Journal of Cleaner Production 242 (2020) 118515

Contents lists available at ScienceDirect

Journal of Cleaner Production

journal homepage: www.elsevier.com/locate/jclepro

Evaluating the effect of urban flooding reduction strategies in

response to design rainfall and low impact development
Pei Hua a, b, Wenyu Yang c, Xiaochen Qi d, Shanshan Jiang a, b, Jiaqiang Xie e, Xianyong Gu f,
Honghao Li g, Jin Zhang d, *, Peter Krebs h
a
SCNU Environmental Research Institute, Guangdong Provincial Key Laboratory of Chemical Pollution and Environmental Safety & MOE Key Laboratory of
Theoretical Chemistry of Environment, South China Normal University, 510006, Guangzhou, China
b
School of Environment, South China Normal University, University Town, Guangzhou 510006, China
c
Chair of Hydrology and Chair of Environmental Economics, Brandenburgische Technische Universita €t, 03046, Cottbus, Germany
d
Institute of Groundwater and Earth Sciences, Jinan University, 510632, Guangzhou, China
e
Low Carbon Ecological Laboratory, Urban Planning and Design Institute of Shenzhen, 518028, Shenzhen, China
f
Key Laboratory of Yangtze River Water Environment, Ministry of Education, College of Environmental Science and Engineering, Tongji University, 200092,
Shanghai, China
g
Institute for Technical and Scientific Hydrology (itwh GmbH), 30167, Hanover, Germany
h
Institute of Urban and Industrial Water Management, Technische Universita €t Dresden, 01062, Dresden, Germany

a r t i c l e i n f o a b s t r a c t

Article history: Rapid urbanisation and climate change-induced watershed hydrologic alteration have increased urban
Received 11 March 2019 flooding risks. As a strategy for urban flood control, low impact development (LID) practices focus on
Received in revised form promoting the infiltration and detention of stormwater runoff. A modelling-based study was conducted
18 September 2019
to analyse LID performance in urban flooding reduction. Two-dimensional simulations were conducted
Accepted 19 September 2019
Available online 20 September 2019
to visualise the flood risk under different precipitation scenarios and LID schemes. An integrated eval-
uation system based on life-cycle cost analysis, the analytic hierarchy process, and regret decision theory
Handling Editor: Yutao Wang were applied to assess the LID performance in multiple aspects. The results indicated that the LID
practices achieved runoff control and flood mitigation to some extent, especially under mild rainfall
Keywords: events. According to the integrated evaluation, urban flooding risk could be sufficiently reduced with the
Low impact development LID combination of bioretention, infiltration trenches, and rain barrels in the given area.
2D simulation © 2019 Elsevier Ltd. All rights reserved.
Urban flooding
Life-cycle cost analysis
Analytic hierarchy process
Regret decision theory

1. Introduction to economies, societies, and human life (Xie et al., 2017).

It is commonly believed that urban flooding is induced by two
Urban flooding is a chronically recurring natural hazard in terms major factors. First, rapid urbanisation has induced the alteration of
of the number of lives lost and the overall cost of damage to watershed hydrology at city scales (Ahiablame and Shakya, 2016),
property, public works infrastructure, and natural resources i.e. natural land is converted into impermeable surfaces, thereby
(Shuster et al., 2005). Flooding accounts for approximately one- resulting in a weakened infiltration process (Zhang et al., 2017).
third of all natural disasters in both developed and developing Second, climate change has caused precipitation patterns to shift
parts of the world (UNISDR, 2012). Floods are also responsible for (Karamouz et al., 2010), i.e. the frequency and magnitude of storm
more than half of all disaster-related fatalities and one-third of the events have heightened (Dong et al., 2017). Both of these factors
economic losses from all natural catastrophes (White, 2000). have led to increased runoff generation and urban flooding risks
Therefore, urban flooding has been regarded as an emerging threat (Baek et al., 2015). Under these circumstances, strategies for urban
flooding reduction have attracted increasing attention during the
past few decades.
Typical flood controls mainly rely on conventional grey infra-
* Corresponding author.
structure, including gutters, storm sewers, tunnels, culverts,
E-mail address: jzhang@jnu.edu.cn (J. Zhang).

https://doi.org/10.1016/j.jclepro.2019.118515
0959-6526/© 2019 Elsevier Ltd. All rights reserved.
2 P. Hua et al. / Journal of Cleaner Production 242 (2020) 118515

