows a simple comparison of the inun-
ence between the simulated and observed flood levels were dation area (ha) and flood depth (m) for each alternative
used as calibration criterion. Based on trial and error model under the 50- and 100-year floods. Without any flood miti-
runs, a Manning’s roughness coefficient of 0.041 and 0.053 gation measure (baseline), the total inundation areas are
produced the best fit against the observed data for the river 44.4 and 49.5 ha, under the 50 and 100-year floods, respec-
main channel and flood plain, respectively. The flood water tively, including 16.2 and 20.1 ha with an inundation depth
depths from the optimal Manning’s n are in a relative agree- over 1.5 m. For the measures of alternatives 3 (channel
ment with those simulated ones (with 26% error of estima- modification by widening and deepening) and 1 (levee con-
tion). Although the error of estimation is high, it is the most struction), the total inundation areas are 38 and 44 ha for a
achievable calibration run because of the lack of observed 50-year flood and 43.1 and 49.3 ha for a 100-year flood,
flood level data in the river basin. This large discrepancy is respectively, which are less than those of the other alterna-
most likely due to the changes in topography and hydraulic tives including baseline. Alternative 3 has the smallest areas
characteristics of the river system, especially as a result of of inundation, showing significant diminution for areas of a
development and man-made activities. The calibrated model depth over 1.5 m (Table 2). As a result, it can be found that
is considered as a baseline for comparison purposes, in spite alternatives 3 and 1 are superior to the other two measures in
of the high degree of uncertainty involve in the model. In terms of flooded area protection.
fact, all flood mitigation alternatives are compared based on Table 3 compares hydraulics parameters in baseline con-
similar model settings. ditions with those reconstructed from flood mitigation alter-
The selected flood mitigation alternatives are positioned natives under 50- and 100-year floods at two specific control
at appropriate locations along the Zaremroud River reach in sections. The effects on flood level reduction by the different
HEC-RAS model runs. Delineation of flood extents and alternatives are presented in Table 3. Alternatives 3 and 4
depths within the flood plain of Zaremroud River was con- have better performance in terms of flood level reduction
ducted in different return periods based on the integration than the other alternatives on the downstream and the
of hydraulic simulation results and GIS analysis using the outside of the river bend control points, respectively.
HEC-GeoRAS. Flood inundation extents, including different However, baseline and levee construction have the highest
model outputs in all cross sections, are extracted after flood levels in selected cross sections. There is no major
running the model for each alternative. Figures 2 and 3 show improvement in flood level reduction using construction of
subsequent maps of predicted 50- and 100-year flood inun- a levee in comparison with the baseline. Although the high-
dation extents and depths for all proposed flood mitigation protection levee may not decrease river flood stages, it could
alternatives. The flood boundary is delineated in blue, with decrease the risk of an overbank flooding disaster in residen-
gradations of blue corresponding to different flood depths. tial areas.
As can be seen in Figure 3, flood-affected areas for the 100- Comparisons between all alternatives in terms of
year flood event are of larger extent than the 50-year flood changes in flow velocity under 50- and 100-year floods are
(Figure 2), as it has affected some parts of Garm Roud resi- presented in Table 3 at two selected control points. There is
dential areas. a variation of flow velocity among the alternatives, depend-
ing on the local river geometry and hydraulic conditions of
river flow. As can be seen in Table 3, alternative 1 causes a
Comparison of flood mitigation alternatives
significant decrease in flow velocity on downstream cross
All proposed flood mitigation alternatives have different section in comparison with the baseline. Generally, levee
environmental and socio-economic consequences. In this construction has the lowest negative effects on flow velocity
part of the paper, we analyse individual impacts of flood on the downstream (Table 3) and is superior to other alter-
control alternatives. However, the final decision should be natives for erosion control and river bank protection. Alter-
made with the consideration of all evaluation criteria native 2 performs better than other options in terms of

J Flood Risk Management •• (2013) ••–•• © 2012 Blackwell Publishing Ltd and The Chartered Institution of Water and Environmental Management (CIWEM)
8 Ghanbarpour et al.

