The research on urban flood resilience will contribute to building a more resilient city and provide valuable reference for municipal decision-makers. There are many frameworks and approaches for empirical studies on what constitutes urban flood resilience and how to evaluate it. In this study, a typical region suffering from flood disaster in China-Jiangsu Province was selected as the study area, including 13 prefecture-level cities. The pressure-state-response (PSR) framework, the projection pursuit based on real-coded accelerated genetic algorithm (RAGA-PP) and the technique for order preference by similarity to an ideal solution based on the Kullback-Leibler distance (KL-TOPSIS) were combined to develop a hybrid multi-criteria approach for assessing urban flood resilience. Then the grey relational analysis obtained the important factors. The results illustrate that (1) the development of each subsystem in the city is uncoordinated, that is, the pressure-subsystem has little influence on urban flood resilience, while the state-subsystem and the response-subsystem have great influence. (2) The urban flood resilience in Jiangsu Province presents obvious polarization trend, that is, Southern Jiangsu is more resilient than Northern Jiangsu. The underlying factors are closely related to the level of economic development. Furthermore, the proposed method provides a practical evaluation approach for other fields.

  • The PSR framework, RAGA-PP and KL-TOPSIS were combined to develop hybrid multi-criteria indices approach for assessing urban flood resilience.

  • The development of each subsystem in the city is uncoordinated.

  • Southern Jiangsu is more resilient than Northern Jiangsu.

  • The underlying factors affecting the urban flood resilience are closely related to the level of economic development.

Graphical Abstract

Graphical Abstract
Graphical Abstract

China has experienced rapid urbanization since the reform and opening up in 1978. The urbanization rate has increased from 20% in 1978 to 64.72% in 2021, leading to a surge in urban built-up areas and a slump in ecological land (Zhang et al. 2019). The city, as a complex social system, is also undergoing constant contradictions and threats, especially the extreme weather caused by rainstorms, floods and other nature-induced disasters (Liang & Ding 2017), which bring serious security risks to already fragile ecosystems. According to the assessment report of World Meteorological Organization (WMO 2020), the future flood risk affected by climate change will mainly be concentrated in cities, and the large-scale flood risk will exist for a long time and become more complex and severe. Many cities in China are built on rivers and lakes, which are more vulnerable to rain and flood disasters. In the flood season of 2020, 378 rivers in the Yangtze River Basin bore over alarm floods, 156 rivers underwent over guaranteed floods and the water levels of 51 rivers have exceeded the highest level in history, involving 14 provinces and cities such as Sichuan, Chongqing and Jiangsu. As rapid urbanization and climate change aggravate urban flood risk, urban flood resilience has become a momentous standard to measure urban safety.

In 2020, the Fifth Plenary Session of the 19th Central Committee of the Communist Party of China proposed the construction of ‘resilient cities’ for the first time, calling for strengthening urban flood control and drainage capacity and building ‘sponge city’ (Jia et al. 2017). The UK government has also urged resilience to be considered at all stages of infrastructure projects (Gallego-Lopez & Essex 2016). The United Nations Human Settlements Programme makes resilience its key strategy (Johnson & Blackburn 2014). Thus, how to improve the ability of urban systems to cope with flood disasters and eliminate the negative impact of flood disasters to the greatest extent, is one of the problems that need to be concerned and properly dealt with in urban development.

The term ‘resilience’ was originally used to describe the ability of a metal to remain stable or return to its original state when impacted by an external force (Alexander 2013). Later, Holling (1973), a theoretical biologist in Canada, defined it as the ability of an ecosystem to maintain normal operation or restore balance when it is destroyed. And then it was introduced into other fields, such as ecological resilience (Zhao et al. 2018), engineering resilience (Wagner & Breil 2013), economic resilience (Wan & Zhai 2017) and institutional resilience (Peyroux 2015). However, there is no consensus on the definition of resilience, as it varies in disciplines and study regions. The importance of resilience has led more and more researchers to apply it to urban flood management, which is of great significance for realizing urban sustainable development and improving the ability of urban flood prevention, mitigation and relief (Liao 2012; Marc 2013). Urban flood resilience has achieved much attention, and refers to the ability of a city to bear floods, restructure physical damage and socio-economic loss, maintain its current identity and adapt to future flood disasters (Bertilsson et al. 2019). How to accurately understand the level of urban flood resilience is an important concern for managers and scholars.

For now, there are some framework and methodology for measuring flood resilience in regulations or guidelines, but different countries and scholars have their own standards and opinions. In addition to some qualitative studies (Sweya & Wilkinson 2020), performance and multi-criteria indices are two types of quantitative approaches for assessing flood resilience, which have been widely advanced in domestic and foreign researches. On the one hand, performance-based methods are often based on the simulation results of hydrological or hydraulic models and are, therefore, more objective, such as flood damage simulation model (Golz et al. 2015), bivariate and statistical models (Ali et al. 2020) and system performance curve (Chen et al. 2021). Wang et al. (2019) used CADDIES model based on two-dimensional cellular automata to simulate urban surface flood, and then to evaluate urban surface flood resilience in urban drainage catchment area. Wang et al. (2021) evaluated flood resilience of urban drainage system based on a ‘do-nothing’ benchmark. Shen et al. (2019) simulated flood impacts with a 1D or 3D hydrodynamic model in coastal urban watersheds and developed a transition index for strengthening flood resilience. Sefton et al. (2022) investigated the hydrological function of the water tank and the acceptability of the water tank to the public in order to improve the urban flood resilience. However, performance-based methods need very high-quality data and the calculations are very complex.