detention basins, pipes, and mechanical devices, which are used evaluation system based on life-cycle cost (LCC) analysis, the ana-
collectively in a system to capture and convey runoff (White, 2000). lytic hierarchy process, and regret decision theory was applied to
However, this strategy is prone to failure, because the drainage assess LID practices according to technical, economic, environ-
burden has increased (Burns et al., 2012; Dong et al., 2017). Owing mental, and operational aspects, which could be helpful for
to the pressures of climate change and urbanisation, expanding the decision-making in LID implementation.
available drainage system to relieve the increasing drainage burden
has proved to be costly and unsustainable, especially for developed 2. Materials and methods
urbanised areas (Qin et al., 2013). Consequently, conventional
strategies for flood control face challenges and an efficient, effective 2.1. Study area
alternative is needed.
As an inspiring strategy for urban flood control, green infra- The study area is located in the downtown of Chaohu City, Anhui
structure has received significant research interest in recent years Province, China, as shown in Fig. 1. It is a 7.4 km2 highly urbanised
(Baek et al., 2015). The term green infrastructure is also known as zone with several different land-use types, including for residences,
low impact development (LID) infrastructure in North America and commerce, public utilities, green spaces, and roads. The study area
New Zealand, sustainable urban drainage systems in the United is jointly served by a storm sewer system and a combined sewer
Kingdom, water-sensitive urban design in Australia, and the sponge system, which relocates stormwater runoff to the surrounding
city in China. The term LID is used in this study. Unlike traditional water body through CSO and storm sewer discharge. The average
techniques, LID practices rely on decentralised units to control imperviousness in the study area exceeds 0.7, and most of the
stormwater runoff near the source by imitating the natural hy- stormwater volume in this area is converted into surface runoff,
drology and promoting infiltration and detention in urban water- which causes a large drainage burden and flood risk during rainfall
sheds (Ahiablame and Shakya, 2016; Eckart et al., 2017). Many periods. Taking into account the balance between modelling ac-
contributions to the development and performance analysis of LID curacy and computation cost, the size of the selected urbanised
practices have been made in previous studies by means of multi- zone was ideal for the present study, and was also comparable with
scale experiments and modelling (Qin et al., 2013). Through previous studies (Bisht et al., 2016; Montalto et al., 2007; Zhou
multi-scale experiments, Rushton (2001) reported that permeable et al., 2013). Furthermore, the Chinese government launched the
pavement (PP) at the full scale reduced surface runoff by 50% on Sponge City program in 2014, which aimed to infiltrate, detain, and
average. DeBusk and Wynn (2011) revealed that bioretention (BR) release water to mimic predevelopment hydrology through LID
showed excellent performance on parking lots with a 97% reduction practices (Wang et al., 2017). Therefore, it was important to conduct
in cumulative runoff. Regarding the modelling method, Zhou et al. research in this area from both a technical and operational
(2013) evaluated the performance of an infiltration trench (IT) in a perspective.
3.77 km2 area in Aarhus, Denmark, and concluded that the cost of
flood-induced damage decreased by over 35% (Zhou et al., 2013). 2.2. Modelling
Montalto et al. (2007) evaluated the performance of a green roof
(GR) in a 7 km2 urbanised land in New York, USA, and reported that An urban drainage model was constructed in a 2D hydrological
the control of 26% of combined sewer overflow (CSO) was achieved model (MIKE URBAN, DHI, www.mikepoweredbydhi.com) based
(Montalto et al., 2007). on three parts of the contents, namely model theories, model
Given the abovementioned research methods, most researchers development, and model calibration and validation. Details are
studied flood control by focusing on LID infiltration, detention, and discussed in sections 2.2.1., 2.2.2., and 2.2.3., respectively.
underground pipe flow (Ahiablame and Shakya, 2016; Koudelak
and West, 2008; Wiles and Levine, 2002), whereas the runoff 2.2.1. Model theories
routing process has not been adequately examined. As a process Three theoretical modules were integrated into the model for
that describes the flow routing of surcharged stormwater on a urban flooding simulation: the rainfall-runoff module, pipe flow
catchment surface, it directly visualises the location and magnitude module, and 2D overland flow module (Chang et al., 2015; DHI,
of urban flooding. This process is helpful for conducting targeted 2011).
improvement of the drainage system according to the flooded The rainfall-runoff module, which is dominated by hydrological
zone; thus, it should be analysed through two-dimensional (2D) computation, dynamically describes the process of runoff genera-
modelling (Xie et al., 2017). Furthermore, previous studies evalu- tion. The time-area method was chosen to determine the concen-
ated LID performance by using technical criteria, e.g. the percentage tration time of rainfall runoff in this module because of minimum
of flood reduction and economic criteria, such as engineering data requirements (DHI, 2014).
expense investigation by means of cost-benefit analysis (Zhou et al., The pipe flow module simulates the hydromechanical process of
2013) and cost-effectiveness analysis (Liao et al., 2015). Martin et al. flow routing through sewers. The flow condition in the pipeline can
(2007) pointed out that other performance criteria from environ- be regarded as a one-dimensional unsteady flow that obeys the
mental and operational perspectives should also be considered, and Saint-Venant equations based on three hypotheses: (1) water has a
an integrated evaluation of LID performance is essential before LID constant density; (2) the transportation at a horizontal scale is
implementation. However, owing to the involvement of different much larger than that at a vertical scale, and the influence of ver-
quantification methods for each specific criterion, the integrated tical acceleration and static pressure is neglected; and (3) the flow
evaluation of LID performance requires a systematic framework to condition is subcritical flow (Preissmann, 1961).
combine all of these methods and criteria (Loc et al., 2017), which The 2D overland flow module computes the surface flow routing
introduces difficulties and uncertainties in the LID decision-making of surcharged storm runoff based on a digital elevation model
process. Thus, the objective of this study was to comprehensively (DEM) and vertically integrated flow equations (Michot et al., 2011).
assess the performance of LID practices in urban flooding reduction
based on 2D modelling and integrated evaluation. For tracking 2.2.2. Model development
runoff routing and visualising urban flooding, 2D runoff routing Four types of data were collected for model development:
processes under different scenarios of LID practices and storm terrain condition, land-use distribution, drainage pipe network,
patterns were simulated in a hydrological model. An integrated and hydrological records. Among these, the terrain condition was
 P. Hua et al. / Journal of Cleaner Production 242 (2020) 118515 3

Fig. 1. Location and land-use distribution of the study area.

stored in a shapefile in the form of numerous survey sites, and was network was stored in a shapefile, which contained drainage pipes,
converted into a DEM in ArcGIS. The land-use distribution was manholes, pumps, basins, and discharge outlets. In order to achieve
designed by the Chaohu Planning and Design Institute (Chaohu, the balance of model accuracy and computation cost, the pipe
2013) and stored in a DWG file in CAD (computer-aided design) network was simplified in ArcGIS by removing unconnected man-
software; it was subsequently digitised in ArcGIS. The drainage pipe holes and dead-end pipes with diameters smaller than 0.3 m or
4 P. Hua et al. / Journal of Cleaner Production 242 (2020) 118515

lengths shorter than 10 m. The hydrological record included four events during March were used for calibration, and four events
measured rainfall and the corresponding water level of the pump during April were used for validation. The results showed that the
station forebay with a time step of 1 min. It was saved in a CSV file coefficient of determination (R2) ranged from 0.918 to 0.935 and the
and used for calibration and validation in Microsoft Excel. All the average relative error (HARE) ranged from 8.82% to 9.34%, which
data were provided by the local authorities. After basic data pro- indicated an acceptable model accuracy (Baffaut and Delleur, 1989;
cessing, the terrain condition, land-use distribution, and drainage Peng et al., 2015). The results of the calibration and validation are
pipe network were imported into the 2D hydrological model to shown in Fig. 3, and the best-fit values and empirical ranges in the
construct the drainage model. The pipe network in the model was calibrated parameters are listed in Table 1.
constructed with 44 km of drainage pipeline, 1367 manholes, 61
outlets, 1 basin, and 1 pump. The whole catchment in the model 2.3. Scenario design
was divided into 1423 sub-catchments based on an integrated
consideration of the Thiessen polygon method and urban street To improve the result representativeness, two types of scenarios
layout. Each sub-catchment was assigned an individual impervi- were designed in this study, namely a storm scenario and LID
ousness, hydraulic slope, and connected node according to the implementation scheme.
land-use type, DEM, and pipe network distribution. The con-
structed model is presented in Fig. 2.
2.3.1. Storm scenario
The precipitation pattern can directly affect the runoff genera-
2.2.3. Model calibration and validation tion and should be fully considered. Based on the regulations and
Regarding the model calibration, four global parameters planning guidelines from local authorities, Eq. (1) was applied for
affected the simulation results: initial loss, hydrological reduction the storm design in Chaohu City, and the time-to-peak ratio of
factor, Manning number, and average runoff velocity. The initial loss design rainfall was assigned as 0.375, which meant that the rainfall
is the reduced stormwater due to wetting and depression storage peak occurred at one-third of the whole duration of the storm event
before surface runoff begins, the hydrological reduction factor is the (r ¼ 0.375) (Chaohu, 2013; MOHURD, 2006).
ratio between the impervious area that contributes to runoff gen-
eration and the total impervious area, the Manning number in- 3600 ð1 þ 0:76lgPÞ
q¼ ; (1)
dicates the roughness of the pipeline, and the average runoff ðt þ 14Þ0:84
velocity influences the overland flow process (El-Kafagee and
Rahman, 2011; Zhou et al., 2012). The empirical ranges of these where q is the rainfall intensity (L/(s∙ha)), P is the return period (y),
parameters are available from previous literature, as shown in and t is the rainfall duration (min).
Table 1. Further calibration and validation were required. Fig. 4 illustrates the intensity-duration-frequency curves of the
Based on the available dataset, the model was calibrated and nine storms. According to relevant regulations, the return period of
validated against the measured rainfall events and corresponding a storm sewer system should be 1.5 y in a normal area, and 3e5 y in
water levels in the pump station forebay from the 12th of March to a vital area. In addition, according to the literature (Hirabayashi
29th of April, 2016. More details about the selection of rainfall et al., 2013), the possibility of southeastern China suffering floods
events were given in S1 of Supplementary Material. Among these, with a 10 y return period is increasing owing to climate change.

