Conceptualizing and Measuring Megacity Resilience with an Integrated Approach: The Case of China

Abstract

Megacities play an essential role in social interaction and relationship formation. There is a need for megacity resilience to achieve both safety and sustainability. This paper set out to develop a contextualized conceptual framework and an applied megacity resilience comprehensive index (MRCI). The study provides a multi-model named the technique for order preference by similarity to ideal solution (TOPSIS), extending the criteria importance through inter-criteria correlation and entropy (CRITIC-Entropy) weight and introducing the time-ordered weighted averaging (TOWA) to a dynamic situation. The results show that, while the performance of resilience in Nanjing was the highest, the growth ratio of resilience in Zhengzhou was the fastest. In addition, a coupling relationship of pressure, state, and response resilience was verified, and response resilience was more correlated and showed similar trends with the MRCI. The findings indicate that response resilience is still an obstacle factor in the criterion layer in Dalian. Moreover, identified key obstacle factors in the index layer may differ by district or functional zones and need to improve unified and point to area operation. Issues around resilient culture and citizenship were found to be common. Improving public service in Zhengzhou, enhancing support for applied research in Nanjing, and optimizing the ecological industry layout in Dalian were identified as key focuses. This study should be of value for similar megacities in developed or developing countries to improve their resilience.

1. Introduction

As physical spaces, economic carriers, and social arenas, cities have attracted considerable interest. The pace of urbanization continues to accelerate worldwide, making the contemporary era the city’s century [1]. Due to the flows of information, capital, and people reshaping urban space, a global transaction network centered on metropolitan regions is emerging [2]. To some extent, a megacity serves as the crystallization and engine of urban evolution. It has been observed that megacities interconnect regions due to the sociological and spatial contexts and have domino effects [3]. Megacities are more likely to experience certain phenomena, such as: concerning overburdened urbanization, edification near industrial zones being at risk, natural sites like as coastal areas, and governance-related social and economic disparities [4]. Efforts to address these disturbances and to adapt to sustainability have become urgent and ever-pressing.

Megacities in developing countries are increasingly seen as global network actors, especially in China. In the Global Cities Index (2021), 31 Chinese cities were on the list. Currently, China has an urbanization rate of 64.72%. Megacities are under immense pressure due to more concentrated urban activities [5]. According to the Human Cost of Disasters, an Overview of the Last 20 Years (2000–2019), the highest number of disaster events, 577, was reported from large and heterogeneous landmasses. Resilience is emerging as a strategy that, when applied, will help support actions geared toward counteracting these issues [6]. The 14th Five-Year Plan (2021–2025) attaches a high priority to the idea of resilience and safety and requires strengthening the risk governance of megacities. Moreover, resilient megacities have been placed high on the construction agenda aiming at “High-Quality Development”. However, each city has different processes that may generate content found to be contextually pertinent and available [7]. Developing methodologically sound procedures for tailoring measurement frameworks is a critical research need.

A conceptual approach is often used in studies that propose urban resilience frameworks or map tools. Diverse bodies, including supranational institutions, have endorsed this approach, such as the Disaster Risk Reduction Initiatives set by UNISDR [8], the 100 Resilient Cities program launched by the Rockefeller Foundation [9], and the New Urban Agenda (2030 SDGs) issued by the United Nations [10]. A transformation paradigm of “engineering resilience—ecological resilience—evolutionary resilience” has emerged. With this conceptual change, the emphasis of solution framing is no longer on simply responding to disasters, but on anticipating and planning for future changes [11,12]. Rational urban development acknowledges environmental uncertainty and limited urban capacity while also adhering to the law of social–ecological systems in a way that provides resilience alongside sustainability [13]. Policy goals and analytical frameworks involve clear normativeness and descriptions of sustainability, resilience, and transformation in various contexts, taking into account scale, stressors, worldviews, and actors [14]. Sometimes, the conceptual framework is too abstract and far from decision-makers’ daily activity, which hampers proper operationalization [15]. Thus, it requires us to transcend beyond theoretical to empirical research. Several attempts have been made to explore infrastructure resilience or other component resilience functional modules and planning tradeoffs in specific disturbances [16,17]. What is known about comprehensive resilience performance may not be directly created. It is primarily fostered through indicators by the system or capability to evaluate if cities are progressing desirably [18,19]. Divided by geographies and climates, city cases are analyzed in relation to resilience in different contexts [20,21]. Methods for assessing resilience show diversification and include hybrid factor analysis [22], the analytic network process model [23], the GA-BP neural network [24], maps and geographic information systems [25], etc. Previous studies generally ignored urban resilience as a dynamic process attribute for integrated scenarios. Moreover, they lacked well-grounded theoretical consideration, which helps to support a resilience process conceptual framework and explores its internal dynamics. In addition, much of previous research was restricted to complex, rigid indicators and did not engage local actors to develop local storylines within a global framework. Although a megacity is an excellent laboratory for resilience, cases are fragmented across individual study sites, and few studies have been limited to considering comprehensive resilience on this scale.

This paper offers several innovations. First, it makes a conceptual link between evolutionary resilience and the pressure–state–response (PSR) process approach. This work aims to develop circular planning toward security and development instead of relying on the static urban system or resilience characteristic capabilities. Second, incorporating a conceptual framework and an applied index system, a dynamic set is generated that may characterize critical issues and make sense of data, weighing local conditions versus global resilience. It greatly improves the creation of flexible resilience indicators for context.