On the other hand, the multi-criteria indices methods, which are comparatively the mainstream of resilience evaluation, usually build a multi-indices system, and then calculate resilience by giving index weight, such as ‘the disaster resilience of place model’ (Cutter et al. 2008), ‘the baseline resilience indicators for community’ (Cutter et al. 2010), ‘the resilience index system of community flood disaster’ (Qasim et al. 2016), ‘the resilience framework for critical infrastructures developed in EU-CIRCLE’ (Vamvakeridou-Lyroudia et al. 2020) and so on. Li et al. (2019) utilized the multi-criteria decision-making method to construct a comprehensive flood resilience evaluation system covering the whole disaster cycle. Zhu et al. (2021) developed a VIKOR and Grey Relational analysis approach to assess the urban flood resilience of Yangtze River Delta in China. Liu et al. (2021) adopted machine learning including support vector regression and the selfish herd optimizer algorithm based on elite reverse learning to evaluate the flood resilience and verified that the method is more reasonable and reliable.

Although these studies make clear that some progress has been made on flood resilience, they are still inadequate. Most studies utilized multi-criteria methods to measure the urban flood resilience from the perspective of engineering resilience, economical resilience, ecological resilience and so on, referring to the famous theoretical framework. Among them, the TOPSIS tool is a common multi-criteria indices approach. The analytic hierarchy process (AHP) is universally applied to calculate the weights of indicators in the TOPSIS tool (Moghadas et al. 2019), which relies on the subjective weight calculation method determined by experts, leading to the uncertainty of the evaluation results. However, the projection pursuit based on real-coded accelerated genetic algorithm (RAGA-PP) can effectively avoid the randomness that the weight of the assessment indicators varies from person to person, and provide marked superiority for the comprehensive assessment of high-dimensional complex problems (Wang & Zhan 2019). And, the Kullback-Leibler (KL) distance with non-symmetry improves TOPSIS tool, so as to avoid the shortage and ensure that the evaluation results are more accurate and reasonable (Hu 2002). Thus, this research represents an attempt to propose a multi-criteria indices-based metric using a hybrid RAGA-PP-KL-TOPSIS approach for assessing urban flood resilience.

Jiangsu, China, a coastal province, is one of the floods and extreme rainfall prone areas. Meanwhile, Jiangsu Province has been trying to improve urban flood resilience due to rapid urbanization. Therefore, taking Jiangsu Province as the study area, this research established the evaluation index system of urban flood resilience based on the pressure-state-response theory (PSR), adopted the RAGA-PP-KL-TOPSIS approach to evaluate urban flood resilience, and discussed the flood resilience level and existing problems of 13 cities in Jiangsu Province.

This research contributes to propose a novel framework and approach for assessing urban flood resilience, as well as some corresponding countermeasures for constructing resilient cities by comparing the spatial distribution and matching of ecological resilience, infrastructure resilience, organizational resilience and the fair distribution of resources.

Study area

Jiangsu Province is located in the Yangtze River Delta, on the eastern coastal area of the Chinese mainland which governs 13 prefecture-level cites (Nanjing, Wuxi, Xuzhou, Changzhou, Suzhou, Nantong, Lianyungang, Huai'an, Yancheng, Yangzhou, Zhenjiang, Taizhou and Suqian). Jiangsu Province has many rivers and lakes across the Yangtze River and Huaihe River. Its topography is mainly composed of plain, water area, low mountains and hills (Figure 1). It belongs to an East Asian monsoon climate and is located in the transition zone between subtropical and warm temperate climate, with abundant rainfall in summer. Due to the relatively complex water system, Jiangsu Province is known as the ‘Flood Corridor’.
Figure 1

Research area.

By the end of 2021, there are 85.05 million permanent residents in Jiangsu Province, where is one of the provinces with the highest population density and comprehensive development level in China. However, with the rapid economic growth and urbanization, the quality of human settlements has been declining. In particular, the continuous expansion of building area leads to a large number of impervious surfaces replacing natural cultivated land, woodland and water surface, which greatly increases the possibility of flood. Table 1 shows the flood disasters in Jiangsu Province in recent years (Ministry of Water Resources of the People's Republic of China 2006-2020).

Table 1

Summary table of flood events in Jiangsu Province, China

YearFlood-affected (10,000 people)Death (people)Affected area of crops (1,000 hec)Economic losses (100 million yuan)
2020 98.00 – 123.6 3.3 
2019 27.1 – – 1.1 
2018 252.71 289.31 22.26 
2017 29.14 – 60.56 4.54 
2016 256.77 – 490.29 107.40 
2015 586.32 – 577.12 148.68 
2014 30.82 – 57.92 1.99 
2013 116.31 – 243.79 9.28 
2012 314.64 – 546.07 76.91 
2011 183.62 292.42 27.48 
2010 233.37 – 527.83 30.30 
2009 186.46 329.40 10.40 
2008 200.05 242.54 15.74 
2007 1,104.20 1,122.70 57.34 
2006 902.50 20 1,173.41 68.28 
YearFlood-affected (10,000 people)Death (people)Affected area of crops (1,000 hec)Economic losses (100 million yuan)
2020 98.00 – 123.6 3.3 
2019 27.1 – – 1.1 
2018 252.71 289.31 22.26 
2017 29.14 – 60.56 4.54 
2016 256.77 – 490.29 107.40 
2015 586.32 – 577.12 148.68 
2014 30.82 – 57.92 1.99 
2013 116.31 – 243.79 9.28 
2012 314.64 – 546.07 76.91 
2011 183.62 292.42 27.48 
2010 233.37 – 527.83 30.30 
2009 186.46 329.40 10.40 
2008 200.05 242.54 15.74 
2007 1,104.20 1,122.70 57.34 
2006 902.50 20 1,173.41 68.28 

The PSR framework

The PSR framework proposed by Organization for Economic Co-operation and Development (OECD 1993) was selected as the theoretical basis for the construction of principal indicators. It embraces three dimensions: pressure, state and response, which can reflect the interaction among economic, social and ecological environment dynamically and systematically. Urban flood resilience is a comprehensive result of stimulation of pressure factors, sensitivity of state factors and adaptability of response factors (Figure 2). The urban flood system will change the state under the stimulation of pressure, which is prone to flood disasters. The critical infrastructures to urban flood can relieve some of the pressure. If the flood pressure is too high, according to the flood signal and warning, the city will take corresponding response to carry out flood risk management, and then react on the pressure-subsystem and state-subsystem.
Figure 2

A theoretical framework of urban flood resilience.