Fig. 2. Constructed drainage model in the 2D hydrologic model.

 P. Hua et al. / Journal of Cleaner Production 242 (2020) 118515 5

Table 1
Results of calibrated parameters.

Key Parameter Calibrated Value Empirical Range Reference

Initial losses (mm) 1.0 0.5e1.5 (Artina et al., 2007; Thorndahl et al., 2006)
Hydrological reduction factor 0.8 0.6e0.9 (Artina et al., 2007; Thorndahl et al., 2006)
Manning number (m1/3/s) 75 5e75 Artina et al. (2007)
Average runoff velocity (m/s) 0.30 0.25e0.30 Xie et al. (2017)

Fig. 3. Results of model calibration and validation: (a) calibration with rainfall on 30/03/2016, and (b) validation with rainfall on 16/04/2016.

Fig. 4. IDF curves of nine design storms with different durations of: (a) 60 min, (b) 120 min, and (c) 180 min.

Thus, the return periods were set as 1, 5, and 10 y (P ¼ 1, 5, and 10 y) pipeline, and storage tank (Uda et al., 2013). The characteristics of
in this study. Through analysing the rainfall events in recent years, commonly adopted LID practices are summarised in Table 2.
the rainfall durations were set as 60, 120 and 180 min (t ¼ 60, 120, For the decision-making discussion, 10 LID schemes were
and 180 min). assigned based on the consideration of LID characteristics and the
current situation of the study area. More specifically, the imple-
mented LID practices should be suitable for the land-use types in
2.3.2. Low impact development scheme the study area. Table 3 lists the 10 LID schemes with detailed
During the last few years, various LID practices have been installation locations and dimensions. The detailed implementa-
developed, such as BR, GR, IT, PP, rain barrel (RB), and vegetative tion locations for the 10 LID options were determined by the in-
swale (VS). Each has specific characteristics (van Roon, 2007). The tegrated consideration of suitable land-use types and the flooded
IT is an excavated trench filled with stone aggregate, which is zone, as presented in Fig. 5. The background map of land-use and
suitable for capturing runoff and promoting infiltration. An RB flood depth was faded to enhance the clarity and recognition of the
collects stormwater from a building rooftop, and contains three LID location. However, owing to the limitation of spatial resolution,
major elements, namely the collecting system, conveyance
6 P. Hua et al. / Journal of Cleaner Production 242 (2020) 118515

Table 2
Characteristics of six commonly adopted LID Practices: bio-retention (BR), green roof (GR), infiltration trench (IT), permeable pavement (PP), rain barrel (RB), vegetative swale
(VS).

LID Major Reference Installing Reference Installing Runoff Reduction Reference

Practice Function Site Rate Rate

BR Infiltration Landscape 5%e30% 40%e81% (Chapman and Horner, 2010; Che and Yan, 2015; Dietz, 2007; Xie et al., 2017)
GR Retention Rooftop 15%e25% 45%e70% (Ahiablame and Shakya, 2016; Dietz, 2007; Martin-Mikle et al., 2015; Qin et al.,
2013)
IT Infiltration Sidewalk, landscape 5%e20% 36%e77% (Che and Yan, 2015; Liao et al., 2015; Uda et al., 2013)
PP Infiltration Roads, parking lots 8%e40% 20%e75% (Dietz, 2007; Liao et al., 2015; Martin-Mikle et al., 2015; Qin et al., 2013)
RB Retention Building 15%e50% 24%e77% (Ahiablame and Shakya, 2016; Guan et al., 2015; Martin-Mikle et al., 2015; Uda
et al., 2013)
VS Infiltration Roads, parking lots 10%e25% 3%e33% (Che and Yan, 2015; Liao et al., 2015; Uda et al., 2013)

Table 3
Implementing schemes of ten LID options.

LID Option Installing location Installing Percentage Land-use area (m2) Installing Dimension (m2)

BR Green space 15% 851100 127665

IT Commerce 15% 1232633 184895
PP Public utility 15% 2166927 325039
RB Residence 15% 1646740 247011
VS Traffic road 15% 1534240 230136
BR þ IT Green space and Commerce 312560
BR þ RB Green space and Residence 374677
PP þ VS Public utility and Traffic road 555175
BR þ IT þ RB Green space and Commerce and Residence 559571
BR þ PP þ RB Green space and Public utility and Residence 699716

it was only possible to indicate the rough location of the LID op- PVC calculation (Liao et al., 2015). The salvage value was calculated
tions. The specific installation sites should be determined according using Eq. (3).
to the actual conditions. The installation dimensions of the LID  
options were calculated by multiplying the installation percentage L
SV ¼ 1  A Ca ; (3)
by the area of the corresponding land use. In addition, the instal- T
lation dimensions of a combined LID option was the sum value of all
the involved LID practices. where SV is the salvage value in the end year of the lifespan (￥), LA
is the interval from the last maintenance year to the end year of the
lifespan (since LID practices are maintained every year, LA was set as
2.4. Life-cycle cost analysis 1), T is the lifespan (y), and Ca is the annual operation and main-
tenance cost (￥).
LCC analysis is an economic analysis of a proposed LID practice
that considers all the associated costs throughout the lifetime 2.5. Integrated evaluation
(Spatari et al., 2011). From an engineering perspective, three parts
of the associated cost should be considered: capital expenses in the Two methods were employed for integrated evaluation in this
construction stage, operation and maintenance expenses in the study, i.e. the analytic hierarchy process and regret decision theory.
operation stage, and the salvage value in the end stage (Fuller and
Petersen, 1996). 2.5.1. Analytic hierarchy process
To take all the associated costs into consideration, a trans- The analytic hierarchy process is a structured technique for
formation based on a proper discount rate was required, and the organising and analysing complex decisions based on mathematics
present value of cost (PVC) in Eq. (2) was used to calculate the LCC and psychology; it decomposes the primary objective of the
(Liao et al., 2013, 2015). decision-making process into a hierarchy of several easily
comprehendible elements, and then systematically evaluates them
X
T
1 1 in a judgement matrix to calculate their relative weights (Saaty,
PVC ¼ C0 þ Ca ,  SV, ; (2)
t¼0 ð1 þ rÞt ð1 þ rÞT 2008). In this study, the primary objective was to assess the per-
formance of LID in urban flooding reduction, which was decom-
where PVC is the present value of the LCC (￥), C0 is the capital cost posed into six evaluation indicators according to technical,
(￥), T is the lifespan (y), Ca is the annual operation and mainte- economic, environmental, and operational aspects, as introduced in
nance cost (￥), r is the discounting rate (5% in this study), t is the Table 4.
time variation (y), and SV is the salvage value in the end year of the The details of the judgment matrix setup and calculation of the
lifespan (￥). relative weight for each indicator are introduced in Tables A1-A3.
As for the salvage value, it is the term of residual life of the LID
practice at the end of its lifespan. Since the usability of a LID 2.5.2. Regret decision theory
practice may not be completely exhausted when it reaches the Regret decision theory has been widely utilised to evaluate al-
design year, the residual life should be taken into account in the ternatives under multiple uncertainties during the past few
P. Hua et al. / Journal of Cleaner Production 242 (2020) 118515 7