The structure of this article is as follows. The PSR theory is presented, which includes the process attribute and gives a new conceptual framework for megacity resilience. Next, the applied quantitative and qualitative megacity resilience index system was developed through expert knowledge and sample data. Moreover, we adopted a multi-model integration method to evaluate the performance of the megacity resilience comprehensive index (MRCI). In particular, improving the criteria importance through inter-criteria correlation and entropy (CRITIC-Entropy) strategies compensated for the deviation of the technique for order preference by similarity to ideal solution (TOPSIS), and introducing the time-ordered weighted averaging (TOWA) created a dynamic contrast. In addition, the obstacle diagnosis and cause analyses of typical cases were carried out to promote resilience.

2. Literature Review

2.1. Definition of Resilience and Urban Resilience

Resilience definitions related to different fields are constantly being broadened and refined. Single equilibrium resilience has a long history in engineering and disaster literature. Meanwhile, multiple equilibria, called ecological resilience, are often cited as the origin of modern resilience theory [26]. Effectively, Holling used ecological resilience rather than engineering resilience to describe the ability of an ecological system to persist when changed, but not necessarily to remain the same [27]. It focuses on “returning to normal” and “resistance”. However, strictly adhering to a state of stability and returning to equilibrium sometimes means returning to poor development and vulnerability and, potentially, catastrophe. Such a goal may lead to maladaptation and not be sustainable [28].

It should be noted that resilience researchers do not focus solely on disaster risk management or climate change issues. The social sciences, politics, and behavioral sciences also study resilience by enhancing resources [29], capital [30]. and so on. Recently, a new social–ecological perspective has pointed out in relation to evolutionary resilience that complex systems in constant change go beyond equilibrium and highlights their ability to undergo “renewal”, “reorganization”, and “development” [31,32,33]. Moreover, adaptation and transformation approaches allow for the possibility of system changes and gradually adjust the system to post-event situations, accidents, or more fundamental shifts over the long run [34].

As a descriptive phrase, resilience refers to a system’s dynamic qualities, including vulnerability drivers, regulating variables, tipping points, and transformational capacity [35]. It is a process that encourages diversity, reflexivity, and invention while also embracing uncertainty and interdependencies [36]. Cities are conceptualized as complex and adaptive, comprising diverse socio-ecological and socio-technical networks that extend across multiple spatial scales [37]. Resilience may improve from resistance, adaptation, and change in social and technical systems and their interaction domains. This process might be embedded in a standard higher-scale system that builds urban resilience over time along a specific development path [38]. In light of these issues, we define urban resilience as how entities and agents build endowment capabilities and utilize them to influence their environment. In a way, it positively identifies disturbances, maintains functioning, and adjusts structure priority, before, during, and following adversity.

2.2. Content Relating to Urban Resilience, Safety, and Sustainability

Resilience has become a new paradigm for city safety in today’s risky society [39]. Drawing on the experience of resilience engineering and disaster response, the six functions of the systemic model were outlined, and some opposing characteristics were maintained without contradiction [40]. This model tested various resilience characteristics using five disaster scenarios. The external disaster relief capability coefficient and the critical parameters were also analyzed [41]. The model aims to understand the impact of perturbation relative to its scale and add a new stress dimension to extend the existing time-dependent resilience function [42]. It connects and contrasts nine system performance concepts that characterize systems withstanding threats. Moreover, a conceptual framing has been proposed that synthesizes the differences and relationships [43]. These criteria involved: community vulnerability, built environments and their infrastructure, natural systems, and risk reduction and planning. The first three criteria reflect prior research, and the fourth and final criterion considers policy and procedural factors [44]. The Resilient Cities Planning Framework consists of four interlinked concepts, the vulnerability analysis matrix, prevention, urban governance, and uncertainty-oriented planning, resulting in guidance on how to integrate multidisciplinary insights to achieve more resilient cities [45].

For empirical and theory-driven urban management studies, urban sustainability and resilience are emerging topics and interconnected strategic orientations [46]. In a previous study, a sustainable and resilient framework was developed through co-design, co-production, and dissemination of a long-term ecological network of U.S. cities [47]. In another study, nature-based solutions entailed a seven-stage assessment associated with climate resilience and health and well-being in urban areas [48]. Further, Agenda 2030 issued aims, frameworks, and standards for sustainability. In particular, the 75th paragraph specifies the requirement for creating indicators. The defining characteristics of diversity, modularity, tight feedback, social cohesion, and innovation have an impact on urban resilience. Moreover, the following indicators match projects’ urban traits [19]. For transferring project results of urban resilience, it discussed possible standardization activities and stakeholder engagement [49]. In three case studies, citizens’ actions, multi-institutional opportunism, and big data scenario modeling were found to be enablers of resilience to confront climate change, simultaneously improving livability and sustainability [50]. Identifying the contrasts and synergies between academic and practitioner perspectives lays the groundwork for better resilience research [51].

Urban resilience content is broadening, ranging from natural, physical, and economic fields to social and institutional fields. Linking short-term emergency objectives with long-term sustainability is a growing need. The former is based on theoretical reasoning, whereas the latter involves modeling attempts or empirical validation. Regarding their conceptual frameworks, they both fail to consider urban resilience as an ongoing procedural phenomenon and its antecedents and consequences. As for empirical research, assessment bridges urban resilience theory and practice. It has been proven to be a valuable tool to quantify goals, measure progress, and benchmark performance for resilience. In practice, building resilient cities tends to involve the same benchmarks for the equal and uniform evaluation of different objects. Sometimes, portability might suffer from inappropriate developmental backgrounds or changeable scenarios. Additionally, excessive numbers and rigid indicators are an issue for implementation.