Figure 2

A theoretical framework of urban flood resilience.

Close modal

In the pressure-subsystem, natural and social factors put pressure on the urban system, such as climate, geography and population. The excessive and long-term rainfall will lead to flood disasters. The low and flat terrain may aggravate flood disasters and the crowded population may bring greater economic losses. The state-subsystem mainly reflects the current state of urban flood system and is related to the stimulation of pressure factors. The changes of per capita public green area, density of drainage pipelines and coverage rate of green area developed make urban flood resilience increase or decrease. If the coverage rate of green area developed is low, it is not conducive to the infiltration of water, resulting in flood disaster. In the response-subsystem, social system makes corresponding adjustment measures according to the signals sent by flood disaster. In order to alleviate the flood pressure, human beings can manage flood disaster from various aspects. Some engineering measures have been implemented, such as consolidating flood control works, and strengthening construction of drainage pipelines. Some non-engineering measures have also been taken into consideration, such as the improvement of the ability of flood control management, the perfection of flood control planning, the rational design of urban development layout. These response measures alleviate the flood pressure and improve the capacity of cities to tolerate floods. Therefore, in this research, urban flood resilience refers to the response ability of a city to take engineering and non-engineering measures to resist floods and restructure physical damage and socio-economic loss in order to maintain its current identity and adapt to future flood disasters when the city faces the flood disaster pressure from natural and social systems.

Index system

The individual indicators of urban flood resilience are referred to the relevant literature at home and abroad (Cutter et al. 2010; Asadzadeh et al. 2015; Moghadas et al. 2019). In addition, the international representative index systems were used for reference, such as the ‘MISR’ framework (Verrucci et al. 2012) and ‘City Resilience Index’ framework (Rockefeller & ARUP 2015). Then, the four experts from the Emergency Management Department of Jiangsu Province and the university in the field of disaster management in China reviewed availability, scalability and scientificalness of the selected indicators carefully. The indicators without one of the above principles were deleted from the primary list of evaluation indicators. Finally, 18 indicators were determined to construct the index system of urban flood resilience based on the PSR framework (Table 2).

Table 2

The index system based on PSR framework

Target LayerRule LayerIndex Layer
IndicatorDescriptionAttribute
Urban flood resilience index Pressure C1: Annual precipitation (mm) Average annual precipitation negative 
C2: Dependent population (%) Percent of the population aged 0–14 and over 65 negative 
C3: Population density (Person/km2Permanent residents/area negative 
State C4: Per capita public green area (m2Public green area/permanent residents positive 
C5: Density of drainage pipelines (km/ km2Length of drainage pipelines/built-up area positive 
C6: Per capita urban road area (m2Urban road area/urban permanent residents positive 
C7: Building density (km2Build-up area/land area negative 
C8: Construction investment (100 million yuan) Investment in fixed assets negative 
C9: Coverage rate of green area developed (%) Green area/build-up area positive 
C10: Per capita housing area (m2Housing area/permanent residents negative 
Response C11: Communication capacity (%) Percent of households with telephone service availability positive 
C12: Medical care capacity (one) No. of health agency beds per 10,000 people positive 
C13: Financial input (%) Percent of public security to government expenditure positive 
C14: Educational attainment equality (%) [(100 - (percent population with no high school – percent population with collage education))/(percent population with no high school + percent population with collage education)] positive 
C15: Per capita GDP (10,000 yuan) GDP/permanent residents positive 
C16: Proportion of tertiary industry (%) Industrial structure positive 
C17: Allocation capacity of flood-fighting materials Allocation capacity of flood-fighting materials decided by experts positive 
C18: Public awareness of disaster emergency Public awareness of disaster emergency decided by experts positive 
Target LayerRule LayerIndex Layer
IndicatorDescriptionAttribute
Urban flood resilience index Pressure C1: Annual precipitation (mm) Average annual precipitation negative 
C2: Dependent population (%) Percent of the population aged 0–14 and over 65 negative 
C3: Population density (Person/km2Permanent residents/area negative 
State C4: Per capita public green area (m2Public green area/permanent residents positive 
C5: Density of drainage pipelines (km/ km2Length of drainage pipelines/built-up area positive 
C6: Per capita urban road area (m2Urban road area/urban permanent residents positive 
C7: Building density (km2Build-up area/land area negative 
C8: Construction investment (100 million yuan) Investment in fixed assets negative 
C9: Coverage rate of green area developed (%) Green area/build-up area positive 
C10: Per capita housing area (m2Housing area/permanent residents negative 
Response C11: Communication capacity (%) Percent of households with telephone service availability positive 
C12: Medical care capacity (one) No. of health agency beds per 10,000 people positive 
C13: Financial input (%) Percent of public security to government expenditure positive 
C14: Educational attainment equality (%) [(100 - (percent population with no high school – percent population with collage education))/(percent population with no high school + percent population with collage education)] positive 
C15: Per capita GDP (10,000 yuan) GDP/permanent residents positive 
C16: Proportion of tertiary industry (%) Industrial structure positive 
C17: Allocation capacity of flood-fighting materials Allocation capacity of flood-fighting materials decided by experts positive 
C18: Public awareness of disaster emergency Public awareness of disaster emergency decided by experts positive 

The pressure-subsystem brings stimulation to the urban flood resilience system. The greater the pressure is, the weaker the urban flood resilience is. The 13 prefecture-level cities in Jiangsu Province are all located on the plain, with an average elevation of less than 50 m. Thus, geographical factors were not taken into account. The rainfall and population were selected as pressure indicators. The state-subsystem reflects the sensitivity of the urban flood resilience, which mainly considers the vulnerability of ecological environment and the exposure of social property. The more sensitive the state is, the stronger the urban flood resilience is. There are 7 indicators such as per capita public green area and density of drainage pipelines in the state-subsystem. The response-subsystem measures the emergency response ability, post-disaster recovery ability and social service ability of cities subjected to flood disasters. The higher the value is, the better the urban flood resilience is. Among them, allocation capacity of flood-fighting materials and public awareness of disaster emergency were scored by the four experts. They measured the level of these two indicators in each city by scoring. The scores range from 1 to 10. Afterward, this study calculated the average values given by the four experts as the final data of these two indicators.