Fig. 5. Implementing location of 10 LID options.

8 P. Hua et al. / Journal of Cleaner Production 242 (2020) 118515

Table 4
Six evaluation indicators from four criteria in LID performance integrated system.

Criteria Technical Economic Environmental Operational

Indicator Runoff Volume Flow Peak Time-to-Peak Delay Life-Cycle Cost Environmental Impact Operational Impact
Removal Reduction
Description The rate of reduced The rate of The proportion of The sum of all associated The level of implemented LID The level of required
runoff volume (m3) reduced peak flow delayed flow peak time expenses for LID changes the current working load for LID
by LID practices (m3/s) by LID point (h) by LID implementation, expressed by surrounding environment in construction and
practices practices present value appearance. maintenance.
Quantification Objective Objective Objective calculation, Objective calculation, Subjective evaluation, Subjective evaluation,
calculation, calculation, presented in % presented in Mil ￥ presented from 1 (low) to 6 presented from 1 (low) to
presented in % presented in % (high) 6 (high)

decades. Instead of obeying the value maximisation principle, practices, the combination of BR þ IT þ RB showed the best runoff
regret decision theory calculates the lost value of alternatives under control capacity among the 10 LID options, with the highest VR of
natural conditions, and the option with the lowest regret value is up to 42.95%, highest flow peak delay of up to 24.22%, and PD of up
chosen as the optimised option (Lindley and Savage, 1973). The to 31.17%. BR had the best performance in VR and PR at mitigation
calculation formula is show in Eq. (4). rates of 18.62% and 16.61%, respectively, which were attributed to
its strong infiltration and evapotranspiration capacity. PP produced
Pi ðbestÞ  Pi the highest PD of up to 36.19% owing to its excellent runoff reten-
Ri ¼ ; (4)
Pi ðbestÞ  Pi ðworstÞ tion and delayed discharge capacities. Overall, the runoff control
capacity of the LID practices generally decreased with the increase
where Ri is the non-unit regret value of the evaluated LID option in storm magnitude, and the combination of BR þ IT þ RB showed
regarding evaluation indicator i, Pi(best) is the best performance the best technical performance among the 10 LID options.
value among the 10 LID options regarding evaluation indicator i, Pi
is the performance value of the evaluated LID option regarding
evaluation indicator i, and Pi(worst) is the worst performance value 3.2. Urban flooding reduction by low impact development practices
among the 10 LID options regarding evaluation indicator i.
The regret value indicates the performance of the evaluated LID The combination of BR þ IT þ RB provided the best performance
option among the 10 LID options; if the regret value is lower, then for runoff control; therefore, its flooding mitigation capacity was
the performance is better. Regarding Pi(best) and Pi(worst), the further verified under nine design storms by means of 2D overland
judging principle can vary among different evaluation indicators. simulations based on the current drainage system and the LID-
For technical indicators, namely the runoff volume removal (VR), implemented drainage system. Through tracking and calculating
flow peak reduction (PR), and time-to-peak delay (PD), the highest the process of surface runoff routing, the flood risk in these two
value among the 10 LID options was Pi(best) and the lowest was systems was visualised in Figs. 6 and 7, respectively. The flooded
Pi(worst). In contrast, for the other three indicators (LCC, environ- area is expressed in blue with varied darkness, which indicates the
mental impact (EI), and operational impact (OI), the lowest value different flood depths above ground level. The terrain condition is
among the 10 LID options was Pi(best) and the highest was presented by contour lines with different numbers, which repre-
Pi(worst). sent the elevation from 0 to 60 m.
As shown in Fig. 6, the flood risk was significant in the current
3. Results and discussion drainage system without LID implementation. Flooding with a
depth range of 0e0.75 m occurred not only around waterlogged
To analyse the capacities of LID practices for runoff control and manholes, but also in some areas far from pipelines owing to the
urban flooding reduction, the constructed model was simulated surface routing process of stormwater runoff. Areas along two
under the designed scenarios followed by integrated evaluation of north-south avenues suffered the most serious flooding because of
LID performances. the relatively low terrain and relatively high surface impervious-
ness. With an average ground elevation of lower than 7 m, most of
3.1. Runoff control by low impact development practices the generated runoff in the surrounding residential and commer-
cial areas flowed into and gathered in this area, which caused a high
Table 5 shows that the specific technical indicators, including drainage burden and finally resulted in waterlogging. In addition,
VR, runoff control capacity, and PD, indicated that the LID practices there was a clear upward trend in the flooding level with the in-
achieved runoff control to some extent. The runoff volume and flow crease in storm magnitude. The area of the flooding zone expanded
peak were reduced by a maximum of 43% and 24%, respectively, and significantly when storm events with higher return periods
the PD was up to 36%. It was proved that the storm magnitude occurred, and a considerable increase in flood depth was observed
directly impacted the runoff control effectiveness. The values of all when the duration of the storm event was prolonged.
three indicators decreased with the increase in rainfall intensity Therefore, an LID combination (BR þ IT þ RB) was implemented
and duration. When a higher-intensity storm occurred, PR became with priority in the flood prone zones. The results in Fig. 7 show
more difficult for the LID practices because of limited infiltration that the implemented LID option achieved flood risk reduction to
capacity. In addition, with the increase in rainfall duration, more some extent. On one hand, implementing LID practices decreased
runoff was generated. Owing to the limitation of retention capacity, the average imperviousness, and thus less surface runoff was
the LID practices were not able to achieve more VR and longer PD generated. On the other hand, the retention function enabled the
under this circumstance. LID practices to capture part of the runoff volume, which relieved
As shown in Table 5, the LID combinations had better perfor- the burden of the drainage system. Meanwhile, as discussed in
mance than individual LID practices because of the larger imple- Section 3.1, the storm magnitude directly affected the effectiveness
mentation scale and mutual complementarity. Among these of flooding reduction. It can be seen that the LID practices were
 P. Hua et al. / Journal of Cleaner Production 242 (2020) 118515 9

Table 5
Runoff control capacity of ten LID options under nine design storms r (tt, PP), which have rainfall duration of tt minutes and return period of PP years.