Figure 2 Flood inundation areas and depths for 50-year flood in different flood mitigation alternatives.

flow velocity on the outside of the river bend. However, the downstream and the outside of the river bend cross
alternatives 4 and 2 have the highest downstream flow sections for each alternative under 50- and 100-year floods.
velocity, which could accelerate downstream river bank As can be seen in Table 3, there are critical and extreme
erosion. Alternative 3 has highest flow velocity at the river stream power changes in alternatives 4 and 2 on the down-
bend control point, which could cause progressive river stream. Generally, alternatives 1 and 2 show better results
bank cut and erosion towards residential area. Table 3 for stream power change on control points, respectively.
shows a simple comparison of the stream power change on Results show that changes in hydraulic parameters of

© 2012 Blackwell Publishing Ltd and The Chartered Institution of Water and Environmental Management (CIWEM) J Flood Risk Management •• (2013) ••–••
Comparative evaluation of flood mitigation alternatives 9

Figure 3 Flood inundation areas and depths for 100-year flood in different flood mitigation alternatives.

stream flow occurred as a direct consequence of the points. Figure 4 shows some examples for comparison of
channel modification practices. flood control alternatives using trade-offs between pairs of
Trade-offs analysis between different decision variables to decision variables. For example, alternative 1 is the most
find the optimum flood control measure is a difficult task, as preferred option in terms of downstream flow velocity
there is conflict between different variables in two control and stream power (Figure 4B) and construction cost

J Flood Risk Management •• (2013) ••–•• © 2012 Blackwell Publishing Ltd and The Chartered Institution of Water and Environmental Management (CIWEM)
10 Ghanbarpour et al.

Table 2 Comparisons of inundation areas (ha) for proposed flood mitigation alternatives based on flood depth in 50- and 100-year floods

Return Flood mitigation alternatives


period (year) Flood depth (m) Alternative 1 Alternative 2 Alternative 3 Alternative 4 Baseline
50 D < 0.5 10.5 10.8 8.4 10.8 10.8
0.5 < D < 1.5 17.3 17.4 14.5 17.4 17.4
1.5 < D 16.2 16.2 15.1 16.2 16.2
Total 44 44.4 38 44.4 44.4
100 D < 0.5 11.1 11.6 9.6 11.6 11.6
0.5 < D < 1.5 17.8 17.8 14.4 17.8 17.8
1.5 < D 20.4 20.1 19.1 20.1 20.1
Total 49.3 49.5 43.1 49.5 49.5

Table 3 Comparisons of hydraulics parameters for proposed flood mitigation alternatives based on 50- and 100-year floods at some
selected control cross sections

Hydraulics Return period Control cross Flood mitigation alternatives


parameter (year) sections Alternative 1 Alternative 2 Alternative 3 Alternative 4 Baseline
Flood level (m) 50 (1)* 159.07 158.80 158.07 158.23 159.01
(2) 157.45 156.85 157.03 156.56 157.47
100 (1) 159.20 159.06 158.45 158.53 159.12
(2) 157.62 157.30 157.17 156.93 157.64
Flow velocity (m/s) 50 (1) 1.25 1.26 2.16 1.97 1.30
(2) 2.84 4.40 3.26 4.86 2.73
100 (1) 1.41 1.25 2.05 1.95 1.48
(2) 3.19 3.93 3.72 4.77 3.06
Stream power (N.m/s) 50 (1) 32.07 15.51 74.29 55.94 36.25
(2) 216.99 615.46 244.61 1134.13 196.74
100 (1) 57.22 17.25 44.65 40.43 54.96
(2) 285.25 430.30 355.46 785.99 248.29

*(1) on the outside of the river bend; (2) on the downstream of the residential area.

(Figure 4C). However, alternative 2 is better than other alter- analysis. A scenario analysis was conducted to show the
natives in terms of outside river bend flow velocity importance of each criterion and how they are integrated in
(Figure 4A) and stream power (Figure 4D). As another the decision process. The performance of flood mitigation
example, alternative 3 has the highest flow velocity on the measures, in terms of preventing flood risk and the resulting
outside of the river bend. But it has the lowest flood level in damage and in terms of their economic and environmental
the same control point, which are obviously in conflict benefits, is important to be considered for an overall
(Figure 4A). It can be concluded that an MCDA analysis is appraisal of acceptability of the alternatives (Bana e Costa
needed in this case, as there is conflicting criteria that make et al., 2004). In this section, we examined the effect of vari-
it difficult for an accurate trade-offs analysis. It should be ations in criterion weightings on the final decision. This will
noted that all variables used in Figure 4 were resulted from have practical implications to show how to incorporate
100-year flood inundation analysis in previous sections. stakeholders’ preferences in decision making, which are
Application of TOPSIS as an MCDA method is discussed in most of the time in conflict. Therefore, it is demonstrated
the next section. how final decision is sensitive to different stakeholders’ or
policy makers’ points of view. In this case, stakeholders
involved in the decision-making process could include local
Scenario-based flood mitigation
residents, non-governmental organisations and governmen-
decision analysis
tal agencies.
TOPSIS, an MCDA method, is used for prioritising the best The first scenario was based on the fact that the most
flood mitigation alternatives. Results of a decision analysis important criterion in flood mitigation plans is flood risk
based on four different scenarios for criteria weighting are including flood level and flood inundation area. Therefore,
shown in Table 4. Only those criteria for 100-year flood in this scenario, flood level and flood inundation area were
in two selected control points are considered for decision assigned a weight twice as high as the other criteria in the

