Science of the Total Environment 696 (2019) 133764

Contents lists available at ScienceDirect

Science of the Total Environment

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

Adaptation as an indicator of measuring low-impact-development

effectiveness in urban flooding risk mitigation
Jing Song a,b, Rui Yang a, Zheng Chang c, Weifeng Li d,e, Jiansheng Wu a,f,⁎
a
The Key Laboratory for Environmental and Urban Sciences, School of Urban Planning and Design, Shenzhen Graduate School, Peking University, Shenzhen 518055, China
b
The Qianhai Institute for Innovative Research (QIIR), Shenzhen, China
c
Department of Architecture and Civil Engineering, City University of Hong Kong, Hong Kong, China
d
Department of Urban Planning and Design, University of Hong Kong, Hong Kong, China
e
Shenzhen Institute of Research and Innovation, The University of Hong Kong, Shenzhen, China
f
College of Urban and Environmental Sciences, Laboratory for Earth Surface Processes, Ministry of Education, Peking University, Beijing 100871, China

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• A time-dependent method with hydro-

dynamic models is applied to evaluate
resilience.
• Resilience is framed into three capaci-
ties coupled with a human–
environment system.
• Infrastructure adaptation is developed
through a resilience-oriented strategy.
• Adaptation is used as a new indicator of
the effectiveness of low-impact devel-
opment.
• Spatial inequality limits the effects of in-
frastructure resilience improvement.

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

Article history: Frequent and intensive urban flooding requires an extensive adoption of low-impact development (LID) to sup-
Received 1 March 2019 plement traditional drainage infrastructures. Our study conceptualizes the resilient infrastructure framework
Received in revised form 8 May 2019 with a particular reference to adaptation, an adjustment capacity in the social–ecological system to withstand
Accepted 3 August 2019
various natural hazards and absorb negative impacts. We argue that adaption is an indicator for measuring LID
Available online 05 August 2019
effectiveness. A methodological framework is adopted using a time-dependent technique with a hydrodynamic
Editor: Ralf Ludwig inundation model to evaluate LID effectiveness. Results of a case study in Gongming, Shenzhen, China, show that
LID projects can effectively reinforce adaptation capacity. However, spatial inequality and accumulation of differ-
Keywords: ent levels of adaptation are evident. This outcome is due to a relatively low absorption capacity because most
Adaptation areas will have a relatively high recovery capacity but retain a low absorption capacity with the construction of
Resilience LID projects. A relatively mild increase in absorption capacity is due to the quality of man-made infrastructural
Climate change development is conflicting across different areas of Gongming, for example some infrastructures are constructed
Hydrodynamic inundation model by the government, whereas others by developers and villagers. In addition, the topographical factor makes some
Low impact development
areas in Gongming lower-lying than others and is therefore increasingly vulnerable to urban flooding during
rainstorms given the difficulty of discharging the surface runoff, thereby limiting the effectiveness of LID projects.
Furthermore, the spatial inequality of adaptation improvement where LID projects cannot be evenly distributed
within the research area leads to the unequal distribution of adaptation. These findings can confirm that the

⁎ Corresponding author at: The Key Laboratory for Environmental and Urban Sciences, School of Urban Planning and Design, Shenzhen Graduate School, Peking University, Shenzhen
518055, China.
E-mail address: wujs@pkusz.edu.cn (J. Wu).