Design Storm Scenario r(60,1) r(120,1) r(180,1) r(60,5) r(120,5) r(180,5) r(60,10) r(120,10) r(180,10)

Intensity (mm/h) 34.96 21.22 15.55 53.53 32.49 23.81 61.53 37.35 27.36

Depth (mm) 34.96 42.44 46.64 53.53 64.99 71.42 61.53 74.70 82.09

Runoff Volume Removal (%) BR 17.78 18.15 18.62 17.40 17.98 18.11 17.21 17.39 17.90
VS 13.69 13.34 13.31 13.68 13.34 13.31 13.68 13.34 13.31
IT 11.78 11.21 11.21 11.78 11.21 11.21 11.78 11.21 11.21
PP 10.29 9.11 8.98 10.28 9.11 8.98 10.27 9.11 8.98
RB 13.39 12.91 12.85 12.42 11.90 11.78 12.06 11.69 11.57
IT þ BR 29.56 29.36 29.83 29.18 29.19 29.32 28.99 28.60 29.11
RB þ BR 31.17 31.06 31.47 29.82 29.88 29.89 29.27 29.08 29.47
PP þ VS 23.97 22.45 22.29 23.96 22.45 22.29 23.96 22.45 22.29
BR þ IT þ RB 42.95 42.27 42.68 41.60 41.09 41.10 41.05 40.29 40.68
BR þ PP þ RB 41.46 40.17 40.45 40.09 38.99 38.87 39.54 38.19 38.45

Flow Peak Reduction (%) BR 16.61 16.53 16.45 16.35 16.22 16.18 16.14 16.06 16.01
VS 12.63 12.63 12.63 12.03 12.33 12.33 11.85 11.85 11.85
IT 10.34 10.34 10.34 9.48 10.06 10.06 9.48 9.48 9.48
PP 3.57 3.57 3.57 3.57 4.65 4.65 5.10 5.10 5.10
RB 10.34 10.29 10.26 10.15 10.02 9.97 9.84 9.62 9.48
IT þ BR 21.43 21.43 21.43 22.22 22.48 22.48 22.45 22.45 22.45
RB þ BR 19.05 19.05 19.05 18.25 18.60 18.60 18.37 18.37 18.37
PP þ VS 14.41 14.41 14.41 13.81 14.66 14.66 14.40 14.40 14.40
BR þ IT þ RB 24.22 24.22 24.22 23.00 23.63 23.63 23.11 23.11 23.11
BR þ PP þ RB 20.83 20.83 20.83 20.04 20.93 20.93 20.92 20.92 20.92

Time-to-peak Delay (%) BR 15.54 6.40 3.88 10.15 4.18 2.54 8.83 3.64 2.21
VS 12.84 5.29 3.21 8.39 3.45 2.10 7.30 3.01 1.82
IT 24.14 10.06 6.03 15.80 6.61 4.02 13.79 5.75 3.45
PP 36.19 14.91 9.04 23.64 9.73 5.90 20.56 8.47 5.14
RB 24.69 10.17 6.17 16.13 6.64 4.03 14.03 5.78 3.51
IT þ BR 34.50 14.21 8.62 22.53 9.28 5.63 19.60 8.07 4.90
RB þ BR 32.51 13.39 8.12 21.23 8.75 5.31 18.47 5.65 4.62
PP þ VS 29.61 12.20 7.40 19.34 9.73 4.83 16.82 6.93 4.20
BR þ IT þ RB 31.17 12.84 7.79 20.35 8.38 5.09 17.71 7.29 4.42
BR þ PP þ RB 33.76 13.91 8.44 22.05 9.08 5.51 19.18 6.61 4.79

relatively effective under storm events with a 1 y return period and the empirical prices in USD were multiplied 3.61 to convert to CNY.
1 h duration, in which the area and depth of the flood were Besides capital cost and annual cost, the land price can also vary
significantly reduced. However, the flooding reduction effective- significantly among different cities. In this study, the land price was
ness under storm events with higher intensity and longer duration determined as 1800 ￥/m2 according to the statistical data from the
was not significant owing to the limitation of the LID capacity. Natural Resource Department of Anhui Province in 2018. The cap-
ital cost, annual cost, and empirical lifespan were collected from the
3.3. Life-cycle cost analysis literature and local regulations. In order to ensure the consistency
of other variables, the assigned capital cost and annual cost were
According to the implementation dimensions shown in Table 3, the median value of the empirical ranges, and the assigned lifespan
the PVC calculation formula in Eq. (2), and the empirical cost ob- of all LID options was set to 40 y. In addition, the LCC of the com-
tained from the references, the LCCs of the 10 LID options were bined LID option was the sum value of all the involved LID practices.
determined, and are listed in Table 6. To facilitate the calculation,
the dimension of the RB was converted into 513 784 L from 3.4. Integrated evaluation
247 011 m2 by assigning one RB with a volume of 208 L to collect
runoff from a 100 m2 building roof. As introduced in Section 2.5, the integrated evaluation of LID
The empirical capital costs and annual costs of the LID practices practices is a multi-criteria analysis process that involves six eval-
were summarised from recent studies, as shown in Table 6. How- uation indicators from the aspects of technical performance,
ever, owing to the lack of comprehensive local price guidelines, the economy, environment, and society.
empirical cost was in USD. Regarding the conversion between USD Through calculation and numerical evaluation, the performance
and CNY, previous studies tended to adopt a real-time currency values of the six indicators for the 10 LID options were determined
exchange rate (Mei et al., 2018; Xie et al., 2017). However, Krugman under storm event r with a rainfall duration of 120 min and return
(2008) pointed out that the real-time currency exchange rate could period of 5 y, as presented in Table 7. The six indicators were VR, PR,
be misleading in the comparison of living standards among coun- PD, LCC, EI, and OI. Among these, three technical indicators were
tries. In contrast, the purchasing power parity (PPP) index was objectively quantified based on the model simulation results. The
more accurate. The PPP index is a method to measure economic economic indicator, namely LCC, was objectively quantified based
variables among countries according to the real gross domestic on the PVC calculation of LID options.
product (GDP), which can avoid the impact of irrelevant exchange The EI and OI were subjectively evaluated by analysing the
rate variations (Reinert et al., 2009). After calculation based on the structures of implemented LID practices. Specifically, the EI de-
GDP of China and the USA in 2018 (IMF, 2019), the purchasing scribes how much the LID practice changes the surrounding envi-
power rate of 1 USD was equal to 3.61 CNY. Thus, in this study, all ronment in appearance and function; it was determined by
10 P. Hua et al. / Journal of Cleaner Production 242 (2020) 118515

Fig. 6. 2D flood risk maps in current drainage system without LID implementation under nine design storms r (tt, PP), which have rainfall duration of tt minutes and return period
of PP years.