© 2012 Blackwell Publishing Ltd and The Chartered Institution of Water and Environmental Management (CIWEM) J Flood Risk Management •• (2013) ••–••
Comparative evaluation of flood mitigation alternatives 11

Figure 4 Comparison of flood mitigation alternatives using pair-wise trade-offs between decision criteria.

Table 4 Priority rating and rank of proposed flood mitigation alternatives based on different decision-making scenarios
Decision-making scenarios for flood risk mitigation
Flood mitigation alternatives Risk-oriented Environmental-oriented Economic-oriented Balance
Alternative 1 (levee construction) 0.716 [1]* 0.637 [2] 0.838 [1] 0.735 [1]
Alternative 2 (channel modification) 0.556 [2] 0.661 [1] 0.518 [2] 0.571 [2]
Alternative 3 (channel modification) 0.298 [3] 0.395 [3] 0.167 [3] 0.278 [3]
Alternative 4 (combination of alternatives 1 and 3) 0.124 [4] 0.184 [4] 0.073 [4] 0.125 [4]

*Numbers in brackets show the rank of each flood mitigation alternative.

decision matrix (Table 1). For the first scenario with higher modification based on alternative 2 performs slightly better
weight to flood risk criteria (inundation area and flood than the levee construction option (Table 4). In the third
level), construction of the levee along Zaremroud River is the scenario, equal weight was assigned to flood risk and envi-
highest rated alternative (0.716). Alternative 4 is the least ronmental impact criteria, but higher weight was considered
preferred practice (0.124), which is the combination of a for economic impact criterion. In this case, levee construc-
levee construction and two stages channel modification. tion is rated as the most preferred flood mitigation alterna-
Channel modifications alternatives 2 and 3 are the second tive (0.838). Scenario four is considered as a balance between
and third rated alternatives (0.556 and 0.298) (Table 4). In all decision criteria including flood risk, environmental and
the second scenario, a higher weight was assigned to envi- economic impacts. In this case, alternative 1 is the most rated
ronmental impact criteria, which included flow velocity and option (0.735). In all scenarios, alternative 4 is the least pre-
stream power at the river bend and downstream control ferred option among flood control measures (Table 4).
points. In this case, alternatives 2 (0.661) and 1 (0.637) are It is shown that the different form of proposed channel
two more preferred options, respectively. However, channel modifications is not a preferred alternative for this river

J Flood Risk Management •• (2013) ••–•• © 2012 Blackwell Publishing Ltd and The Chartered Institution of Water and Environmental Management (CIWEM)
12 Ghanbarpour et al.