All in all, there are 12 positive indicators and six negative indicators in the index system. For the positive indicators, the larger the value is, the stronger the urban flood resilience will be. On the contrary, for the negative indicators, the smaller the value is, the weaker the urban flood resilience will be.

Methodology

RAGA-PP model

Projection pursuit (PP) is a statistical method used in processing and analyzing high-dimensional data, especially nonlinear and non-normal data (Wang et al. 2017). Common optimization algorithms cannot converge to the global optimal solution, but the real-coded accelerated genetic algorithm (RAGA) can accelerate the cyclic process and increase the ability of global optimization in case of high dimension. Thus, the RAGA-PP model was built to determine the weights of urban flood resilience indicators. The modeling steps are as follows (Wang et al. 2019):

Step 1: Determination of sample set and data processing. The sample set is , in which is the index in the sample. n and m are the sample and index number, respectively. Then all data are processed by the min-max normalization.

Step 2: Construction of the projection index function. The m dimension data is transformed to the optimal direction vector , then projection value can be expressed as:
formula
(1)
When the multidimensional data is transformed to a one-dimensional space, the local projection points should be kept as dense as possible, so that a number of point groups are formed, while these point groups as a whole dispersed as possible. Based on this, the projection index function can be expressed as:
formula
(2)
where and represent the local density and standard deviation of , respectively.
formula
(3)
formula
(4)
formula
(5)
where represents the mean value of ; stands for the window radius of , represents the sample interval, and is a unit step function. When ,; when , .
Step 3: Optimization of the projection index function.
formula
(6)
The solution of the best projection direction is a very complex nonlinear optimization problem. In this study, RAGA was selected, which simulates the rule of survival of the fittest and the mechanism of chromosome information exchange within the population. The specific steps are shown in Figure 3.
Figure 3

The flow chart of RAGA model.

Figure 3

The flow chart of RAGA model.

Close modal

Step 4: Determination of weight. The best projection direction vector is obtained. Therefore, can be used as the weight of each evaluation index.

KL-TOPSIS model

TOPSIS is a technology widely used in multi criteria decision-making process, which can consider an unlimited number of alternatives and criteria. The conventional TOPSIS tool cannot distinguish the advantages and disadvantages of the points on the vertical line between positive and negative ideal point. However, the KL- TOPSIS model can solve this problem, and the modeling steps are as follows (Hu 2002):

Step 1: The construction of decision matrix. B is a decision matrix composed of . Then, normalized decision matrix can be calculated.

Step 2: The construction of weighted normalized decision matrix. The weights derived from the calculation of RAGA-PP model.
formula
(7)
Step 3: Calculation of the positive ideal and negative ideal solutions.
formula
(8)
formula
(9)
where is the set of income-type indicators and is the set of cost-type indicators.
Step 4: Calculation of the KL distance.
formula
(10)
formula
(11)
Step 5: Calculation of the relative closeness of each city to the positive ideal solution. When is closer to 1, it is closer to the positive ideal point, indicating that the urban flood resilience level is higher; when is closer to 0, it is closer to the negative ideal point, indicating that the urban flood resilience level is lower.
formula
(12)

Grey relational analysis

It is necessary to judge the important indicators effecting urban flood resilience. This study used grey relational analysis to diagnose important factors. Grey relational analysis is a method to measure the correlation degree between factors. If the correlation degree between the resilience and all indicators is greater than 0.5, the correlation is close. Moreover, the indicators with correlation degree greater than 0.7 will be regarded as important factors affecting urban flood resilience (Liu & Li 2006).

Data preparation

This study collected rainfall data, socio-economic data and urban infrastructure data of 13 prefecture-level cities in Jiangsu Province. The data of educational attainment equality consists of the amount of resident population, percent population with no high school and percent population with college education were mainly derived from the ‘Sixth National Population Census’. The data on the proportion of the population by age group was also from the ‘Sixth National Population Census’. Other data were obtained from Jiangsu Statistical Yearbook in 2021 (http://tj.jiangsu.gov.cn/index.html) and cities statistical yearbook (http://tjj.nanjing.gov.cn/bmfw/njtjnj/). Other qualitative data was obtained by the Delphi method, including allocation capacity of flood-fighting materials and public awareness of disaster emergency.

Index weight analysis

According to the RAGA-PP steps, MATLAB2018b was applied for programming and modeling, in which some parameters were set: the population size N = 400, the cross-probability Pc = 0.8, the variation probability Pm = 0.2, the number of optimization variables n = 18, the random number required for the variation direction M = 10, the acceleration times = 7, DaiNo = 2, and ads = 1 (Wang et al. 2019). DaiNo shows that after two generations of evolution, it will speed up the limit number at one time. Then it can obtain the optimal projection direction vector and the projection index function through several iterations. The larger the value of the optimal projection vector is, the greater the contribution of the index to the evaluation system is. Then, according to the fact that the sum of squares of the projection direction sub-vector is 1, the weight of each index can be obtained, as shown in Table 3, which reflects the fluence of each indicator on the urban flood resilience.