2.3. PSR Framework Application and Its Relationship to Urban Resilience

PSR relies on the causality concept: human activities and climate change place pressure on the environment, affecting the quality and quantity of resources and altering the functions of ecosystems, which in turn leads to human efforts to implement measures. PSR involves three steps. “Pressure” is the trigger for problems and is also included in the negative phase. “State” refers to the system condition and reflects the support potential. “Response” is the term used to describe how a system reacts to pressure and involves self-organized activity and human intervention [52,53]. It reveals the circular process among drivers, systems, and human activities. The PSR model is a mature theory used widely in ecological security and environmental performance [54,55]. In conjunction with geographic characteristics and socio-ecological contexts, the PSR model offers a solid analytical foundation for system and human factor challenges. Cities are exemplars of complex systems, and evolutionary resilience energizes the study of cities. In addition, the PSR framework originated in ecology and constantly collides with city boundaries due to the real needs of construction and governance. Several applications of the PSR framework have been extended, including the analysis of urban vulnerability and urban functional optimization [56,57]. However, research on urban resilience has been restricted to specific hazard scenarios or system compositions [58,59], such as flood resilience under heavy rain or infrastructure resilience. Neither comprehensive scenarios nor circular thought have been taken into account.

The pressure–state–response framework is utilized to dismantle system behavior, and its logic fits the resilience process attributes. In addition, applying the PSR theory to comprehensive resilience satisfies the expandable boundary principle.

Sources of disturbance are uniformly referred to as “pressure resilience”. Its dynamics stem from both external and endogenous disturbances. External disturbances are typically caused by the physical characteristics of natural hazards, such as their frequency, intensity, and duration. At the same time, self-generated issues in the urban system are generally the result of human activities and production demands. Pressure resilience can serve as a rough guide to prepare a knowledge base for building comprehensive scenarios. A system is under strain until a new response cycle takes effect. A passive bearing phase occurs, requiring the system to withstand energy shock. Here, the robustness and redundancy of the city’s intrinsic properties come into play. This is mainly reflected in the strength of the support provided by entities and actors. Thus, we call this part “state resilience”. Cities with higher resilience experience disorder slower, allowing them more time to cope with certain pressures. Critical services and functions are also preserved more frequently. Using theory and prior experience, agents collaborate to reverse negative situations and pursue development amid a crisis. This is referred to as “response resilience”. As flexible variables in the process, governments, corporations, and individuals have tremendous potential to interact and bring about a higher degree of dynamic equilibrium. They analyze the pressure incident and apply that learning towards the actual construction, optimizing the system structure for long-term sustainability.

The PSR conceptual framework provides a holistic view and identifies the parameters driving the feedback processes. Megacity resilience can be evaluated according to the disturbance inputs, changes in states, and feedback to complement the research subject in terms of its process features. Prediction–adaptation–resilience (PAR) projects for future urban growth and vulnerability adopt demands and supply chains and connect SDGs [60]. This approach has similarities with PSR and may help us in the future to strengthen resilience.

3. Materials and Methods

3.1. Conceptual Framework

We identified megacity traits that may make them special and even provoke managing them differently than other cities. Although these manifest themselves variously in different places, several main concerns appear typical to megacities and arise often. Firstly, according to the World Cities Risk Index 2015–2025 from the Cambridge Risk Research Centre, eight of the top ten most-at-risk cities are megacities. Regarding their geographical position, megacities, such as those located in coastal areas, are sensitive to historical climate change incidents. Secondly, large populations may put a strain on per capita resources in megacities. Due to their density, the unequal distribution of geographical infrastructure areas, and increased needs for quality and quantity in public services, some functions of megacities are weaker. Thirdly, city renewal and population movement cause divisions between strong and vulnerable people, with spatial juxtapositions often being pushed to the limit. Social space inequality is intertwined and generates security hazards in megacities. Fourthly, even though services in Chinese megacities comprise more than 50% of China’s GDP, industry remains a pillar that generally adds more than 20% to GDP. Megacities can become fragile ecological environments if they encounter severe pollution threats from industrialization. Lastly, megacities’ indistinct physical limits complicate defining regulatory boundaries, rendering planning and regulating measures exceedingly tricky. A flat management approach is inappropriate for megacities of dense processes and actors. To cope with these issues, we considered integrating entities and agents with ongoing processes to identify impact pressures and create adequate responses.

In this study, a process resilience structure, based on evolutionary resilience and PSR theory, was put in place. This derived framework, shown in Figure 1, involved a holistic approach. Presented in

Table 1. Key terminology in the framework.