reach under study, with the exception in environmental- developments alter streamflow and sediment loads, causing
oriented scenario. Scenario analysis confirms the superiority channel degradation or sedimentation.
of alternative 1 to the other proposed alternatives as well as Other proposed alternatives, such as channel modifica-
the strong weakness of alternative 4. On the other hand, tion with different dimensions, failed to be optimal options
channel modification practices with less volume of machin- in the study area. Although simple dredging as a short-term
ery channel excavation are more preferred, as alternative 2 is solution for reducing flood damages could be considered
superior to alternative 3, considering all weighting scenarios annually, it is neither effective nor cost-efficient. As the
(Table 4). Although increasing the size or depth of the results of this study show, channel modification activities
channel or decreasing its roughness can lead to a reduction change the upstream and downstream stream power balance
in flood levels because of the additional channel capacity, and bring about a greater possibility of channel degradation
channel modifications can also have negative effects, such as downstream because of an increase in flow velocity
an increase in flow velocity and stream power (Table 3). As (Brookes, 1987). Moreover, channel modification activities
Wyzga (1993) stated, an increase in flow velocity and stream diminish the quality and diversity of aquatic and streamside
power owing to river modification results in a progressive habitats by altering river beds and channels, and increase
outwashing of finer grains from bed material, which causes sediment loads below channelisation works (Schoof, 1980;
negative impacts near residential area and downstream. Sherwood et al., 1990). Such negative impacts of channel
From this synthetic assessment, we can conclude that the modification support the superiority of levee construction.
levee construction plan is superior to the other three flood The application of the hydraulic model, HEC-RAS, com-
mitigation measures. As a result, when taking risk, environ- bined with GIS analysis and an MCDA method has allowed
mental and economic impact considerations into account, for the efficient analysis of proposed flood hazard mitigation
alternative 1 is beneficial in protecting residential areas and alternatives. The combination of GIS analysis and hydraulic
contributes less to sedimentation and increasing down- simulation facilitates the process of producing flood hazard
stream stream power and flow velocity. However, in terms of maps (Salimi et al., 2008) and other hydraulics parameters
economic considerations, the construction of a levee adja- like flow velocity, stream power and flood level at any cross
cent to the residential area may not cost as much as the section point along the river reach. The outputs can be trans-
channel excavation for river reach widening and deepening. ferred to an MCDA approach, which provides a basis for
Different criteria-weighting scenarios enable us to take into applying a scenario analysis. The proposed approach in this
consideration all of the evaluation criteria, such as cost- research can be used for flood risk assessment under climate
effectiveness and least environmental impact as well as effec- change and urban development scenarios with the improv-
tiveness in reducing the flood hazard in residential areas. ing of flood damage details (Veerbeek and Zevenbergen,
2009). Moreover, further analysis, such as flood damage
assessment, can be carried out for insurance studies, land use
planning and hydraulics design purposes. The multicriteria
Conclusions approach allows us to include different objectives in the
This paper focused on integrating flood plain inundation decision-making process. Therefore, the evaluation of flood
mapping and MCDA. As flood hazard control measures are mitigation measures performed in this study not only per-
complex, they require a comprehensive evaluation frame- tains to flood risk reduction but also covers economic and
work. The appropriate flood mitigation alternatives in environmental objectives.
reducing the flood risk along Zaremroud River were sug-
gested, modelled and compared. The first step was the devel-
opment of flood hazard maps, which show inundated areas References
and flood depths. The second step was to select the most Bana e Costa C.A., Antao de Silva P. & Correia F.N. Multicrite-
appropriate alternative based on a multicriteria approach. ria evaluation of flood control measures: the case of Ribeira
The effects of the suggested alternatives were evaluated in do Livramento. Water Resour Manag 2004, 18, 263–283. doi:
terms of flood hazard reduction, and environmental and 10.1023/B:WARM.0000043163.19531.6a.
economic impacts under 50- and 100-year flood inundation Beven K.J. Towards a coherent philosophy for modelling the
areas and depths. The most appropriate measure to tackle environment. Proc. Roy. Soc. Lond. A 2002, 458, (2026), 2465–
flood problems along Zaremroud River and Garm Roud 2484.
residential area is to construct a levee as a structural flood Brookes A. River channel adjustments downstream from chan-
mitigation measure. However, it is important to apply the nelization works in England and Wales. Earth Surf Process
best management practice and watershed management as Landf 1987, 12, 337–351. doi: 10.1002/esp.3290120402.
long-term non-structural measures (Simonovic and Carson, Chang H., Franczyk J. & Changhwan K. What is responsible for
2003; Ghanbarpour et al., 2005) because intensive watershed increasing flood risks? The case of Gangwon Province, Korea.

© 2012 Blackwell Publishing Ltd and The Chartered Institution of Water and Environmental Management (CIWEM) J Flood Risk Management •• (2013) ••–••
Comparative evaluation of flood mitigation alternatives 13