https://doi.org/10.1016/j.scitotenv.2019.133764
0048-9697/© 2019 Published by Elsevier B.V.
2 J. Song et al. / Science of the Total Environment 696 (2019) 133764
government can practically use adaptation as an indicator in evaluating LID effectiveness and identifying the
problematic stages of drainage resilience in urban flooding risk mitigation.
1. Introduction
Urban areas are hubs of economic and political activities and house
half of the world's population (Hay et al., 2005); however, these urban
areas are increasingly affected by damaging urban flooding events asso-
ciated with global environmental change (Liao, 2012; White, 2013).
Densely built and low-lying coastal cities are particularly vulnerable to
flooding events, which cause substantial economic damage and even
human misery (Mirza, 2003). In this context, the resilience and sustain-
ability of complex urban systems are becoming a major research focus
(Ahern, 2011; Holling, 1973, 1986; Leichenko, 2011).
Holling (1973) defined ecological resilience as the ability of the eco-
system to return to equilibrium or steady state after a disturbance. This
concept was further developed to incorporate a situation that may not
simply revert to the original status.
Resilience denotes withstanding to retain essential processes, struc-
tures, and feedback after a disturbance (Berkes, 2007; Davoudi et al.,
2012; Gunderson, 2000). Essentially, resilience is a comprehensive
strategy of urban hazard mitigation aimed at creating resilient cities. Re-
silient cities can withstand natural hazards through three capacities,
namely, absorption, recovery, and adaptation (Folke, 2006). Moreover,
adaptation is an adjustment capacity in the social–ecological system to
withstand various natural hazards and absorb negative impacts by im-
proving its absorption and recovery capacities (Cosens et al., 2013;
Folke, 2006; Walker et al., 2002). An urban system is a socioecological
system with a certain degree of resilience that allows urban system to
adapt to disturbances to a certain extent (Ahern, 2013; Allen, 2012;
Pattee, 2012). However, given frequent and intensive external distur-
bances, such as urban flooding associated with climate change, explor-
ing the strategies of improving urban system resilience and evaluating
the effectiveness of the strategies that enable the urban system to
adapt to various external disturbances have become urgent concerns
(Kalnay and Cai, 2003; Li et al., 2016a; Ziska et al., 2003).
The resilience of an urban system to urban flooding depends on the
density, quality, and efficiency of man-made and natural drainage infra-
structure (Ahern, 2011; Li et al., 2016b; Smith, 2014; Vedachalam et al.,
2014). When urban flooding occurs, the interactions between natural
and built drainage infrastructure determine a city's functionality and
survival. This occurrence is due to such infrastructure possesses absorp-
tion capacity and recovery capacity; the former helps the city absorb
rainwater during urban flooding, whereas the latter helps the city self-
organize and revert to a new steady state after urban flooding. The gov-
ernment is responsible for learning from their experience with urban
floods to reinforce the adaptation capacity of drainage systems and
thus enable them to absorb and recover from various unpredictable
urban floods (Berkes, 2007; Godschalk, 2003; Li et al., 2016a). Strategies
and technologies that can help the urban surface (e.g., paved streets,
parking lots, sidewalks, and roofs) adapt to surface runoff during
storm events are currently being explored for urbanized areas (Qin
et al., 2013). Low-impact development (LID) projects are resilience-
oriented strategies used to reinforce the adaptation resilience of drain-
age infrastructures (Nicolosi et al., 2018; Pyke et al., 2011).
The Chinese government initiated a strategic measure called
“Sponge City,” which incorporates LID concepts to improve urban drain-
age infrastructure and address urban flooding. The Sponge City initiative
envisions cities as sponges that can absorb, store, drain, and purify rain-
water and then release and recycle the saved rainwater during water
shortages (Chen and Yu, 2015; Qiu, 2015a, 2015b; Yu et al., 2015). Sev-
eral zones in cities are expected to be selected as “sponge bodies” to be
© 2019 Published by Elsevier B.V.
incorporated with LID projects. These zones are projected to be paved
by permeable materials or used as green lands. Cavities and roofs that
can absorb water during rainstorms will also be used (Che et al., 2015;
Chen and Li, 2017; Chen and Yu, 2015; Yu et al., 2015). Although the
Chinese government does not formally state the concept of resilience
in its official documents, the Sponge City initiative has strong implica-
tions on resilience improvement because it strives to enhance the
country's adaptation capability to different urban flooding events.
In contrast to normal construction, constructing LID projects is ex-
pensive. In addition, not every place is suitable for these expensive pro-
jects. Therefore, evaluating the effectiveness of LID practices to improve
drainage infrastructure resilience is an emerging but under-researched
topic. Our study bridges this research gap and quantitatively evaluates
the effectiveness of LID projects for drainage infrastructure with adapta-
tion capacity as the indicator. We hypothesize that (1) LID practices can
improve the absorption and recovery capacities of drainage infrastruc-
ture. Thus, they can reinforce the adaptation capacity of drainage infra-
structure, rather than its basal capacity, which is assumed to be zero. In
addition, the (2) absorption and recovery capacities have spatial varia-
tions for capacity improvement, thereby leading to varying adaptation
capacity of drainage infrastructure.
2. Methodology
2.1. Study area
The study area (Gongming) is located in Guangming New District of
Shenzhen, China (Fig. 1). Gongming is a densely populated urban region
located in the northwest of Shenzhen and close to the eastern bank of
Pearl River estuary. The south subtropical monsoon climate causes sev-
eral rainfall events in Gongming every year. Furthermore, extreme rain-
falls associated with climate change frequently occur in Gongming, thus
resulting in frequent and intensive urban floods (Cai, 2017; Qin et al.,
2013). Gongming covers a total land area of 37.68 km2, of which
26.31 km2 (69.8%) is a built-up region dominated by industrial and res-
idential lands.
Small property housing (urban villages) is its main residential land
use, and most houses are rental homes occupied by peasants. Gongming
was selected as the pilot of Sponge City construction in Shenzhen. Re-
cently, the Ad hoc Plan of Sponge City Construction (Shenzhen
Municipal Government, 2016) is proposed. The plan elaborates the im-
plementation guidelines for Sponge City in several following decades,
including the required percentage of different types of LID projects on
various land uses. The design for such constructions provides us a sce-
nario for simulating the processes of urban flooding with and without
LID practices on defined drainage infrastructure. All these factors con-
tributed to the selection of Gongming as the research area for evaluating
the effectiveness of LID practices in enhancing the adaptation of drain-
age infrastructure.
2.2. Analytical framework
The resilience of drainage infrastructure against urban flooding can
be framed into the following capacities. (1) Absorption indicates the ca-
pacity of drainage infrastructure to absorb a defined amount of rainwa-
ter to maintain its original state and prevent rainwater runoff on urban
surfaces. (2) Recovery refers to the capacity of the drainage infrastruc-
ture to revert to a new equilibrium without runoff again after a
rainwater-led runoff on urban surfaces. (3) After implementing an
 J. Song et al. / Science of the Total Environment 696 (2019) 133764 3

Fig. 1. Study area.

anthropogenic strategy to improve the resilience of drainage infrastruc-
ture, adaptation can be identified; adaptation refers to the capacity of
improved drainage infrastructure to adapt to various unpredictable
heavy rains. Adaptation reflects the potential to absorb and recover to
a new steady state that is close to the original state and prevent damage
from the same type of disturbance (Li et al., 2016b). Here, adaptation
describes a measure, an indicator, and “an adjustment capacity.” This
term denotes a measure that adjusts the coping capacity of the city
and that this capacity can be quantified through an indicator
(i.e., adaptation).