Key Terminology	Justification	References
Natural Hazards	Consider the frequency and magnitude of hazards.	[61,62]
System Service	Require a system to meet shared needs, such as availability of key services and societies’ well-being and security.	[16,63]
Entities	Involve infrastructure and more material endowment.	[64]
Agents	Consider at-risk people as capable agents rather than passive victims. Individuals, organizations, and governments work through their characteristics and preparedness from the micro to the macro level.	[65,66]
Recovery	Focus on agents: self-determination and tradeoffs in collaborating to achieve rapid rescues.	[67]
Adaptability	Take the time to reflect and learn about approaches, assets, and strategies for strengthening long-term rules.	[68]
Scenario Identification	Clarify whether megacity scenarios show similarities by comparing past conditions and exploring uncertainty.	[69]
Specific Components	Some components are sensitive to specific pressures, such as drainage systems under frequent rainfall.	[70,71]
Generic Structures	An overall configuration that does not significantly change its service level.	[70,71]
Cognitive	Both constructive conceptual orientation and profound knowledge and expertise help megacities cope with pressures.	[72]
Behavioral	Combining practical habits with behavioral preparedness generates options rather than constraints.	[73]
Prior Knowledge Base	Design an appropriate structure for valid case contents, and retain new solutions in the case base for solving similar future problems.	[74]
Contextual Reinforcement	When a threshold of regime shift occurs or knowledge is obtained from new practices, updating is undertaken by revising indexes to adapt to scenario trends.	[75]
Expressive Analysis	Explore the relationship between occurring pressure events and changes in the resilience of urban processes.	[76]
Envisioning Analysis	A consistent qualitative approach describes how the future may unfold, bridging the gap between locally developed scenarios and higher-level scales.	[69]

are the key terms.

Pressure gives rise to scenarios that unfold across entities and agents. Disruptions to specific components of the system and adverse impacts on provided services are two examples of these. These refer to the system’s specific ability to cope with disturbances in its components and self-organization. Related to the generic structure, influential and diverse stakeholder participation is integral to designing a polycentric and reflexive system. Stakeholders collaborate equitably, matching initiatives taken in various sectors. The cognitive and behavior agents interacted with during disturbances shape organizing for subsequent disturbances. Their improved governance capabilities can foster responses to drift and aid in anticipated actions. Ongoing adaptation occurs as disturbances are processed and addressed and linked to the prior knowledge base, with contextual reinforcement taking place. This is where core aspects of process attributes are naturally nested. Moreover, the feedback loop of expressive and envisioning analyses meets the needs of both short-term emergencies and long-term development. Thus, necessary information is gathered to plan for future scenarios, and generic objectives and relative parameters are generated for mapping counterpart actions on structure and composition.

Each process resilience connotation is described in detail, and the groundwork for locating operable indicators is laid in this way.

Pressure resilience relates to an interaction between a hazard source and its social context [77]. Typically, pressure is external or endogenous to the system. Among the external shocks, there are mainly natural hazards that impact cities temporarily; internal shocks are mainly pressures on urban services that affect cities on a long-term basis. Natural hazards are tied to the regional geographic characteristics of cities, from historical incidents to new threats posed by climate change. Natural hazards, such as floods and earthquakes, can occur over large areas randomly. Climate change triggers frequent extreme weather events, which lead to severe secondary risks [78,79]. It is imperative to analyze such events’ scope and anticipated impacts, using them as external archetypes of disturbance objects. Long-term goals of system services, such as limited urban carrying capacity, increased dangers to social stability, and demand issues for green urban growth, are putting pressure on cities [80,81,82]. Megacity resilience explores what value they provide and for whom. Focusing on the corresponding system, service delivery makes pressure apparent since it can be measured directly or by proxy. A megacity is one in which resources, markets, information, and other elements are densely concentrated and frequently overburdened. Lifestyles in these cities are fraught with latent social risk processes that conflict with citizens’ rights, breeding artificial instability [83]. Measures of improving citizens’ essential health and related well-being, which are advocated by livable cities, are limited. In megacities, the pressure experienced by system services reflects several characteristics and long-term development requirements.

State resilience allows for encompassing agents and entities while examining their degree of structure and identity. Scholars consider entities as resource endowments of interconnected infrastructure and economic conditions [37,84]. The economy as a particular entity lays a solid foundation for the other entities. The provision of infrastructure, including buildings and communication, can provide a natural barrier to eliminating impacts. Key components are spatially distributed and functionally linked, but can be restructured [7]. Redundancy and modularity that extra capacities do not use daily can accommodate natural disasters and unexpected services. Specific components and multiple pathways provide alternative options for delivery, with the availability of critical functions becoming maximized under pressure. Regarding agents, resilience depends on different groups and their socially differentiated capacities. Age, education level, and occupation have been documented as contributing to undertaking pressure through features such as access to services and social networks. Individually, resilience varies across demographic factors, indicating differences in physical and psychological disaster tolerance [85]. As a spontaneous social force, potential members of institutions represent the awareness and responsibility of the general population [86]. Local action is enabled and shaped by larger governance structures [87]. Government foresight is an essential condition by which an emergency action guide can be achieved.

Response resilience involves carrying out emergency reactions, learning and summarizing, and obtaining updated rule configurations from experience [88]. Collaboration emergencies are used here to examine recovery. The availability of resources is positively related to educational attainment. Individuals with a high education level are more likely to respond correctly and have access to livelihood security [89]. Social ties often act as first responders under pressure before professional and formal rescue efforts can begin. As a result of informal and voluntary networks, stakeholders can engage in political processes outside the confines of restraining procedures [90]. Decision-making capacity on the part of governments interacts with the shared nature of information communication to facilitate cross-sectoral linkages and to speed up the urban recovery process [91]. In many instances, actions taken by megacities have ripple effects on neighboring municipalities, requiring their cooperation [92]. Within the region, megacities can pool resources to conduct pressure assessments and share practice innovations. Through internal processes and with external drivers, the term “adaptability” refers to the ability to make adjustments. Livelihood diversification is widely utilized to deal with disturbances [93]. An array of activities and social support capabilities, such as more effortless occupational shifts, are developed for survival. Education and culture contribute to the establishment of common urban ideals [94]. In doing so, residents are encouraged to develop a resilient mind, visualize a plan of action, and actively engage in public affairs. The vertical and horizontal management abilities of the government inspire, facilitate, and shape local adaptations for overcoming common barriers [95]. An organized and effective policy system can clarify the roles and responsibilities of agents, thus serving as a safeguard [96]. A thorough review of emergency response shows that the keywords “optimization” and “demand” are emerging topics [97]. Gaining knowledge from prior catastrophes and rebuilding and expanding infrastructure to optimize megacities are also major needs. Additionally, technological innovation is intensifying. An integrated platform is on the agenda for early warning, monitoring, and management systems. There is evidence that green spaces provide health benefits throughout the human lifespan [98,99]. Such routes for achieving human health and well-being are in line with sustainability.