analysing the structure of the implemented LID practice. For water in the bucket after rainfall events, so it was assigned the
instance, when an IT is installed in a sidewalk area, part of the lowest value of 1 among the single LID practices. The OI of the LID
sidewalk is converted into a grass filter strip, curbs, and gravel layer, combination was the sum value of all the involved single LID
which completely changes the appearance and function of the practices.
sidewalk, so it was assigned the highest value of 3 among the single As shown in Table 7, for a single LID practice, the highest values
LID practices. In contrast, a RB is a type of bucket connected to a of EI and OI were assigned as 3.00 due to the consideration of the
building roof through rainwater pipes. It is usually placed next to sum value for LID combinations. More explicit, the highest EI and OI
the building and does not substantially alter the appearance and among LID single practices were 3.00, and the possible highest EI
function of the building; thus, it was assigned the lowest value of 1 and OI among LID combinations were 6.00.
among the single LID practices. The EI of the LID combination was By adopting regret decision theory, the performance values in
the sum value of all the involved LID practices. Table 7 were transformed into regret values in Table 8 from 0.00 to
The OI describes how much work is required for LID construc- 1.00, which indicate the regret level from the lowest to the highest,
tion and maintenance; it was determined by analysing the struc- respectively. Subsequently, the relative weights of the six involved
ture of the implemented LID practice. For instance, a typical IT indicators were calculated using the analytic hierarchy process to
consists of three layers; the surface layer is the grass filter strip and represent how much importance was accounted for by each indi-
curbs, the second layer is the media layer with gravel and soil, and cator during the evaluation. Combining the regret values and
the third layer is the drainage layer with perforated pipes and drain relative weights, the integrated regret values of the 10 LID options
rocks. A monitoring well goes through all three layers. The complex were determined, which are listed in Table 8, and the detailed
structure of the IT leads to a large OI in excavation, material calculation is presented in Table A4.
installation, and daily maintenance; thus, it was assigned the As demonstrated by the results, the 10 LID options performed
highest value of 3 among the single LID practices. In contrast, a RB differently according to the six indicators. Among the individual
has a much simpler structure. Since most buildings are equipped options, BR showed good technical performance, while the RB and
with a vertical rainwater pipe, it is easy to install a RB and connect it PP were low-impact in the aspects of economy, environment, and
to the vertical rainwater pipe to receive stormwater from the society. As for the combined options, the combination of
rooftop; the only required maintenance is regularly emptying the BR þ IT þ RB performed best in runoff control capacity. Although
 P. Hua et al. / Journal of Cleaner Production 242 (2020) 118515 11

Fig. 7. 2D flood risk maps in drainage system with LID implementation under nine design storms r (tt, PP), which have rainfall duration of tt minutes and return period of PP years.

Table 6
Life-cycle cost of ten LID options.

LID Option Installing Unit Empirical Empirical Empirical Local Land Assigned Assigned Assigned Salvage Life- Reference
Dimension Capital Cost Annual Cost Lifespan Price Capital Cost Annual Cost Lifespan Value cycle
(￥/unit) (￥/unit) (year) (￥/m2) (￥/unit) (￥/unit) (year) (￥/unit) Cost (Mil
￥)

BR 127665 m2 729e1151 14.6e23 20e50 1800 940 18.8 40 18.3 393.05 (Liao et al., 2015; Mei
et al., 2018; Uda et al.,
2013)
IT 184895 m2 648e995 13e20 20e50 1800 822 16.5 40 4.6 540.07 (Liao et al., 2015;
Montalto et al., 2007;
Uda et al., 2013)
PP 325039 m2 355e541 7.1e11 8e40 1800 450 9 40 8.8 784.05 (Montalto et al., 2007;
Uda et al., 2013; Xie
et al., 2017)
RB 513784 L 8e12 0.3e0.5 20e40 1800 10 0.4 40 0.4 10.08 (Liao et al., 2015; Uda
et al., 2013; Xie et al.,
2017)
VS 230136 m2 62e180 1.2e3.6 20e50 1800 120 2.4 40 2.3 451.82 (Liao et al., 2015; Mei
et al., 2018; Uda et al.,
2013)
BR þ IT 312560 933.13
BR þ RB 374677 403.14
PP þ VS 555175 1235.87
BR þ IT þ RB 559571 943.21
BR þ PP þ RB 699716 1187.19
12 P. Hua et al. / Journal of Cleaner Production 242 (2020) 118515

Table 7
Performance value of six indicators for ten LID options under design storm event r (120, 5) which has a rainfall duration of 120 min and a return period of 5 years.

LID Option VR PR PD LCC (Mil￥) EI OI

BR 17.98% 16.22% 4.18% 393.05 1.50 2.00

IT 11.21% 10.06% 6.61% 540.07 3.00 3.00
PP 9.11% 4.65% 9.73% 784.05 2.00 2.50
RB 11.90% 10.02% 6.64% 10.08 1.00 1.00
VS 13.34% 12.33% 3.45% 451.82 2.50 1.50
IT þ BR 28.35% 22.48% 9.28% 933.13 4.50 5.00
RB þ BR 29.05% 18.60% 8.75% 403.14 2.50 3.00
PP þ VS 22.45% 14.66% 9.73% 1235.87 4.50 4.00
BR þ IT þ RB 40.26% 23.63% 8.38% 943.21 5.50 6.00
BR þ PP þ RB 38.16% 20.93% 9.08% 1187.19 4.50 5.50

Table 8
Regret value of six indicators for ten LID options under rainfall r (120, 5), which has a rainfall duration of 120 min and a return period of 5 years.

Indicator VR PR PD LCC EI OI Integrated

Regret Value
Weight 0.2622 0.2622 0.2622 0.1031 0.0663 0.0440

LID Option Regret Value

BR 0.72 0.39 0.88 0.31 0.11 0.20 0.57

IT 0.93 0.72 0.50 0.43 0.44 0.40 0.66
RB 0.91 0.72 0.49 0.63 0.00 0.00 0.62
PP 1.00 1.00 0.00 0.00 0.22 0.30 0.55
VS 0.86 0.60 1.00 0.36 0.33 0.10 0.71
BR þ IT 0.38 0.06 0.07 0.75 0.78 0.80 0.30
BR þ RB 0.36 0.26 0.16 0.32 0.33 0.40 0.28
PP þ VS 0.57 0.47 0.00 1.00 0.78 0.60 0.45
BR þ IT þ RB 0.00 0.00 0.22 0.76 1.00 1.00 0.25
BR þ PP þ RB 0.07 0.14 0.10 0.96 0.78 0.90 0.27