Fig. 2. Analytical framework to understand the creation and formulation of the adaptation of drainage infrastructures.
4 J. Song et al. / Science of the Total Environment 696 (2019) 133764
Drawn from the existing research (Ouyang and Dueñas-Osorio,
2012), Fig. 2 provides an analytical framework to understand the crea-
tion and formulation of the adaptation of drainage infrastructures.
Each engineering system exhibits its own performance, which is a
state required to maintain the necessary functions under a given condi-
tion at a specified time or time interval assuming that all required exter-
nal resources are provided. Suppose the performance of the existing
drainage infrastructure is maintained at a steady-state A0. At t0, an ex-
ternal disturbance, that is, heavy rain in this research, is supposed to at-
tack the drainage infrastructure, and the performance of drainage
infrastructure is graduated to decrease at the post-disturbance transient
state A1 at t1 and then arrive at a new equilibrium state (post-distur-
bance steady state) A2 at t2. The performance of the new equilibrium de-
pends on the emergency repair capacity and the extent of damage
caused to the drainage systems. Typically, the performance of the new
steady state cannot return to the original level given constant loss of
drainage systems.
Effective political intervention by a learning-by-doing government
can improve drainage infrastructure to make it adapt to various external
disturbances. Another external disturbance attacks the improved drain-
age system at to’, which is the same as t0. In this case, the steady-state
performance of the improved drainage system increases to A0’ at t0’,
and the post-disturbance steady state increases to A2’ at t2’. Accordingly,
the performance of the post-disturbance transient state for the im-
proved drainage system is lifted to A1’ at t1’.
On the one hand, absorption stage (Ab = t0 ≤ tAb ≤ t1 and Ab’ = t0’
≤ tAb’ ≤ t1’, before and after improving the drainage system) refers to
the length of time from the beginning of rainfall (at this point, the per-
formance of the drainage system is A0 and A0’, respectively) to the first
appearance of waterlogging (at this point, the performance of the drain-
age system is A1 and A1’, correspondingly). On the other hand, recovery
stage (Re = t1 ≤ tRe ≤ t2 and Re’ = t1’ ≤ tRe’ ≤ t2’, before and after improv-
ing the drainage system) is the length of time from the first appearance
to the fading of waterlogging (at this point, the performance of the
drainage system is A2 and A2’, respectively).
Moreover, adaptation (Ad = t[(Ab’−Ab)+(Re−Re′)], after improving the
drainage system) is the sum of the differences in the absorption (Ab’
− Ab) and recovery time before and after the rainfall (Re − Re’). After
the improvement, the absorption stage lengthens (Ab’ N Ab), and the re-
covery stage will be shortened (Re’ b Re). Therefore, adaptation in the
present research is referred to as a time-dependent evaluation, which
uses the duration of adaptation to reflect the effectiveness of the anthro-
pogenic strategy that is implemented to improve the drainage system.
In the present research, the hydrodynamic inundation model that
couples the stormwater management model (SWMM, 1D) (Huong
and Pathirana, 2013; Wu et al., 2013) and the integrated urban flood
modeling system (IFMS Urban, 2D)1 can be applied to simulate the
time process from the beginning of a rainstorm to the first appearance
of urban flooding until reaching the disappearance of urban flooding
with and without implementing LID practices. The simulated duration
of urban flooding is directly provided through the result tables of the
coupling model.
2.3. Data and model building
First, we must generalize the drainage system and underlying sur-
faces for simulation (Qin et al., 2013). The study area with 4502 pipe-
lines and 1175 nodes was generalized into 577 sub-catchments with
597 pipelines and 653 nodes, which contained 56 outlets and 597 in-
spection nodes. The following types of land use were obtained: (1) resi-
dential land (commercial property), (2) residential land (urban village),
1
A software developed by the China Institute of Water Resources and Hydropower Re-
search, Nanjing Hydraulic Research Institute, Hehai University, Shandong University and
Dongnan University. http://www.iwhr.com/zgskyww/rootfiles/2017/06/16/
1496391626804103-1496391626806566.pdf (Chinese version).
(3) industrial land, (4) park and green land, (5) roads, and (6) others
(Fig. 3). We obtained the spatial distribution of the sewerage pipe net-
work of Gongming (2014) and land use map of Gongming (2013)
from the Shenzhen government.
Second, we set the elevation values for drainage systems to realize
the simulation. The coarse 30 m resolution Digital Elevation Model
(DEM) data are utilized in our research for the initial discretization of
a sub-catchment. The high-resolution DEM in our study area is observed
by applying interpolations (Kriging) to measured scattered elevation
points with high accuracy (0.001 m). The measured scattered elevation
points are survey data, which are nodes of a pipe network with high ac-
curacy (0.001 m) elevation information provided by the Shenzhen
government.
Third, we set the LID scenarios for model simulation. Using 2014 as
the base year, the response time processes to urban flooding were sim-
ulated under the scenarios with and without LID projects to evaluate the
effectiveness of LID projects in urban flooding risk mitigation. We se-
lected the design standards of LID projects settled down in the Ad hoc
Plan of Sponge City Construction (Shenzhen Municipal Government,
2016) as references for designing the scenarios for the simulation and
thus obtaining a similar picture of the LID project construction that
will be gradually constructed in Gongming in the future (Table 1).
Table 1 lists the scenarios of implementing LID projects for the land
use types of our study area. CH denotes the construction proportion of
concave herbaceous field; GR implies the coverage of green roof, and
PP indicates the permeable pavement proportion of sidewalks, parking
lots, and squares of the study area. CHs, GRs, and PPs are all LID strate-
gies applied to supplement the drainage system for improving the
drainage capacities. The percentage listed in Table 1 represents the con-
struction proportion of each type of LID strategies.
Finally, we selected the rainfall events and the corresponding mon-
itoring depths of urban flooding to model, calibrate, and validate the
simulation. A typical rainfall event on May 11, 2014, was selected in ac-
cordance with the year of sewerage system data (2014) for the urban
flooding simulation and parameter calibration. This event led to serious
urban flooding in the study area. After calibration, we used data from a
similar rainfall event in 2016 (May 10, 2016) to validate the simulation.
Rainfall event (2016) is similar to the selected occurrence (2014) that
can lead to similarities in the spatial distribution of the urban flooding
events and also contained the monitoring depths of urban flooding
points to validate the coupling model. Both rainfall records were gath-
ered from the Shenzhen Meteorological Data System (https://data.
szmb.gov.cn/) and covered the entire precipitation processes on an
hourly basis. Their intensities, including the duration and precipitation
per hour of the heavy rain events, are summarized in Table 2. The corre-
sponding urban flooding monitoring data in the 2014 and 2016 events
were obtained from the Shenzhen SanFang (flood, drought, and wind
defense) headquarters of the Guangming New District Urban Construc-
tion Bureau.
3. Results
3.1. Calibration and validation results of the simulation
The precipitation on May 11, 2014, was used to simulate the urban
flooding process. Data of the measured depths and spatial distribution
of the urban flooding on May 11, 2014, were used to calibrate the pa-
rameters and those on May 10, 2016, to validate the model. Data of
the depths of urban flooding in 2014 were collected from seven loca-
tions (Fig. 4), which were applied to determine the parameters of the
SWMM through several calibration interactions. After the calibration,
we obtained the value for each parameter of the coupling model (Ap-
pendix A). Then, we validated the simulation with the depths of urban
flooding in 2016. Three urban flooding points in 2016 (Fig. 6) were se-
lected to validate the simulation. The locations of these points were
identified in accordance with those in 2014. The three points are located
 J. Song et al. / Science of the Total Environment 696 (2019) 133764 5