3.2. Index System Construction

A megacity refers to a city that has a vast population, advanced urbanization, a complicated social organization, and a vulnerable ecological environment. The China Bureau of Statistics defines megacities as cities with permanent resident populations of between 5 million and 10 million. Megacities generally occupy a prominent position in regional politics and economies. As cases, we selected Zhengzhou, Nanjing, and Dalian, which respectively head the Zhongyuan, the Yangtze River Delta, and the mid-southern Liaoning urban agglomerations. Here is a map showing the research’s context; see Figure 2.

Considering how global and national drivers interact with local conditions, we sought a process of knowledge construction encompassing different technical, political, and stakeholder perspectives. We, therefore, invited 15 experienced experts to work together on a comprehensive evaluation system of megacity resilience. They were chosen based on age, experience, and influence diversity in their organizations. They were from diverse fields, such as meteorology, sociology, water conservancy, urban planning, and disaster management. A profile of the experts is shown in

Table 2. Profile of the experts.

ID.	Role	Department	Background	Experience
E1	Planning manager	Emergency Management Agency	Civil engineering	25
E2	Consultant	Emergency Management Agency	Meteorological engineering	18
E3	Coordinator	Emergency Management Agency	Information management	10
E4	General affairs manager	Housing and Urban-Rural Development Bureau	Urban and rural planning	28
E5	Technical office manager	Housing and Urban-Rural Development Bureau	Civil engineering	23
E6	Regulation office manager	Housing and Urban-Rural Development Bureau	Policy studies	20
E7	Section chief	Human Resources and Social Security	Public administration	16
E8	Deputy section chief	Human Resources and Social Security	Finance	12
E9	Technical office manager	Industry and Information Technology	Computer science	20
E10	Engineer	Water and Sewerage Administration	Environmental engineering	15
E11	Engineer	Transportation Bureau	Civil engineering	14
E12	Professor	University	Urban and rural planning	27
E13	Professor	University	Civil engineering	20
E14	Associate professor	University	Public administration	15
E15	Associate professor	University	Meteorological engineering	12

. Firstly, we relied on and combined the research findings of previous scholars. The Rockefeller Foundation City Resilience Index and Guide for Safety Resilient City Evaluation (GB/T40947-2021) served as a mature reference. Secondly, some resilience-related project reports from Chinese cities were undertaken. By means of scientific, comprehensive, and flexible principles, the specific connotations were enhanced and matched indicators were completed. Finally, the revision had to take into account actual needs and data availability. Complex scientific data were converted into qualitative forms of informative local issues and perceptions [100]. Faced with disturbance, the Chinese government can quickly concentrate its efforts on major issues. Due to the peculiarities of the Chinese system, its decision-making skill was considered. In addition, some objective indicators that cannot be measured directly due to resource constraints were identified and then replaced with approximate indicators. Because of the scattered distribution of per capita shelter areas, we substituted the labor productivity of the construction industry to show its potential capacity for emergency protection. This study constructed a megacity resilience comprehensive index system. It includes three primary indicators (pressure resilience, state resilience, and response resilience), as well as 28 secondary indicators. Moreover, 22 qualitative and 6 quantitative data can serve as input for mixed approaches. The specific indicators and justifications are shown in

Table 3. The megacity resilience comprehensive index system. Note: “*”represents no unit; “+”means positive indicator; “−“means negative indicator.