the combination of BR þ RB was not as advantageous as urban flooding reduction, 2D modelling with multiple scenarios
BR þ IT þ RB, it still possessed significantly low regret values in the and an integrated evaluation concerning different criteria were
other three aspects because of its lower engineering expense, conducted in the present study.
smaller implementation scale, and easier maintenance. 2D overland simulations were conducted to visualise the urban
Regarding the integrated regret value, the LID combinations had flooding risk under different scenarios. On the one hand, the 2D
smaller values than those of the single LID options, which meant flood maps of the current drainage system showed that urban
that the combined use of LID practices performed better than in- flooding occurred not only near waterlogged manholes, but also in
dividual implementation. The optimal choice was BR þ IT þ RB with areas far from pipelines owing to the surface routing process of
an integrated regret value of 0.25. Despite the high EI and OI of the stormwater runoff. On the other hand, the 2D flood maps of the
IT, the excellent runoff control capacity of BR and low EI and OI of modified drainage system demonstrated the effectiveness of urban
the RB guaranteed the best performance of this combination flooding reduction by LID implementation. Despite the varied
among the 10 LID options. However, owing to its mediocre per- performances among the different LID practices, the flooding
formance in all evaluation aspects, the VS with an integrated regret reduction effectiveness generally displayed an upward trend with
value of 0.71 was ranked in the last position among the 10 LID decreasing storm magnitude and increasing LID implementation.
options. The integrated evaluation based on the LCC analysis, analytic
hierarchy process, and regret decision theory comprehensively
assessed the LID performance according to technical, economic,
3.5. Policy implications
environmental, and operational aspects. The combination of
BR þ IT þ RB was determined as the optimal option with an inte-
This study compared the integrated performances of different
grated regret value of 0.25.
LID practices in urban flooding reduction by means of 2D runoff
The evaluation and modelling approach provided in this study
routing modelling and multi-criteria evaluation. Specifically, the 2D
could be used to assess the effects of engineering drainage projects
simulation results showed the current flooding risk and demon-
and to optimise the final practices to assist with decision-making.
strated the necessity of drainage system improvement, thereby
The potential engineering practices could be tested, analysed, or
illustrating the necessity of implementing LID practices. Moreover,
trained in different scenarios where real-world systems or concepts
the integrated evaluation quantified the effectiveness of LID prac-
can be represented by the models.
tices in urban flooding reduction from technical, economic, envi-
ronmental, and operational aspects, which was helpful for selecting
the optimised LID practice. Generally, this research will provide
Acknowledgements
guidance during the LID decision-making process.
This work was jointly supported by the Managing Water Re-
4. Conclusions sources for Urban Catchments project in the framework of the Sino-
German “Innovation Cluster Major Water” (Grant No.
To comprehensively assess the performance of LID practices in 02WCL1337A-K) and the COLABIS project (Collaborative Early
 P. Hua et al. / Journal of Cleaner Production 242 (2020) 118515 13