Fig. 3. Generalization map of land use.

in the old town of Gongming, which did not experience land use
changes and construction in the two years (2014–2016), as confirmed
through a field trip and interview with local residents. Finally, the re-
sults were obtained (Table 3). The simulations demonstrate a relative
error that ranges from 5% to 20% (rainfall event in 2016), excluding
the third point, which satisfies the data requirements of the Standard
for Hydrologic Forecasting in China (GB/T 22482-2008), whose ac-
cepted relative error (%) ranges from 0% to 20%.
The final model was built after calibration and validation. The dura-
tions and spatial distributions of absorption and recovery can be directly
identified from the result tables of “time to concentrate floodwater” and
“flooded duration” in the coupling model. The adaptation duration can
be calculated on the basis of absorption and recovery durations. The
total duration was equally divided into three continuous instances in ac-
cordance with the original data for dichotomizing the duration based on
existing similar research (Barua et al., 2016; Johnson et al., 2016).
Absorption, recovery, and adaptation capacities could be analyzed on
the basis of duration partition (Table 4).
3.2. Effectiveness of the Sponge City initiative in improving infrastructure
resilience
Fig. 5 demonstrates the change in the depths of urban flooding with
(a) and without (b) constructing the LID projects under the assumption
of a rainfall pattern identical to the selected one in 2014.
The spatial distribution of the simulated depths of urban flooding
(Fig. 8) illustrates that several places are previously affected by urban
flooding, mainly in the northern part of the study area, and subse-
quently appear to be unaffected by the assumed scenario of LID projects.
Thus, the LID measures are effective to a certain degree. In addition, the
areas highlighted in red (i.e., regions with a surface runoff deeper than

Table 2
Rainfall intensity for events on May 11, 2014, and May 10, 2016, in the study area.
Table 1
Scenarios of implementing LID projects during a simulation in Gongming. Time (h) 2014-05-11 (mm/h) 2016-05-10 (mm/h)

1 8.4 1.8
Control Residential Residential Industrial Park and
2 35.6 19.8
variablesa land land land green
3 10.1 3.4
(urban village) (commercial lands
4 2.1 12.4
property)
5 0.4 10.0
CH 40% 40% 40% 40% 6 5.0 37.0
GR – 20% 30% – 7 4.6 2.7
PP 50% 50% 50% 50% 8 18.9 0.2
9 23.6
Note: water areas, vacant land, and agricultural land are excluded in the LID projects in ac-
10 3.7
cordance with the Ad hoc Plan of Sponge City Construction (Shenzhen Municipal Govern-
11 3.2
ment, 2016).
a 12 3.2
Note: construction proportion of CH: areas covered by LID projects with elevations
13 11.6
lower than that of surrounding catchments (including CHs, rainwater gardens, infiltration
14 11.3
facilities, and water with poundage action) that account for the total area of green lands.
15 2.8
Coverage of GR: the proportion of GR areas accounts for the total construction area of
16 0.4
the roof. *Industrial lands must construct at least 30% GRs. Coverage of PP: the proportion
17 0.0
of PP areas accounts for the total area of sidewalks, parking lots, and squares.
6 J. Song et al. / Science of the Total Environment 696 (2019) 133764

Fig. 4. Spatial location of urban flooding points in the rainfall events of 2014 and 2016.

0.4 m) are reduced after the LID projects are assumed to be incorporated
into the model. In accordance with the statistical data of the model sim-
ulation, 35% of the places previously affected by surface runoff becomes
free from urban flooding when constructing the LID projects is realized
as the assumption.
For the places where urban flooding is eliminated by LID projects,
absorption capacity will be enlarged, and the absorption duration is pos-
itive for a storm identical to or less intensive than that which occurred
on May 11, 2014. The second stage of resilience cycle (i.e., the recovery
stage) will not happen because urban flooding does not occur at all.
Thus, the adaptation capacity of these places will be enhanced. Conse-
quently, these areas can adapt well to heavy rainstorms, such as that
on May 11, 2014, if the LID projects are realized as the assumption.
Urban flooding will still occur in places other than those mentioned
previously after implementing the LID projects, as assumed. The next
subsection explores and identifies the possible changes in the infra-
structure resilience of these places.
3.3. Performance of absorption capacity
construction, the duration will increase to 2.15–11.21 h. Fig. 6(a) and
(b) depicts the absorption durations with and without constructing
the LID projects, respectively.
In these figures, the durations that reflect the absorption capacity of
places still experiencing urban flooding after constructing the LID pro-
jects will be improved significantly. Places that are highlighted in red
(i.e., those with absorption durations longer than 4.5 h) will increase
in number and are distributed across the entire study area.
The absorption capacities of various places are classified in accor-
dance with different durations into low, intermediate, and high levels.
On the one hand, most places (99.90%) have low absorption capacity
without the LID projects. On the other hand, with LID projects, 11.37%
and 2.05% of the places will have high and intermediate absorption ca-
pacities, correspondingly. Places with a low absorption capacity will
be reduced to 86.58%. Approximately 87% of Gongming will still have
low absorption capacity under the assumed scenario. However, the
total area with a relatively high absorption capacity will be increased
to approximately 283 times its original area.