Dimension	Measurement Items	Justifications	Unit Index	Attribute
Pressure Resilience	Number of warning messages issued	The probability magnitude of comprehensive natural disasters	Freq.	−
	Direct economic losses from natural disasters	The impact degree of comprehensive natural disasters	100 million CNY	−
	Population density	System service availability, whether they are within 500 m of basic public services	persons/km2	−
	Public safety rankings	Safeguarded services for citizen security issues involve social security, traffic accidents, and so on	*	−
	Ratio of industrial solid wastes comprehensively unutilized	Related to human health and well-being	%	−
State Resilience	Civil building quality	Structural and shelter functions and seismic protection strength	*	+
	Labor productivity of construction industry	Construction industry emergency security capabilities, such as temporary resettlement	CNY/person	+
	Density of road network	Transportation support capacity	km/km2	+
	Total annual volume of passengers transported by buses and trolley buses	Traffic operation capability	10,000 person-times	+
	Density of sewers	Drainage support capacity	km/km2	+
	Ratio of wastewater centralized treated by sewage work	Wastewater treatment capacity	%	+
	Number of mobile telephones per 100 households	The degree of communication coverage	pcs/100 households	+
	Beds of health care institutions per 1000 persons	Medical resources guarantee	pcs/1000 persons	+
	Proportion of tertiary sector of GDP	Comprehensive economic strength	%	+
	Local general public budget expenditure	Financial support	CNY 100 million	+
	Aged 0–14, aged 65, and over-proportion	Physical and psychological vulnerability of these groups should be considered	%	−
	Proportion of staff in public management and social security organizations	Back-up forces that can be deployed in an emergency	%	+
	Preparation of pre-disaster planning and organization of emergency drills	Whether the government is forward-looking and prepared in terms of emergency procedures	*	+
Response Resilience	Number of students in regular HEIs per 10,000 persons	Educational attainment positively correlates with access to resources (self-rescue)	person	+
	Density of social organizations	Social network relationship strength and coverage (volunteer rescue)	pcs/10,000 persons	+
	Government decision-making capability	Disaster situation analysis and appropriate actions (government rescue)	*	+
	Regional collaboration capability	Rapid and effective access to manpower and resources from neighboring cities	*	+
	Insurance penetration	Livelihood diversity protection	%	+
	Collections of public libraries per 100 persons	Urban culture construction and cultivation of citizens’ disaster prevention cognition	copies/100 persons	+
	Investment in municipal public infrastructure construction	Investment in infrastructure upgrades, such as renovations and expansions	CNY 10,000	+
	Integrated government capacity	The government’s institutional advantages should be utilized for vertical and horizontal collaborative governance	*	+
	R&D internal outlay	The ability to monitor and detect threats is dependent on technological innovation	CNY 10,000	+
	Green covered area as %	People-oriented and sustainable development pursuits	%	+

.

The data from 2010 to 2019 were used as the research basepoint. Some of these recommended indicators came directly from the China statistical yearbook, China city statistical yearbook, and China urban construction statistical yearbook. Reports on the work of local government or designed composite indicators from the City Blue Book (e.g., government decision-making capabilities, public safety rankings) were adopted in the later research. The individual missing data were imputed by the average values or the growth rate interpolation method. As for other indicators, experts scored them by analyzing the relevant data of norms, policies, and news reports over the years, using a Likert scale.

3.3. Methodology

3.3.1. The Improved CRITIC-Entropy Weight and TOPSIS

The scientificity of the resilience evaluation largely depends on the rationality of the weight of each index. For a better indicator weight, we combined multiple approaches. According to CRITIC, indexes are judged objectively by evaluating their degree of comparative strength and internal conflict [101]. Scholarship has shown that the standard deviation is dimensional, and the correlation coefficient makes no difference between positive and negative signs. We used the absolute value of the correlation coefficient to improve the model [102]. Despite this, it cannot measure the degree of dispersion among the indicators. A better result is achieved when the entropy weight rule is added to the model [103]. It was assumed that there are $m$ evaluation objects and $n$ item evaluation indexes; $i=1,\dots ,m$; $j=1,\dots ,n$. Specifically, the operations of standardized treatment are:

Standardized measures were applied to eliminate dimensional differences between indexes. $x_{ij}^{\prime }$ is the processed data.

$$x_{ij}^{\prime }=\{\begin{array}{l}\frac{x_{ij}-\min x_j}{\max x_j-\min x_j}, \mathrm{positive} \mathrm{indicator} \hfill \\ \frac{\max x_j-x_{ij}}{\max x_j-\min x_j}, \mathrm{negative} \mathrm{indicator}\hfill \end{array}$$

(1)

where positive indicator means the larger the indicator, the better the evaluation result is and max ${\mathrm{x}}_j$ is the maximum values of the $j$ index, and vice versa.

Using CRITIC to calculate the weight of the j index:

Both fluctuation and conflict were taken into account. $c_j$ represents the information content contained in the $j$ index.

$$c_j=\frac{{\sigma }_j}{{\overline{x}}_j}{\displaystyle \sum _{i=1}^m\left(1-\left|r_{ij}\right|\right)}$$

(2)

where ${\sigma }_j$ and ${\overline{x}}_j$ are the standard deviation and average of the $i$ index and $r_{ij}$ is the correlation coefficient between the $i$ and $j$ index.

The normalized weight for the $j$th feature ($w_1$) was determined.

$$w_1=\frac{c_j}{{\displaystyle \sum _{j=1}^n{c}_j}}$$

(3)

Using entropy to calculate the weight of the $j$ index:

$e_j$ means information entropy of the $j$th index.

$$e_j=-\ln {(m)}^{-1}{\displaystyle \sum _{i=1}^m{p}_{ij}}\ln p_{ij}$$

(4)

where $p_{ij}=\frac{x_{ij}^{\prime }}{{\displaystyle \sum _{i=1}^m{x}_{ij}^{\prime }}}$ is the proportion of the $j$ index occurring in the $i$th object.

The entropy weight ($w_2$) of the $j$th index can be obtained.

$$w_2=\frac{1-e_j}{{\displaystyle \sum _{j=1}^{n}1}-e_j}$$

(5)

where $1-e_j$ is the information utility value of item $j$.

Calculate the portfolio weights:

$$w_j=\beta w_1+(1-\beta )w_2$$

(6)

TOPSIS has been effectively utilized in various application areas and industrial sectors, but it needs to emphasize multidisciplinary and social decision issues [104]. The subject of megacity resilience fits perfectly with it. TOPSIS is a technique that detects the distance between multiple objectives and positive or negative ideal solutions and then ranks the superiority of these solutions based on the proximity of these distances [105]. The improved CRITIC-Entropy weight method and the TOPSIS method can overcome this technique’s traditional flaw of failing to record the correlation and importance of variables. The resulting inverse-order problem is therefore effectively solved [106].