Table 3

Influence of indicators

IndicatorsC1C2C3C4C5C6C7C8C9
Optimum projection vector 0.113793 0.313258 0.072868 0.233674 0.15616 0.268319 0.075296 0.068901 0.358862 
Weight 0.012949 0.09813 0.00531 0.054603 0.024386 0.071995 0.00567 0.004747 0.128782 
IndicatorsC10C11C12C13C14C15C16C17C18
Optimum projection vector 0.115623 0.258685 0.329097 0.258731 0.315769 0.344369 0.164198 0.183759 0.2426232 
Weight 0.013369 0.066918 0.108305 0.066942 0.09971 0.11859 0.026961 0.033767 0.0588660 
IndicatorsC1C2C3C4C5C6C7C8C9
Optimum projection vector 0.113793 0.313258 0.072868 0.233674 0.15616 0.268319 0.075296 0.068901 0.358862 
Weight 0.012949 0.09813 0.00531 0.054603 0.024386 0.071995 0.00567 0.004747 0.128782 
IndicatorsC10C11C12C13C14C15C16C17C18
Optimum projection vector 0.115623 0.258685 0.329097 0.258731 0.315769 0.344369 0.164198 0.183759 0.2426232 
Weight 0.013369 0.066918 0.108305 0.066942 0.09971 0.11859 0.026961 0.033767 0.0588660 

Among the evaluation indicators selected, the impact of coverage rate of green area developed (C9) is stronger, which explains that C9 is extremely significant to urban flood resilience, while the impact of construction investment (C8) is the weakest. The order of weights of all indicators from large to small is as follows: coverage rate of green area developed (C9), per capita GDP (C15), medical care capacity (C12), educational attainment equality (C14), dependent population (C2), per capita urban road area (C6), financial input (C13), communication capacity (C11), public awareness of disaster emergency (C18), per capita public green area (C4), allocation capacity of flood-fighting materials (C17), proportion of tertiary industry (C16), density of drainage pipelines (C5), per capita housing area (C10), annual precipitation (C1), building density (C7), population density (C3), and construction investment (C8).

By adding the corresponding index weights, the weights of pressure, state and response subsystems on the urban flood resilience evaluation are 11.64%, 30.36% and 58.00%, respectively. In general, the pressure-subsystem has little influence on urban flood resilience, but the state and the response subsystems have great influence. This also reflects the characteristics of cities, that is, in the face of external disasters, cities can only resist risks by reducing sensitivity and improving adaptability, and it is difficult to change the stimulation of pressure. Meanwhile, it also reveals that the state of urban infrastructure, the emergency response and the post-disaster recovery ability are important factors to measure the level of urban flood resilience. Therefore, urban infrastructure, ecological environment protection and economic development are the key to improve urban flood resilience.

Resilience score and ranking

After assigning weights to all indicators by RAGA-PP technique, the KL-TOPSIS method was adopted to calculate the relative closeness, so as to measure the flood resilience level of each city comparatively. Table 4 demonstrates the final scores and ranking for the total resilience and the results of its three subsystems. The ranking of urban flood resilience is in descending order.

Table 4

Resilience scores and the ranking of cities

CitiesResilienceRankPressureRankStateRankResponseRank
Nanjing 0.851928 0.17452 13 0.407262 0.982585 
Wuxi 0.758439 0.360437 10 0.861143 0.771005 
Xuzhou 0.180499 0.7757 0.147802 11 0.139351 
Changzhou 0.618328 0.437378 0.45841 0.671765 
Suzhou 0.647723 0.232571 11 0.402417 0.752339 
Nantong 0.510399 0.795137 0.888902 0.40005 
Lianyungang 0.159744 11 0.73011 0.101217 12 0.123183 11 
Huai'an 0.155019 10 0.592557 0.363296 0.082424 12 
Yancheng 0.136187 12 0.496806 0.066423 13 0.127253 10 
Yangzhou 0.51101 0.429425 0.610777 0.496487 
Zhenjiang 0.480989 0.161319 12 0.426051 0.522461 
Taizhou 0.751287 0.652696 0.544374 0.796982 
Suqian 0.076714 13 0.771927 0.21992 10 0.020462 13 
CitiesResilienceRankPressureRankStateRankResponseRank
Nanjing 0.851928 0.17452 13 0.407262 0.982585 
Wuxi 0.758439 0.360437 10 0.861143 0.771005 
Xuzhou 0.180499 0.7757 0.147802 11 0.139351 
Changzhou 0.618328 0.437378 0.45841 0.671765 
Suzhou 0.647723 0.232571 11 0.402417 0.752339 
Nantong 0.510399 0.795137 0.888902 0.40005 
Lianyungang 0.159744 11 0.73011 0.101217 12 0.123183 11 
Huai'an 0.155019 10 0.592557 0.363296 0.082424 12 
Yancheng 0.136187 12 0.496806 0.066423 13 0.127253 10 
Yangzhou 0.51101 0.429425 0.610777 0.496487 
Zhenjiang 0.480989 0.161319 12 0.426051 0.522461 
Taizhou 0.751287 0.652696 0.544374 0.796982 
Suqian 0.076714 13 0.771927 0.21992 10 0.020462 13 

Figure 4 shows Nanjing is the most resilient city in Jiangsu Province according to the criteria used here. The ranking of state-subsystem is at medium level, but Nanjing has the least pressure and the highest response. The main reasons for this result are that Nanjing as capital of Jiangsu Province has better infrastructure, the highest level of educational attainment equality, the ascendant capacity to allocate flood-fighting materials, strong public awareness of disaster emergency, and most developed medical care and communication services.
Figure 4

Scores and ranking.

Figure 4

Scores and ranking.

Close modal

In addition, Wuxi, Taizhou, Suzhou and Changzhou follows closely. These cities all have a high level of economic development and good infrastructure. Although the population density is high and the rainfall is heavy, the independent population accounts for the very great proportion, which makes the population more stable. The success of cities in adapting to these pressures has resulted in less impact on the urban flood resilience system.

Suqian is the least resilient city, while Yancheng, Lianyungang, Huai'an, and Xuzhou are ranking as the poorer cities, comparatively. This is closely related to the urban infrastructure and economic level, but has little to do with the pressure given by the natural environment. In the pressure-subsystem, the proportion of dependent population in Suqian is as high as 29.19%, and the other four cities are around 27% (only 18.71% in Nanjing). Due to the outflow of the independent population, urban construction and economic development are hindered. In the response-subsystem, the level of educational attainment equality in Suqian is only 0.21, and the other four cities are only around 0.3 (as high as 0.92 in Nanjing). The lack of education will not only weaken public awareness of flood emergency, but also affect the sustainable development of the city, leading to the loss of talents, which will fall into a vicious cycle. Consequently, these cities are vulnerable to floods.