The statistical data of the simulation reveal that the absorption dura- 3.4. Performance of recovery capacity
tion without constructing the LID projects is 0.00–10.40 h. With the
The statistical results of the model simulation show that the recov-
Table 3 ery durations with and without the LID projects are 0.23–14.83 and
Observed and simulated urban flooding depths. 0–14.64 h, respectively. Fig. 7(a) and (b) displays the spatial simulations
Urban Rainfall events in 2014 Rainfall events in 2016 of the recovery duration with and without the LID projects.
flooding
Observed Modeled RE Observed Modeled RE
site
(unit: m) (unit: m) (%) (unit: m) (unit: m) (%)
ID

1 0.25 0.22 −12 0.20 0.23 15

2 0.55 0.64 16 0.70 0.72 3 Table 4
3 0.50 0.22 −56 0.40 0.38 −5 Capacity levels of different capacities comprising resilience.
4 0.45 0.40 −11 – – –
Capacity levels Low Intermediate High
5 0.20 0.18 −10 – – –
6 0.20 0.17 −15 – – – Absorption capacity 0–3.73 3.74–7.47 7.48–11.21
7 0.20 0.22 10 – – – Recovery capacity 9.85–14.76 4.93–9.84 0–4.92
Adaptation capacity 0.03–7.24 7.25–14.46 14.47–21.65
Note: RE means relative error. “–” means missing data.
 J. Song et al. / Science of the Total Environment 696 (2019) 133764 7

Fig. 5. Simulated maximum depths of urban flooding with (a) and without (b) constructing LID projects.

If the LID projects are constructed, then the recovery duration of the
places with urban flooding will also be reduced, as illustrated in the
comparison of Fig. 7(a) and (b). Areas highlighted in red, which repre-
sent those with durations longer than 8 h, also shrink.
Recovery capacity in accordance with the recovery duration is di-
vided into low, intermediate, and high levels. The total area with low
recovery capacity will reduce from 7.50% to 0.80% after constructing
the LID projects. The total area with a high recovery capacity will in-
crease from 76.38% to 98.19%. These findings demonstrate an evident
improvement in recovery capacity.
These results suggest that more than half of the places in the study
area currently have high recovery capacity, thereby indicating that a