Calculate the weighted matrix:

$$Z=X_{ij}\times W_{\mathrm{j}}=\left[\begin{array}{cccc}{z}_{11}& z_{12}& \cdots & z_{1n}\\ z_{21}& z_{22}& \cdots & z_{2n}\\ ⋮& ⋮& ⋮& ⋮\\ z_{m1}& z_{m2}& \cdots & z_{mn}\end{array}\right],i=1,2,\cdots ,m;j=1,2,\cdots ,n$$

(7)

where $X_{ij}$ is a matrix consisting of $x_{ij}^{\prime }$ and $W_j$ is a matrix consisting of $w_j$.

Determine the positive and negative ideal solutions:

$$Z^{+}=\left[z_1^{+},z_2^{+},\dots ,z_m^{+}\right]=\left[\max \left\{z_{11},z_{21},\dots ,z_{n1}\right\},\max \left\{z_{12},z_{22},\dots ,z_{n2}\right\},\dots ,\max \left\{z_{1m},z_{2m},\dots ,z_{nm}\right\}\right]$$

(8)

$$Z^{-}=\left[z_1^{-},z_2^{-},\dots ,z_m^{-}\right]=\left[\min \left\{z_{11},z_{21},\dots ,z_{n1}\right\},\min \left\{z_{12},z_{22},\dots ,z_{n2}\right\},\dots ,\min \left\{z_{1m},z_{2m},\dots ,z_{nm}\right\}\right]$$

(9)

where $Z^{+}$ means the largest number in each column constitutes the ideal positive solution, and vice versa.

Calculate the distance from the evaluation object to the positive and negative ideal solution:

$$D_i^{+}=\sqrt{{\displaystyle \sum _{j=1}^n{\left(z_{ij}-z_j^{+}\right)}^2}}$$

(10)

$$D_i^{-}=\sqrt{{\displaystyle \sum _{j=1}^n{\left(z_{ij}-z_j^{-}\right)}^2}}$$

(11)

where $D_i^{+}$ means the distance of the $j$ scheme to the positive ideal solution, and vice versa.

Calculate the relative closeness between the $i$th evaluation object and the ideal solution:

$${\epsilon }_i=\frac{D_i^{-}}{D_i^{+}+D_i^{-}}$$

(12)

Here, ${\epsilon }_i$ is the result of the static evaluation. As ${\epsilon }_i$ becomes larger, so does the level of megacity resilience.

3.3.2. TOWA and Dynamic Comprehensive Evaluation

Considering the panel data’s temporal characteristics, we propose introducing time-ordered weighted averaging (TOWA) to utilize it fully. A TOWA operator of dimension n is a mapping F [107]. $N=\{1,2,\cdots ,s\}$, $\langle t_k,a_k\rangle (k\in N)$ are called TOWA tuples. Within these pairs, $t_k$ is the time-order-inducing value and $a_k$ is the argument value.

$$F\left(\langle t_1,a_1\rangle ,\langle t_2,a_2\rangle ,\cdots ,\langle t_k,a_k\rangle ,\cdots ,\langle t_s,a_s\rangle \right)={\displaystyle \sum _{k=1}^s{u}_k}{b}_k$$

(13)

where $U={\left(u_1,u_2,\cdots ,u_s\right)}^{\mathrm{T}}$ is the weighted vector associated with F and $b_k$ is the $k$th time of collection of the aggregated objects $a_k$.

In this case, the time weight vector is computed using two methods: entropy value [108,109] and minimum variance [110]. The entropy of the time weights $I$, minimal variability $D_{}^2$, and the time degree $\lambda$ are core concepts involved.

$I$ measures the quality of information contained in the weight of the time dimension from the aggregated data. A larger $I$ means less difference between the weights of time dimensions, which implies a lower dispersion and uncertainty.

$$I=-{\displaystyle \sum _{k=1}^p{u}_k}\ln u_k$$

(14)

where $u_k$ is the weights of time dimension of the $k$th moment, and ${\displaystyle \sum _{k=1}^p{u}_k}=1,u_k\in [0,1],k=1,2,\cdots ,p$.

$D_{}^2$ is utilized to reflect weighting factor fluctuation in time series and is aimed at the most stable pair. The results obtained by using this method have a solid order-preserving property.

$$D_{}^2=\frac{1}{n}{\displaystyle \sum _{k=1}^n{v}_k^2}-\frac{1}{n^2}$$

(15)

where $v_k$ is the weights of the time dimension of the $k$th moment, and ${\displaystyle \sum _{k=1}^n{v}_k}=1,v_k\in [0,1],k=1,2,\cdots ,n$.