Spatial distribution characteristics

ArcGIS was used to visualize the results in order to aid intuitive understanding of the level of urban flood resilience and the influence degree from the three dimensions, as well as the spatial distribution of flood resilience of 13 cities in Jiangsu Province. Z-scores were applied to sort level of urban flood resilience under five grades visualized as different colors. Cities with scores greater than 1.5 were classified as high level, between 0.5 and 1.5 as relatively high level, between −0.5 and 0.5 as moderate level, between −1.5 and −0.5 as relatively low level, and less than −1.5 as low level (Moghadas et al. 2019).

As shown in Figure 5, urban flood resilience and its three subsystems displayed distinct spatial agglomeration. Cities with moderate resilience are mainly near the center of Jiangsu. Besides Nanjing with high resilience, the resilience level in the southern areas of the Yangtze River (Southern Jiangsu) is relatively high, while that in the northern part (Northern Jiangsu) is relatively low, where polarization is obvious. The most likely explanation is that the Southern Jiangsu is a more developed region in Jiangsu Province, with relatively adequate supporting facilities and a better livable environment. This indirectly reveals that there is a serious unfairness in the distribution of urban infrastructure construction, education, medical care and other services in Jiangsu Province. It is necessary for these cities to obtain fair development and social well-being, only in this way can Jiangsu Province better improve the overall level of resilience. In the pressure-subsystem, the spatial distribution of pressure level is almost opposite to that of resilience level. No city in Jiangsu Province is under low pressure. Nanjing and Zhenjiang bear the highest pressure, which is due to concentrated heavy rainfall and dense hills resulting in large-scale runoff. As a result of less rainfall and small population density, the pressure levels in Northern Jiangsu are medium or low. In the state-subsystem, most cities are at a middle level. However, it is evident that the state of Southern Jiangsu is more resilient than that of Northern Jiangsu, among which Wuxi and Nantong are the best. This is because the state-subsystem mainly reflects the ability of urban infrastructure to withstand floods. The density of drainage pipelines in Wuxi and Nantong is the highest, which is 24.6% and 18.8% respectively. The per capita public green area and the coverage rate of green area developed of these cities are in the forefront of Jiangsu Province, which are important factors affecting the urban flood resilience. Hence, the results reveal the significance of strengthening physical infrastructure. The spatial distribution of response-subsystem is basically consistent with urban flood resilience. This may be related to the education, medical care and economic level of the city. The results show that the factors of response-subsystem have the greatest influence on urban flood resilience.
Figure 5

Urban flood resilience level for 13 prefecture-level cities along with its three dimensions: (a) urban flood resilience; (b) urban flood pressure; (c) urban flood state; and (d) urban flood response.

Figure 5

Urban flood resilience level for 13 prefecture-level cities along with its three dimensions: (a) urban flood resilience; (b) urban flood pressure; (c) urban flood state; and (d) urban flood response.

Close modal

Finally, the results illustrate a relatively moderate level of urban flood resilience for Jiangsu Province in terms of synthetical comparison. A total of five cities are at a low resilient level, and three cities are at a moderate resilient level, accounting for 61.54% of 13 prefecture-level cities. Nanjing is the only city with the best flood resilience and only four cities are clustered as relatively high flood resistance. The results obtained clearly summarize the level of urban flood resilience and emphasize which cities and factors are most in need of intervention.

Important factors

The grey relational analysis was adopted for the empirical results so as to obtain important factors, as shown in Figure 6. The indicators selected by the model have a large correlation value (above 0.55) with urban flood resilience. The results reveal that the selected indicators have a great influence on urban flood resilience.
Figure 6

Correlation degree.

Figure 6

Correlation degree.

Close modal

Furthermore, the more important factors of urban flood resilience include the building density, the population density, the educational attainment equality, the communication capacity and the per capita GDP, whose grey correlation degree is greater than 0.75, which are closely related to urban economic development and are in accordance with the evaluation results. Therefore, the government should not only improve these factors, but also balance the economic development among cities, so as to better build resilient cities and realize the sustainable development of the city.

Development of urban flood resilience

Urban flood resilience is the result of the comprehensive impact of pressure, state and response subsystems, and these factors are interrelated and interactive. Rainstorm pressure is the root of urban flood resilience. Environmental pressure and population pressure fundamentally influence the frequency and damage degree of floods, and also affect the post-disaster reconstruction and socio-economic recovery. However, the response measures are the most important factors affecting urban flood resilience, which should be paid more attention to.

In the evaluation and analysis, the flood resilience of Jiangsu Province presents two uncoordinated characteristics. On the one hand, the development of each subsystem in the city is uncoordinated. The levels of pressure, state and response are inconsistent. Taking Nanjing as an example, the pressure level is the lowest, and the state level ranks seventh, while the response level is the highest. Because of the largest weight of the response-subsystem, Nanjing has the best flood resilience. It is difficult to relieve the urban environmental and population pressure. Only by improving the state-subsystem and response-subsystem can the level of urban flood resilience be improved. The economic impacts and efforts are made to directly affect urban infrastructure. The educational attainment equality can affect public flood risk perceptions. The medical care affects urban disaster-relief abilities. They all indirectly affect the level of urban flood resilience. At present, Nanjing has made great efforts in the construction of economy, medical care and education. However, the investment in urban ecosystem and flood control works is relatively weak, which also makes its flood resilience uncoordinated. In view of the fact that the building density, the population density, the educational attainment equality, the communication capacity and the per capita GDP are important factors for improving urban flood resilience. In order to improve the ability to cope with flood disasters as a whole, the city can further strengthen the construction and improvement of urban ecosystem, flood control works and public flood-fighting awareness and capacity.