Fig. 6. Spatial distributions of absorption durations with (a) and without (b) constructing LID projects.
8 J. Song et al. / Science of the Total Environment 696 (2019) 133764
Fig. 7. Spatial distributions of recovery durations with (a) and without (b) constructing the LID projects.
relatively short period is required for areas with urban flooding to re-
turn to their normal status. If the vision of a Sponge City is achieved,
then the recovery capacity can be further improved.
3.5. Performance of adaptation capacity
Statistical data from the model simulation indicate that the adapta-
tion duration ranges from 0.03 h to 21.65 h. Thus, the adaptation capac-
ity of the drainage system with LID projects will be improved. Fig. 8
plots the spatial distribution of the drainage system's adaptation to
urban flooding. We find that most flooded places will improve upon
completing the LID projects under the Sponge City initiative, but spatial
variations will remain.
The modeling data show that 90.05%, 9.39%, and 0.56% of these
places have relatively low, intermediate, and high adaptation capacities,
correspondingly. Thus, most of the study area demonstrate adaptations
after implementing the LID projects, but the effects are only slightly
evident.
We conclude that, although all places in Gongming show an in-
creased adaptation capacity after constructing the LID projects, the spa-
tial inequality and accumulations of different levels of adaptation
capacities of drainage infrastructure suggest that other attributes hinder
the adaptation capacity improvement attributed to LID project
construction.
First, the LID is suitable only for a few types of housing. For example,
in this research, we assume that industrial lands and commercial prop-
erties can construct the GRs. Urban villages are exempted by the gov-
ernment to implement GRs. Fig. 8 displays that the adaptation around
the urban villages is not highly reinforced in comparison with that
around industrial lands and commercial properties. This result is due
to the construction quality of urban villages may not satisfy the stan-
dards of GRs, even for renewed village buildings. Another reason relates
to the complicated land ownerships associated with urban village prop-
erties. This outcome makes the identification of the organization that
will pay for the construction cost difficult.
Combining the findings of absorption and recovery, we conclude
that this outcome is due to a relatively low absorption capacity given
that most areas will have a relatively high recovery capacity but retain
a low absorption capacity after constructing the LID projects. A rela-
tively mild increase in absorption capacity is due to the quality of
man-made infrastructural development is conflicting across different
areas of Gongming (some are constructed by the government, whereas
others by developers and villagers). Some areas in Gongming have inad-
equate and unsound drainage systems, thereby making the discharge of
rainwater difficult during urban flooding even after implementing the
LID projects. Furthermore, the topographical factor makes some areas
in Gongming lower-lying than others and is therefore highly vulnerable
to urban flooding during rainstorms given the difficulty of discharging
the surface runoff. For these places, the LID projects have limited effec-
tiveness. Another important reason for the spatial inequality of adapta-
tion improvement is that the LID projects cannot be evenly distributed
within the study area. Such construction must conduct some require-
ments. For example, buildings that are up to the design standards and
suitable structures can build the green on their roofs, thereby increasing
the difficulty for an even LID project distribution.
4. Discussion
4.1. Divergent but complementary information derived from the compo-
nents of drainage infrastructure resilience
Most studies have assessed resilience using only one index, thereby
ignoring that resilience is a dynamic process (Davoudi et al., 2012;
Folke, 2006). Our research strengthens the arguments of previous calls
to assess resilience by considering it as a dynamic process with various
metrics that show different stages of drainage infrastructure, namely,
absorption, recovery, and adaptation. Here, a time-dependent method
is imperative to allow the separation of absorption, recovery, and adap-
tation for a multilevel assessment of resilience in a particular area. The
methodological approach represents a shift toward such a multilevel as-
sessment through a time-dependent evaluation of different stages of
 J. Song et al. / Science of the Total Environment 696 (2019) 133764
Fig. 8. Spatial distribution of adaptation duration after constructing the LID projects.
resilience, which can be transferred to other regions to conduct similar
research.
Our research confirms that assessing resilience based on only one of
these components will inadequately reflect the complex process of re-
silience and cannot help identify the specific problems that limit drain-
age infrastructure resilience. For example, a major finding of this
research is that the spatial inequality in allocating man-made infrastruc-
ture is the main reason for low drainage infrastructure resilience consid-
ering that most places will remain at the low level of absorption with
LID projects but constantly at the high level of recovery with and with-
out LID projects.
4.2. Hydrodynamic inundation model for evaluating drainage infrastruc-
ture resilience
Although extant research has assessed the different components of
resilience (Norris et al., 2008; Walker et al., 2004), to the best of our
knowledge, the present research is the first to assess absorption, recov-
ery, and adaptation empirically with a hydrodynamic inundation model
that couples 1D SWMM and 2D IFMS. In particular, the absorption dura-
tion is the time from the beginning of rainfall up to the first appearance
of waterlogging, which can be identified through the table of “time to
concentrate floodwater” in the integrated hydrodynamic inundation
model. The recovery duration is the time from the first appearance to
the fading of waterlogging, which can be ascertained through the
table of “flooded duration” in the integrated hydrodynamic inundation
model.
The adaptation duration is the sum of duration differences between
the present absorption and recovery and those with an assumed institu-
tional intervention from a learning-by-doing government being real-
ized. Scientific literature has seldom addressed the quantitative
assessment of adaptation capacity in terms of resilience; every system
9
has a certain extent of adaption, but we cannot identify it until an inter-
vention is implemented from the outside through learning to improve
the resilience of the urban system (Davoudi et al., 2012). The present re-
search focuses on urban flooding, a natural disturbance that threatens
human health, and the hydrodynamic inundation model that simulates
such flooding. On the one hand, the hydrodynamic inundation model
includes 2D IFMS, which can be used to simulate the spatial distribution
of urban flooding (Huong and Pathirana, 2013). On the other hand, as an
internationally accepted strategy to improve the resilience of drainage