Nat Hazards 2009, 48, 339–354. doi: 10.1007/s11069-008- Noman N.S., Nelson E.J. & Zundel A.K. Improved process for
9266-y. floodplain delineation from digital terrain models. J Water
Chen S.J. & Hwang C.L. Fuzzy multiple attribute decision Resour Plan Manage 2003, 129, 427–436. doi: 10.1061/
making, methods and applications. Berlin: Springer-Verlag, (ASCE)0733-9496(2003)129:5(427).
1992, 536. Pappenberger F., Beven K., Horritt M. & Blazkova S. Uncer-
Chow V.T. Open channel hydraulics. New York: McGraw-Hill, tainty in the calibration of effective roughness parameters in
Inc., 1959. HEC-RAS using inundation and downstream level observa-
Correia F.N., Saraiva M.G., Nunes da Silva F. & Ramos I. Flood- tions. J Hydrol 2005, 302, (1–4), 46–69. doi: 10.1016/
plain management in urban developing areas. Part II. GIS- j.jhydrol.2004.06.036.
based flood analysis and urban growth modeling. Water Salimi S., Ghanbarpour M.R., Solaimani K. & Ahmadi M.Z.
Resour Manag 1999, 13, 23–37. doi: 10.1023/ Floodplain mapping using hydraulic simulation model in
A:1008045419517. GIS. J Appl Sci 2008, 8, (4), 660–665.
Deputy of Technical Affairs (DTA). Cost and quantity estimat- Schoof R. Environmental impact of channel modification.
ing manual, for watershed management and natural Water Resour Bull 1980, 16, (4), 697–701.
resources. National Deputy of Planning and Monitoring Sellin R.H.J., Bryant T.B. & Loveless J.H. An improved method
Technical Report No. 100/86952, 146, 2009. (in Persian). for roughening floodplains on physical river models. J
Dilley M., Chen R.S., Deichmann U., Lerner-Lam A.L., Arnold Hydraul Res 2003, 41, 3–14. doi: 10.1080/
M., Agwe J. Natural disaster hotspots: a global risk analysis. 00221680309499924.
Washington, DC: International Bank for Reconstructionand Sherwood C.R., Jay D.A., Harvey R., Hamilton P. & Simenstad
Development/The World Bank and Columbia University, C. Historical changes in the Columbia River Estuary. Progr
2005. Oceanogr 1990, 25, 299–352. doi: 10.1016/0079-
Ghanbarpour M.R. & Hipel K.W. Multi-criteria planning 6611(90)90011-P.
approach for ranking of land management alternatives at Simonovic S.P. & Carson R.W. Flooding in the Red River Basin
different spatial scales. Res J Environ Earth Sci 2011, 3, (2), – lessons from post flood activities. Nat Hazards 2003, 28,
168–177. (2–3), 345–365. doi: 10.1023/A:1022921823614.
Ghanbarpour M.R., Hipel K.W. & Abbaspour K.C. Prioritizing United Nation Development Programme (UNDP). Reducing
long-term watershed management strategies using group disaster risk: a challenge for development. New York: United
decision analysis. Int J Water Resour Dev 2005, 21, (2), 297– Nations Development Programme, Bureau for Crisis Preven-
309. doi: 10.1080/07900620500108528. tion and Recovery, 2004, 146.
Ghanbarpour M.R., Salimi S., Saravi M.M. & Zarei M. Calibra- Veerbeek W. & Zevenbergen C. Deconstructing urban food
tion of river hydraulic model combined with GIS analysis damages: increasing the expressiveness of food damage
using ground-based observation data. Res J Appl Sci Eng models combining a high level of detail with a broad
Technol 2011, 3, (5), 456–463. attribute set. J Flood Risk Management 2009, 2, 45–57. doi:
Hicks D.M. & Mason P.D. Roughness characteristics of New 10.1111/j.1753-318X.2009.01021.x.
Zealand rivers. Wellington, NZ: New Zealand National Insti- Walker W.S. & Maidment D.R. Geodatabase design for FEMA
tute of Water and Atmospheric Research, (Also published by flood hazard studies, CRWR. Online report 06-10, Center for
Water Resources Publications, Highlands Ranch, CO, 1998). Research in Water Resources, University of Texas at Austin.
(1991). 2006.
Hill A.R. The environmental impacts of agricultural land drain- Williams T.M. Incorporating GIS in river hydraulic modeling:
age. J Environ Manage 1976, 4, 251–274. assessing the ability to predict ecological consequences of
Hwang C. & Lin M. Group decision making under multiple crite- river modification on floodplain forests. Hydrology and
ria. Berlin: Springer, 1987. Management of Forested Wetlands Proceedings of the Inter-
Hydrologic Engineering Center (HEC). HEC-GeoRAS, GIS national Conference, 8–12 April. 2006.
tools for support of HEC-RAS. User’s Manual (version: 4.2), Wyzga B. River response to channel regulation: case study of the
U.S. Army Corps of Engineering, Davis, CA., September Raba River, Carpathians, Poland. Earth Surf Process Landf
2009. 2009. 1993, 18, 541–556. doi: 10.1002/esp.3290180607.
Hydrologic Engineering Center (HEC). HEC-RAS river analysis Yi C., Lee J. & Shim M. GIS-based distributed technique for
system. Hydraulic Reference Manual (version: 4.1), U.S. assessing economic loss from flood damage: pre-feasibility
Army Corps of Engineering, Davis, CA., January 2010. 262. study for the Anyang Stream Basin in Korea. Nat Hazards
2010. 2010, 55, 251–272. doi: 10.1007/s11069-010-9524-7.

J Flood Risk Management •• (2013) ••–•• © 2012 Blackwell Publishing Ltd and The Chartered Institution of Water and Environmental Management (CIWEM)

You might also like