On the other hand, the development of flood resilience among cities is uncoordinated. The urban flood resilience in Jiangsu Province shows obvious polarization, that is, the resilience level of Southern Jiangsu is higher than that of Northern Jiangsu. Due to the characteristics of geographical environment and economic basis, there are differences in infrastructure investment, government management level and response ability among cities. Southern Jiangsu is an economically developed area, and its economy is in the forefront of the whole province and even China, which has laid a solid foundation for urban infrastructure construction and post-disaster recovery. However, the levels of economy, education and medical care in Northern Jiangsu is lower, and the construction standard of flood control works and drainage facilities is generally weak, which affects the process of post-disaster recovery. Consequently, the government must balance urban development.

Implication

Jiangsu Province, as an important part of the Yangtze River economic belt, is the focus of urban agglomeration construction. Facing severe flood disasters, Jiangsu Provenance should take necessary measures to balance the development of urban flood resilience. Thus, in view of the above analysis results, this study proposed some policy recommendations to improve the urban flood resilience from four aspects: ecological resilience, infrastructure resilience, organizational resilience and the fair distribution of resources.

First, the administrators should put the ecological protection, restoration and construction in a prominent position, and combine ecological infrastructure with gray infrastructure organically. Regional ecological networks should be built to enhance the ability of self-regulation and natural ecosystems to resist nature-induced disasters. Ecological space such as green space, rivers and lakes should be fully utilized to partition the disaster prevention zones for a systematic and complete disaster prevention spatial layout with timely response, so as to effectively prevent the spread of disasters. Second, urban infrastructure is as important as human bones and a key component in maintaining urban flood resilience. Therefore, it is necessary to incorporate urban infrastructure construction into the national strategy to get better urban planning and more financial expenditure. To promote the spatial optimization of infrastructure, the rationality of various infrastructure layout is examined from the perspective of integrity, systematisms and relevance. The sponge city concept is being developed to enhance the infrastructure resilience to flood hazards, including wetlands for rainwater storage, plant-covered rooftops and permeable pavements to store excess runoff water (Li et al. 2019). Third, through the establishment of a multi-dimensional collaborative governance system, a resilient urban governance system with multi-scale, multi-agent, full-space and whole-process defense mechanisms will be formed. Meanwhile, the community service management ability should be improved, more resources should be allocated to the grassroots, and comprehensive awareness of safety and resilience should be cultivated.

Last, justice, as an ethical assertion of governance, underpins urban resilience strategies, which is often formulated to deal with underlying inequalities (Douglass & Miller 2018). Therefore, whether it is the need of urban flood resilience construction or the moral proposition of fair governance, Jiangsu Province should focus on strengthening the urban construction in Northern Jiangsu, through increasing capital investment and talents introduction, so as to counteract the development gap between Southern and Northern Jiangsu. Northern Jiangsu now attaches importance to the improvement of medical care and education to ensure an appropriate balance between resource availability and population. The level of urban flood resilience in Southern Jiangsu as a whole is greater, but the government needs to pay more attention to the balanced development of three subsystems. In order to realize the common vision and mission, the government can reinforce public participation, take communities as experimental units and concentrate financial, material and human resources to build resilient communities by analyzing the problems existing in communities faced with flood disasters, so as to gain experience to better build resilient cities.

A comprehensive evaluation of the urban flood resilience is a broader extension of resilience research, which is of great significance to the sustainable development of cities. As a flood prone area, the study of urban flood resilience in Jiangsu Province is helpful to make appropriate strategies and policies for its resilient city construction. This study proposed an index system of urban flood resilience evaluation based on the PSR framework and a hybrid RAGA-PP-KL-TOPSIS approach. The assessment results show a relatively moderate urban flood resilience level in Jiangsu Province, as well as different and uncoordinated urban flood resilience levels among 13 cities. The construction density, the population density, the educational attainment equality, the communication capacity and the per capita GDP are important factors affecting urban flood resilience in Jiangsu Province, which is related to economic development. In order to coordinate the levels of flood resilience within and among cities, Jiangsu Provincial government should not only pay attention to the construction of state-subsystem and response-subsystem, but also consider the balanced development of cities.

This study proposed the index system of urban flood resilience based on PSR framework, which can be divided into pressure-subsystem, state-subsystem and response-subsystem, and had a clear description of each subsystem and the relationship between each system in detail. This expands the assessment framework of urban flood resilience to some extent. The hybrid RAGA-PP-KL-TOPSIS approach can provide more ideas for evaluation in other fields, such as sustainability development, resources, innovation capability. In addition, some indicators can be considered, such as housing age, sturdier housing type and worn-out urban textures. The authors are confident that these constraints will be effectively addressed in the future.

This study has been funded by the Postgraduate Research & Practice Innovation Program of Jiangsu Province (Grant No. KYCX21_0443), the Fundamental Research Funds for the Central Universities (Grant No. B210207036), National Natural Science Foundation of China (Grant No. 42071278), the National Key R&D Program of China (Grant No. 2019YFC0409000).

The authors declare there is no conflict.

All relevant data are included in the paper or its Supplementary Information.