systems, LID projects can be simulated using the hydrodynamic inunda-
tion model (Wu et al., 2013). All these components facilitate adaptation
assessment with the integrated hydrodynamic inundation model.
4.3. Adaptation as a factor for evaluating the effectiveness of LID projects in
urban flooding risk mitigation
The quantitative assessment of adaptation capacity facilitates the
evaluation of the effectiveness of LID projects in urban flooding risk mit-
igation. Existing research has aimed to verify the effectiveness of LID
projects using inundation depth, area, or time decrease after the LID
projects on a regional scale (Ahiablame et al., 2012; Zhao et al., 2013).
However, the indicator applied for the comparison is constantly a static
factor in a certain moment or a period, such as when the maximum in-
undation depth or the inundation time occurs due to heavy rain
(Ahiablame et al., 2012; Qin et al., 2013). In contrast to the traditional
method for judging the effectiveness of LID projects, adaptation, as an
indicator, can show the dynamic changes in absorption and recovery
after an intervention.
The spatial distribution of adaptation capacity with LID projects and
landscape patterns can be combined using the 2D inundation model,
thereby making the effectiveness judgment intuitive and visualized. At
this point, using adaptation capacity provides an innovative approach
10 J. Song et al. / Science of the Total Environment 696 (2019) 133764
to visualizing data that can reflect the spatial patterns of the effective-
ness of LID projects in urban flooding risk mitigation.
Therefore, mismatches that limit the effectiveness of LID projects
and generate spatial inequality can be identified in terms of adaptation.
For example, in the present research, the spatial distribution of adap-
tation suggests that low-quality buildings, such as those in urban vil-
lages or those that are lower-lying relative to others, will not develop
high adaptation. However, these places must be improved urgently to
reduce the risk of urban flooding because the people who are living in
these areas are economically sensible and have low-quality sewerage
systems. This condition indicates that LID projects alone are insufficient,
especially for these impoverished areas. These findings can motivate
policymakers to explore supplementary strategies to help impoverished
areas cope with and adapt to urban flooding and reduce the environ-
mental inequality encountered by these social groups.
5. Conclusion
The main novelty of this research is to apply a coupling hydrody-
namic inundation model into defining and quantifying a resilience
framework. The multiple stages of resilience as a dynamic process
were considered and evaluated in this research. A GIS and hydrody-
namic inundation model-based perspective provides an opportunity
for exploring new means of assessing and managing drainage resilience
and the effectiveness of LID projects in resilience building. A quantita-
tive analysis of the capacity of different stages of infrastructure resil-
ience for addressing urban flooding, particularly the adaptation
capacity, can provide a quantitative indicator for identifying the spatial
distribution of infrastructure resilience in a locality.
On an empirical basis, the effectiveness of a resilience-oriented ini-
tiative from the central government of China was assessed on a local
basis and used to dispel doubts on effectively implementing this
nation-level measure given the diversity of local contexts in different
communities across the nation. We applied the proposed time-
dependent method to evaluate infrastructure resilience to urban
flooding and explore the adaptation capacity by assuming that con-
structing LID projects will work on drainage infrastructures using the
urban renewal scheme. Although findings suggested the effectiveness
of LID projects, these projects remain limited because the changes to ad-
aptation were made by assuming that all the areas have been under
urban renewal. The findings also concluded that the problems related
to drainage systems must be addressed and that these problems cannot
be offset only by constructing LID projects. Therefore, the priorities of
the strategies of managing urban flooding events must not be confused,
and gray infrastructure with low quality and design standards must be
improved. By then, the LID projects can be built to improve the situation.
Although a local case was adopted in the current research, LID pro-
ject construction, as a measurement, is applied to improve the resilience
of many cities to urban flooding globally. Reasonable arrangements
must be identified for such construction, including the spatial distribu-
tion and types of LID projects because LID is expensive. This method
can be used in other places of various cities, regions, and countries to
evaluate the effectiveness of LID projects and thus provide an opportu-
nity to optimize its construction.
Declaration of Competing Interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influ-
ence the work reported in this paper.
Acknowledgements
This research was supported by the Programs of Science and Tech-
nology of Shenzhen (JCYJ20170412150910443) and the National Natu-
ral Science Foundation of China (Grant No. 41671180).
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.scitotenv.2019.133764.
References
Ahern, J., 2011. From fail-safe to safe-to-fail: sustainability and resilience in the new urban
world. Landsc. Urban Plan. 100 (4), 341–343. https://doi.org/10.1016/j.
landurbplan.2011.02.021.
Ahern, J., 2013. Urban landscape sustainability and resilience: the promise and challenges
of integrating ecology with urban planning and design. Landsc. Ecol. 28 (6),
1203–1212. https://doi.org/10.1007/s10980-012-9799-z.
Ahiablame, L.M., Engel, B.A., Chaubey, I., 2012. Effectiveness of low impact development
practices: literature review and suggestions for future research. Water Air Soil Pollut.
223 (7), 4253–4273. https://doi.org/10.1007/s11270-012-1189-2.
Allen, P.M., 2012. Cities and Regions as Self-organizing Systems: Models of Complexity.
Routledge, London.
Barua, U., Akhter, M.S., Ansary, M.A., 2016. District-wise multi-hazard zoning of Bangladesh.
Nat. Hazards 82 (3), 1895–1918. https://doi.org/10.1007/s11069-016-2276-2.
Berkes, F., 2007. Understanding uncertainty and reducing vulnerability: lessons from re-
silience thinking. Nat. Hazards 41 (2), 283–295. https://doi.org/10.1007/s11069-
006-9036-7.
Cai, H., 2017. Decoding Sponge City in Shenzhen: Resilience Program or Growth Policy?
Massachusetts Institute of Technology, Cambridge
Che, W., Zhao, Y., Li, J., 2015. Explanation of sponge city development technical guide:
basic concepts and comprehensive goals. China Water&Wastewater 31 (8), 1–5.
Chen, Y., Li, N., 2017. “Sponge” concept-discussion on the theory of stormwater manage-
ment construction in urban area. World Construction 5 (2), 6–9.
Chen, Y., Yu, K., 2015. Thought of ancient “sponge city”: experiences of applying water-
adaptive landscape. China Water Resources 17, 19–22.
Cosens, B., Gunderson, L., Arnold, A., Benson, M., Craig, R., DeCaro, D., ... Schlager, E., 2013.
Social-ecological system resilience, climate change, and adaptive water governance.