Warning Information Systems for Urban Infrastructures; Grant No. Krugman, P.R., 2008. International Economics: Theory and Policy, 8/E. Pearson
Education India.
03G0852A), which were funded by the German Federal Ministry of
Liao, Z., Chen, H., Huang, F., Li, H., 2015. Costeeffectiveness analysis on LID measures
Education and Research (BMBF). The mention of trade names or of a highly urbanized area. Desalin. Water Treat. 56 (11), 2817e2823.
commercial products does not constitute endorsement or recom- Liao, Z.L., He, Y., Huang, F., Wang, S., Li, H.Z., 2013. Analysis on LID for highly ur-
mendation for use. This manuscript has not been subjected to the banized areas' waterlogging control: demonstrated on the example of Caohej-
ing in Shanghai. Water Sci. Technol. 68 (12), 2559e2567.
above agencies' required peer and policy review, and thus does not Lindley, D., Savage, L.J., 1973. The foundations of statistics. Math. Gaz. 57 (401), 220.
reflect the views of the above agencies and no official endorsement Loc, H.H., Duyen, P.M., Ballatore, T.J., Lan, N.H.M., Das Gupta, A., 2017. Applicability of
should be inferred. The authors' deepest gratitude goes to the sustainable urban drainage systems: an evaluation by multi-criteria analysis.
Environ. Syst. Decis. 37 (3), 332e343.
anonymous reviewers for their careful work and thoughtful sug- Martin-Mikle, C.J., de Beurs, K.M., Julian, J.P., Mayer, P.M., 2015. Identifying priority
gestions, which helped to improve this paper substantially. sites for low impact development (LID) in a mixed-use watershed. Landsc.
Urban Plan. 140, 29e41.
Martin, C., Ruperd, Y., Legret, M., 2007. Urban stormwater drainage management:
Appendix A. Supplementary data the development of a multicriteria decision aid approach for best management
practices. Eur. J. Oper. Res. 181 (1), 338e349.
Supplementary data to this article can be found online at Mei, C., Liu, J., Wang, H., Yang, Z., Ding, X., Shao, W., 2018. Integrated assessments of
green infrastructure for flood mitigation to support robust decision-making for
https://doi.org/10.1016/j.jclepro.2019.118515. sponge city construction in an urbanized watershed. Sci. Total Environ. 639,
1394e1407.
References Michot, B., Meselhe, E.A., Rivera-Monroy, V.H., Coronado-Molina, C., Twilley, R.R.,
2011. A tidal creek water budget: estimation of groundwater discharge and
overland flow using hydrologic modeling in the Southern Everglades. Estuarine.
Ahiablame, L., Shakya, R., 2016. Modeling flood reduction effects of low impact
Coastal Shelf Sci. 93 (4), 438e448.
development at a watershed scale. J. Environ. Manag. 171, 81e91.
MOHURD, 2006. Code for Design of Outdoor Wastewater Engineering (in Chinese)
Artina, S., Bolognesi, A., Liserra, T., Maglionico, M., 2007. Simulation of a storm
[S].
sewer network in industrial area: comparison between models calibrated
Montalto, F., Behr, C., Alfredo, K., Wolf, M., Arye, M., Walsh, M., 2007. Rapid
through experimental data. Environ. Model. Softw 22 (8), 1221e1228.
assessment of the cost-effectiveness of low impact development for CSO con-
Baek, S.-S., Choi, D.-H., Jung, J.-W., Lee, H.-J., Lee, H., Yoon, K.-S., Cho, K.H., 2015.
trol. Landsc. Urban Plan. 82 (3), 117e131.
Optimizing low impact development (LID) for stormwater runoff treatment in
Peng, H.-Q., Liu, Y., Wang, H.-W., Ma, L.-M., 2015. Assessment of the Service Per-
urban area, Korea: experimental and modeling approach. Water Res. 86,
formance of Drainage System and Transformation of Pipeline Network Based on
122e131.
Urban Combined Sewer System Model.
Baffaut, C., Delleur, J., 1989. Expert system for calibrating SWMM. J. Water Resour.
Preissmann, A., 1961. Propagation of translatory waves in channels and rivers. In:
Plan. Manag. 115 (3), 278e298.
Proc., 1st Congress of French Association for Computation. AFCAL, Grenoble,
Bisht, D.S., Chatterjee, C., Kalakoti, S., Upadhyay, P., Sahoo, M., Panda, A., 2016.
France, pp. 433e442.
Modeling urban floods and drainage using SWMM and MIKE URBAN: a case
Qin, H.-p.p., Li, Z.-x.x., Fu, G., 2013. The effects of low impact development on urban
study. Nat. Hazards 84 (2), 749e776.
flooding under different rainfall characteristics. J. Environ. Manag. 129,
Burns, M.J., Fletcher, T.D., Walsh, C.J., Ladson, A.R., Hatt, B.E., 2012. Hydrologic
577e585.
shortcomings of conventional urban stormwater management and opportu-
Reinert, K.A., Rajan, R.S., Glass, A.J., Davis, L.S., 2009. The Princeton Encyclopedia of
nities for reform. Landsc. Urban Plan. 105 (3), 230e240.
the World Economy.(Two Volume Set). Princeton University Press.
Chang, T.-J., Wang, C.-H., Chen, A.S., 2015. A novel approach to model dynamic flow
Rushton, B.T., 2001. Low-impact parking lot design reduces runoff and pollutant
interactions between storm sewer system and overland surface for different
loads. J. Water Resour. Plan. Manag. 127 (3), 172e179.
land covers in urban areas. J. Hydrol. 524, 662e679.
Saaty, T.L., 2008. Decision making with the analytic hierarchy process. Int. J. Serv.
Chaohu, U.P.I., 2013. Engineering Plan of Storm Water and Anti-flood in Chaohu (in
Sci. 1 (1), 83e98.
Chinese).
Shuster, W.D., Bonta, J., Thurston, H., Warnemuende, E., Smith, D.R., 2005. Impacts
Chapman, C., Horner, R.R., 2010. Performance assessment of a street-drainage
of impervious surface on watershed hydrology: a review. Urban Water J. 2,
bioretention system. Water Environ. Res. 82 (2), 109e119.
263e275.
Che, S., Yan, W., 2015. Study and Application of Sponge City Based on Urban
Spatari, S., Yu, Z., Montalto, F.A., 2011. Life cycle implications of urban green
Greenland Construction in Shanghai. Shanghai Jiao Tong University Press (in
infrastructure. Environ. Pollut. 159 (8), 2174e2179.
Chinese).
Thorndahl, S., Johansen, C., Schaarup-Jensen, K., 2006. Assessment of runoff
DeBusk, K.M., Wynn, T.M., 2011. Storm-water bioretention for runoff quality and
contributing catchment areas in rainfall runoff modelling. Water Sci. Technol.
quantity mitigation. J. Environ. Eng. 137 (9), 800e808.
54 (6e7), 49e56.
DHI, 2011. Mike Flood 1D-2D ModellingdUser Manual. DHI Water & Environment.
Uda, M., Van Seters, T., Graham, C., Rocha, L., 2013. Assessment of Life Cycle Costs for
DHI, 2014. MIKE URBAN TUTORIALS.
Low Impact Development Stormwater Management Practices. Sustainable
Dietz, M.E., 2007. Low impact development practices: a review of current research
Technologies Evaluation Program. Toronto and Region Conservation Authority,
and recommendations for future directions. Water Air Soil Pollut. 186 (1e4),
Toronto, ON, Canada, pp. 112e112.
351e363.
UNISDR, 2012. How to Make Cities More Resilient: A Handbook for Local Govern-
Dong, X., Guo, H., Zeng, S., 2017. Enhancing future resilience in urban drainage
ment Leaders. Author Geneva.
system: green versus grey infrastructure. Water Res. 124, 280e289.
van Roon, M., 2007. Water localisation and reclamation: steps towards low impact
Eckart, K., McPhee, Z., Bolisetti, T., 2017. Performance and implementation of low
urban design and development. J. Environ. Manag. 83 (4), 437e447.
impact development e a review. Sci. Total Environ. 607e608, 413e432.
Wang, Y., Sun, M., Song, B., 2017. Public perceptions of and willingness to pay for
El-Kafagee, M., Rahman, A., 2011. A study on initial and continuing losses for design
sponge city initiatives in China. Resour. Conserv. Recycl. 122, 11e20.
flood estimation in New South Wales. In: Proceedings of the 19th International
White, W., 2000. Water in Rivers: Flooding, A Contribution To the World Water
Congress on Modelling and Simulation (MODSIM2011): Sustaining Our Future:
Vision. IAHR, Madrid.
Understanding and Living with Uncertainty: Perth Convention and Exhibition
Wiles, J.J., Levine, N.S., 2002. A combined GIS and HEC model for the analysis of the
Centre, Perth, Western Australia, 12-16 December 2011, pp. 3782e3787.
effect of urbanization on flooding: the Swan Creek watershed, Ohio. Environ.
Fuller, S., Petersen, S., 1996. Life-cycle Costing Manual for the Federal Energy
Eng. Geosci. 8 (1), 47e61.
Management Program, NIST Handbook 135.
€a
€, N., Koivusalo, H., 2015. Assessment of LID practices for restoring Xie, J., Chen, H., Liao, Z., Gu, X., Zhu, D., Zhang, J., 2017. An integrated assessment of
Guan, M., Sillanpa
urban flooding mitigation strategies for robust decision making. Environ.
pre-development runoff regime in an urbanized catchment in southern Finland.
Model. Softw 95, 143e155.
Water Sci. Technol. 71 (10), 1485e1491.
Zhang, J., Wu, J., Hua, P., Zhao, Z., Wu, L., Fan, G., Bai, Y., Kaeseberg, T., Krebs, P., 2017.
Hirabayashi, Y., Mahendran, R., Koirala, S., Konoshima, L., Yamazaki, D.,
The influence of land use on source apportionment and risk assessment of
Watanabe, S., Kim, H., Kanae, S., 2013. Global flood risk under climate change.
polycyclic aromatic hydrocarbons in road-deposited sediment. Environ. Pollut.
Nat. Clim. Chang. 3 (9), 816.
229, 705e714.
IMF, 2019. Report for Selected Country Groups and Subjects (PPP Valuation of
Zhou, Q., Mikkelsen, P.S., Halsnæs, K., Arnbjerg-Nielsen, K., 2012. Framework for
Country GDP). Retrieved. (Accessed 10 April 2019).
economic pluvial flood risk assessment considering climate change effects and
Karamouz, M., Hosseinpour, A., Nazif, S., 2010. Improvement of urban drainage
adaptation benefits. J. Hydrol. 414, 539e549.
system performance under climate change impact: case study. J. Hydrol. Eng. 16
Zhou, Q., Panduro, T.E., Thorsen, B.J., Arnbjerg-Nielsen, K., 2013. Adaption to
(5), 395e412.
extreme rainfall with open urban drainage system: an integrated hydrological
Koudelak, P., West, S., 2008. Sewerage network modelling in Latvia, use of Info-
cost-benefit analysis. Environ. Manag. 51 (3), 586e601.
Works CS and storm water management model 5 in liepaja city. Water Environ.
J. 22 (2), 81e87.