Alexander
D. E.
2013
Resilience and disaster risk reduction: an etymological journey
.
Nat. Hazards Earth Syst. Sci.
13
(
11
),
2707
2716
.
Bertilsson
L.
,
Wiklund
K.
,
Tebaldi
I. M.
,
Rezende
O. M.
,
Veról
A. P.
&
Miguez
M. G.
2019
Urban flood resilience a multi-criteria index to integrate flood resilience into urban planning
.
J. Hydrol.
573
,
970
982
.
Cutter
S. L.
,
Barnes
L.
,
Berry
M.
,
Burton
C.
,
Evans
E.
,
Tate
E.
&
Webb
J.
2008
A place-based model for understanding community resilience to natural disasters
.
Glob. Environ. Chang.
18
,
598
606
.
Cutter
S. L.
,
Burton
C. G.
&
Emrich
C. T.
2010
Disaster resilience indicators for benchmarking baseline conditions
.
J. Homeland Secur Emerg. Manage.
7
(
1
),
1
24
.
Douglass
M.
&
Miller
M. A.
2018
Disaster justice in Asia's urbanizing Anthropocene
.
Environ. Plann. E: Nat. Space
1
(
3
),
271
287
.
Gallego-Lopez
C.
&
Essex
J.
2016
Designing for Infrastructure Resilience
.
UK Government, London, UK
.
Golz
S.
,
Schinke
R.
&
Naumann
T.
2015
Assessing the effects of flood resilience technologies on building scale
.
Urban Water J.
12
(
1
),
30
43
.
Holling
C. S.
1973
Resilience and stability of ecological systems
.
Annu. Rev. Ecol. Syst.
4
(
1
),
1
23
.
Hu
Y. H.
2002
The improved method for TOPSIS in comprehensive evaluation
.
Math Pract. Theory
32
(
4
),
572
575
.
Jia
H. F.
,
Wang
Z.
,
Zhen
X. Y.
,
Clar
M.
&
Yu
S. L.
2017
China's sponge city construction: a discussion on technical approaches
.
Front. Environ. Sci. Eng.
11
,
18
.
Li
Z. M.
,
Zhang
X. X.
,
Ma
Y. F.
,
Feng
C. Y.
&
Hajiyev
A.
2019
A multi-criteria decision making method for urban flood resilience evaluation with hybrid uncertainties
.
Int. J. Disaster Risk Reduct.
36
,
101140
.
Liu
S. F.
&
Li
Y.
2006
Grey Information: Theory and Practical Applications
.
Spring Verlag
,
London, UK
.
Liu, D., Wang, C. Q., Yi, J., Fu, Q., Li, M., Ali, S., Li, T. X. & Cui, S. 2021 Measurement and analysis of regional flood disaster resilience based on a support vector regression model refined by the selfish herd optimizer with elite opposition-based learning. J. Environ. Manag. 300, 113764. https://doi.org/10.1016/j.jenvman.2021.113764.
Marc
W.
2013
Resilience and responsibility: governing uncertainty in a complex world
.
Geogr J
180
(
1
),
15
26
.
Ministry of Water Resources of the People's Republic of China
2006–2020
China Flood and Drought Disaster Bulletin
.
China Statistics
,
Beijing, China
.
Moghadas
M.
,
Asadzadeh
A.
,
Vafeidis
A.
,
Fekete
A.
&
Ktter
T.
2019
A multi-criteria approach for assessing urban flood resilience in tehran, iran
.
Int. J. Disaster Risk Reduct.
35
,
10196
.
OECD
1993
Environmental Indicators: Basic Concepts and Terminology. Background Paper No. 1. Group on the State of the Environment, Environment Directorate
.
Organization for Economic Co-operation and Development
,
Paris, France
.
Qasim
S.
,
Qasim
M.
,
Shrestha
R. P.
,
Khan
A. N.
&
Tun
K.
2016
Community resilience to flood hazards in Khyber Pukhthunkhwa province of Pakistan
.
Int. J. Disaster Risk Reduct.
18
,
100
106
.
Rockefeller, ARUP
2015
City Resilience Index. Accessed April 11, 2021
.
Sefton
C.
,
Sharp
L.
,
Quinn
R.
,
Stovin
V.
&
Pitcher
L.
2022
The feasibility of domestic raintanks contributing to community-oriented urban flood resilience
.
Clim. Risk Manag.
35
,
100390
.
Vamvakeridou-Lyroudia
L. S.
,
Chen
A. S.
,
Khoury
M.
,
Gibson
M. J.
,
Kostaridis
A.
,
Stewart
D.
,
Wood
M.
,
Djordjevic
S.
&
Savic
D. A.
2020
Assessing and visualising hazard impacts to enhance the resilience of critical infrastructures to urban flooding
.
Sci. Total Environ.
707
,
136078
.
Verrucci
E.
,
Rossetto
T.
,
Twigg
J.
&
Adam
B.
2012
Multi-disciplinary indicators for evaluating the seismic resilience of urban areas. Iitk.ac.in.
Wagner
I.
&
Breil
P.
2013
The role of ecohydrology in creating more resilient cities
.
Ecohydrol. Hydrobiol.
13
(
2
),
113
134
.
Wan
B.
&
Zhai
G.
2017
Research on the post-disaster regional economic growth model based on resilience: taking the Wenchuan earthquake as an example
.
Mod. City Res.
9
,
41
47
.
Wang
Q.
&
Zhan
L.
2019
Assessing the sustainability of the shale gas industry by combining dpsirm model and raga-pp techniques: an empirical analysis of Sichuan and Chongqing, China
.
Energy
176
,
353
364
.
https://doi.org/10.1016/j.energy.2019.03.158
.
Wang
W. C.
,
Chau
K. W.
,
Xu
D. M.
,
Qiu
L.
&
Liu
C. C.
2017
The annual maximum flood peak discharge forecasting using hermite projection pursuit regression with SSO and LS method
.
Water Resour. Manage.
31
(
1
),
1
17
.
WMO
2020
Global Climate in 2015–2019
.
World Meteorological Organization
,
Geneva, Switzerland
.
Zhang
X.
,
Song
J.
,
Peng
J.
&
Wu
J.
2019
Landslides-oriented urban disaster resilience assessment—a case study in Shenzhen, China
.
Science of The Total Environment
661
,
95
106
.
https://doi.org/10.1016/j.scitotenv.2018.12.074
.
Zhao
F. K.
,
Yang
L.
,
Chen
L. D.
,
Li
G.
,
Sun
L.
&
Li
S. J.
2018
Soil security in peri-urban ecosystems: problems and challenges
.
Acta Ecol. Sin.
38
(
12
),
4109
4120
.
Zhu
S.
,
Li
D.
,
Huang
G.
,
Chhipi-Shrestha
G. K.
&
Sadiq
R.
2021
Enhancing urban flood resilience: a holistic framework incorporating historic worst flood to Yangtze River Delta, China
.
Int. J. Disaster Risk Reduct.
61
,
102355
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/).