Project of the NSF Funded National Socio-Environmental Synthesis Center. University
of Maryland, Baltimore, Maryland, USA.
Davoudi, S., Shaw, K., Haider, L.J., Quinlan, A.E., Peterson, G.D., Wilkinson, C., Davoudi, S.,
2012. Resilience: a bridging concept or a dead end? “Reframing” resilience: chal-
lenges for planning theory and practice interacting traps: resilience assessment of a
pasture management system in Northern Afghanistan urban resilience: what does
it mean in planning practice? Resilience as a useful concept for climate change adap-
tation? The politics of resilience for planning: a cautionary note: edited by Simin
Davoudi and Libby Porter. Plan. Theory Pract. 13 (2), 299–333. https://doi.org/
10.1080/14649357.2012.677124.
Folke, C., 2006. Resilience: the emergence of a perspective for social–ecological systems
analyses. Glob. Environ. Chang. 16 (3), 253–267. https://doi.org/10.1016/j.
gloenvcha.2006.04.002.
Godschalk, D.R., 2003. Urban hazard mitigation: creating resilient cities. nat. Hazard. Rev 4
(3), 136–143. https://doi.org/10.1061/(ASCE)1527-6988(2003)4:3(136).
Gunderson, L.H., 2000. Ecological resilience—in theory and application. Annu. Rev. Ecol.
Syst., 425–439 https://doi.org/10.1146/annurev.ecolsys.31.1.425.
Hay, S.I., Guerra, C.A., Tatem, A.J., Atkinson, P.M., Snow, R.W., 2005. Opinion—tropical in-
fectious diseases: urbanization, malaria transmission and disease burden in Africa.
Nat. Rev. Microbiol. 3 (1), 81. https://doi.org/10.1038/nrmicro1069.
Holling, C.S., 1973. Resilience and stability of ecological systems. Annu. Rev. Ecol. Syst. 4
(1), 1–23. https://doi.org/10.1146/annurev.es.04.110173.000245.
Holling, C.S., 1986. The resilience of terrestrial ecosystems: local surprise and global
change. In: Clark, W.C., Munn, R.E. (Eds.), Sustainable Development of the Biosphere.
Cambridge University Press, Cambridge, pp. 292–320.
Huong, H., Pathirana, A., 2013. Urbanization and climate change impacts on future urban
flooding in Can Tho city, Vietnam. Hydrol. Earth Syst. Sci. 17 (1), 379–394. https://doi.
org/10.5194/hess-17-379-2013.
Johnson, K., Depietri, Y., Breil, M., 2016. Multi-hazard risk assessment of two Hong Kong
districts. Int. J. Disaster Risk Reduct. 19, 311–323. https://doi.org/10.1016/j.
ijdrr.2016.08.023.
Kalnay, E., Cai, M., 2003. Impact of urbanization and land-use change on climate. Nature
423 (6939), 528. https://doi.org/10.1038/nature01675.
Leichenko, R., 2011. Climate change and urban resilience. Curr. Opin. Environ. Sustain. 3
(3), 164–168. https://doi.org/10.1016/j.cosust.2010.12.014.
Li, Y., Degener, J., Gaudreau, M., Li, Y., Kappas, M., 2016a. Adaptive capacity based water
quality resilience transformation and policy implications in rapidly urbanizing land-
scapes. Sci. Total Environ. 569, 168–178. https://doi.org/10.1016/j.
scitotenv.2016.06.110.
Li, Y., Li, Y., Wu, W., 2016b. Threshold and resilience management of coupled urbanization
and water environmental system in the rapidly changing coastal region. Environ.
Pollut. 208, 87–95. https://doi.org/10.1016/j.envpol.2015.08.042.
Liao, K.H., 2012. A theory on urban resilience to floods—a basis for alternative planning
practices. Ecol. Soc. 17 (4). https://doi.org/10.5751/ES-05231-170448.
Mirza, M.M.Q., 2003. Climate change and extreme weather events: can developing coun-
tries adapt? Clim. Pol. 3 (3), 233–248. https://doi.org/10.3763/cpol.2003.0330.
Nicolosi, E., Medina, R., Feola, G., 2018. Grassroots innovations for sustainability in the
United States: a spatial analysis. Appl. Geogr. 91, 55–69. https://doi.org/10.1016/j.
apgeog.2017.12.024.
Norris, F.H., Stevens, S.P., Pfefferbaum, B., Wyche, K.F., Pfefferbaum, R.L., 2008. Community
resilience as a metaphor, theory, set of capacities, and strategy for disaster readiness.
 J. Song et al. / Science of the Total Environment 696 (2019) 133764
Am. J. Community Psychol. 41 (1–2), 127–150. https://doi.org/10.1007/s10464-007-
9156-6.
Ouyang, M., Dueñas-Osorio, L., 2012. Time-dependent resilience assessment and im-
provement of urban infrastructure systems. Chaos: An Interdisciplinary Journal of
Nonlinear Science 22 (3), 033122. https://doi.org/10.1063/1.4737204.
Pattee, H.H., 2012. Evolving self-reference: matter, symbols, and semantic closure. In:
Howard, H., Rączaszek-Leonardi, J. (Eds.), Laws, Language and Life: Howard Pattee's
Classic Papers on the Physics of Symbols With Contemporary Commentary. Springer,
Berlin, pp. 211–226.
Pyke, C., Warren, M.P., Johnson, T., LaGro Jr., J., Scharfenberg, J., Groth, P., Main, E., 2011.
Assessment of low impact development for managing stormwater with changing
precipitation due to climate change. Landsc. Urban Plan. 103 (2), 166–173. https://
doi.org/10.1016/j.landurbplan.2011.07.006.
Qin, H., Li, Z., Fu, G., 2013. The effects of low impact development on urban flooding under
different rainfall characteristics. J. Environ. Manag. 129, 577–585. https://doi.org/
10.1016/j.jenvman.2013.08.026.
Qiu, B., 2015a. Concept, pathway, and future development of Sponge City (LID). Construc-
tion. Science and Technology 41 (3), 1–7.
Qiu, B., 2015b. Focus on Sponge City Development - concept, pathway and future devel-
opment of Sponge City (LID). China Investigation and Design 7, 28–41.
Shenzhen Municipal Government, 2016. Ad Hoc Plan of Sponge City Construction.
Smith, N.R., 2014. Building resilient cities in China: the nexus between planning and sci-
ence. Town Plan. Rev. 85 (1), 127–142. https://doi.org/10.3828/tpr.2014.8.
Vedachalam, S., John, M.E., Riha, S.J., 2014. Spatial analysis of boil water advisories issued
during an extreme weather event in the Hudson River Watershed, USA. Appl. Geogr.
48, 112–121. https://doi.org/10.1016/j.apgeog.2014.02.001.
11
Walker, B., Carpenter, S., Anderies1b, J., Abel1b, N., Cumming, G., Janssen, M., ... Pritchard,
R., 2002. Resilience management in social-ecological systems: a working hypothesis
for a participatory approach. Conserv. Ecol. 6 (1), 14. http://www.consecol.org/
vol6/iss1/art14/.
Walker, B., Holling, C.S., Carpenter, S.R., Kinzig, A., 2004. Resilience, adaptability and
transformability in social–ecological systems. Ecol. Soc. 9 (2). http://www.
ecologyandsociety.org/vol9/iss2/art5/.
White, I., 2013. Water and the City: Risk, Resilience and Planning for a Sustainable Future.
Routledge, London.
Wu, J., Thompson, J.R., Kolka, R.K., Franz, K.J., Stewart, T.W., 2013. Using the Storm Water
Management Model to predict urban headwater stream hydrological response to cli-
mate and land cover change. Hydrol. Earth Syst. Sci. 17 (12), 4743. https://doi.org/
10.5194/hess-17-4743-2013.
Yu, K., Li, D., Yuan, H., Fu, W., Qiao, Q., Wang, S., 2015. “Sponge city”: theory and practice.
City Planning Review 39 (6), 26–36.
Zhao, X., Yang, K., Peng, S., Xu, Q., Meng, C., 2013. The Study of Urban Rainstorm
Waterlogging Scenario Simulation Based on GIS and SWMM Model — Take the Exam-
ple of Kunming Dongfeng East Road Catchment Area. Paper Presented at the 2013
21st International Conference on Geoinformatics.
Ziska, L.H., Gebhard, D.E., Frenz, D.A., Faulkner, S., Singer, B.D., Straka, J.G., 2003. Cities as
harbingers of climate change: common ragweed, urbanization, and public health.
J. Allergy Clin. Immunol. 111 (2), 290–295. https://doi.org/10.1067/mai.2003.53.