$\lambda$ describes the importance of time series at different moments and characterizes specific parameters in the process aggregation. As a rule, a smaller $\lambda$ means that more attention is given to recent data ($t_k$), and vice versa. When $\lambda =0.1$, this shows strong attention to recent data. When $\lambda =0.9$, this represents strong attention to distant data. A similar logic applies for intermediate values between the two adjacent judgments.

$$\lambda =\frac{1}{p-1}{\displaystyle \sum _{k=1}^n(p-k)}{u}_k$$

(16)

Dynamic evaluation is a form of integrated evaluation that incorporates the time dimension. It is used for city-related measures, such as urban sustainability and urban underground space resources [109,111]. The static and dynamic method of evaluating objects deals with the “utility measurement” and “rank” problems. While static evaluation addresses comparability among evaluated objects, dynamic evaluation addresses the comparability of a given evaluated object itself over time [112]. We chose the twice-weighted evaluation form. The first weighting emphasizes the role of the evaluation index at different times, in other words, the comprehensive evaluation values $y_i\left(t_p\right)$ at $t_p$ time. The second weighting stresses the function of the time factor to obtain $h_i$.

$$h_i=F\left(\langle t_1,y_i\left(t_1\right)\rangle ,\langle t_2,y_i\left(t_2\right)\rangle ,\cdots ,\langle t_p,y_i\left(t_p\right)\rangle \right)={\displaystyle \sum _{k=1}^p{u}_k}{b}_{ik}$$

(17)

where $U={\left(u_1,u_2,\cdots ,u_s\right)}^{\mathrm{T}}$ is the time weighting vector and $b_{ik}$ is the TOWA tuples’ $y_i\left(t_p\right)$ at the $k$ time.

An important step is figuring out the time weight vector. In mathematical terms, we described the entropy value and minimum variance method separately and converted them into nonlinear problems. Due to the complexity of nonlinear problems, the lingo 18 software was used to handle them.

$$\{\begin{array}{c}\max \left(-{\displaystyle \sum _{k=1}^p{u}_k}\ln u_k\right)\\ \mathrm{s}. \mathrm{t}. \lambda =\frac{1}{p-1}{\displaystyle \sum _{k=1}^n(p-k)}{u}_k\\ {\displaystyle \sum _{k=1}^p{u}_k}=1,u_k\in [0,1],k=1,2,\cdots ,p\end{array}$$

(18)

$$\{\begin{array}{c}\min \left(\frac{1}{n}{\displaystyle \sum _{k=1}^n{v}_k^2}-\frac{1}{n^2}\right)\\ \mathrm{s}. \mathrm{t}. \lambda =\frac{1}{n-1}{\displaystyle \sum _{k=1}^n(n-k)}{v}_k\\ {\displaystyle \sum _{k=1}^n{v}_k}=1,v_k\in [0,1],k=1,2,\cdots ,n\end{array}$$

(19)

3.3.3. The Obstacle Diagnosis Model

The obstacle diagnosis model often distinguishes factors hindering performance improvement [113,114,115]. We applied the obstacle diagnosis model to obtain the restricted megacity resilience and its obstacle factors to provide a basis for targeted decision-making.

$$M_{tj}=\frac{R_j\cdot P_{tj}}{{\displaystyle \sum _{j=1}^n\left(R_j\cdot P_{tj}\right)}}\times 100\%$$

(20)

The factor contribution degree $P_{tj}$ can be expressed by the portfolio weight ${\mathrm{w}}_j$ of the $j$th indicator. The indicator deviation degree $R_j$ is $1-{\mathrm{x}}_{tj}$, which represents the difference between the $j$th indicator and the ideal value.

4. Results

4.1. Megacity Resilience Comprehensive Assessment

4.1.1. Weights, MRCI, and Rank

Assuming $\beta =0.4$, we implemented Formulas (2)–(6) in Python and calculated the combined weights and total weight, as shown in

Table 4. Indicator weight of megacity resilience.

Measurement Items	Symbol	CRITIC Weight	Entropy Weight	Combined Weight	Total Weight
Number of early warning messages issued	P1	0.0457	0.0233	0.0323	0.1601
Direct economic losses from natural disasters	P2	0.0314	0.0184	0.0236
Population density	P3	0.0462	0.0294	0.0361
Public safety rankings	P4	0.0459	0.0358	0.0399
Ratio of industrial solid wastes comprehensively unutilized	P5	0.0377	0.0220	0.0283
Civil building quality	S1	0.0289	0.0322	0.0309	0.4071
Labor productivity of construction industry	S2	0.0340	0.0237	0.0278
Density of road network	S3	0.0443	0.0540	0.0501
Total annual volume of passengers transported by buses and trolley buses	S4	0.0441	0.0192	0.0292
Density of sewers	S5	0.0312	0.0392	0.0360
Ratio of waste water centralized treated by sewage work	S6	0.0312	0.0138	0.0208
Number of mobile telephones per 100 households	S7	0.0253	0.0219	0.0232
Beds of health care institutions per 1000 persons	S8	0.0394	0.0215	0.0286
Proportion of tertiary sector in GDP	S9	0.0266	0.0348	0.0315
Local general public budget expenditure	S10	0.0248	0.0298	0.0278
Aged 0–14, aged 65, and over-proportion	S11	0.0467	0.0300	0.0367
Proportion of staff in public management and social security organizations	S12	0.0381	0.0294	0.0329
Preparation of pre-disaster planning and organization of emergency drills	S13	0.0304	0.0324	0.0316
Number of student in regular HEIs per 10,000 persons	R1	0.0451	0.0475	0.0466	0.4328
Density of social organizations	R2	0.0348	0.0669	0.0541
Government decision-making capabilities	R3	0.0301	0.0273	0.0284
Regional collaboration capability	R4	0.0301	0.0457	0.0395
Insurance penetration	R5	0.0300	0.0469	0.0402
Collections of public libraries per 100 persons	R6	0.0390	0.0691	0.0570
Investment in municipal public infrastructure construction	R7	0.0344	0.0552	0.0468
Integrated government capacity	R8	0.0255	0.0413	0.0350
R&D internal outlay	R9	0.0473	0.0709	0.0615
Green covered area as %	R10	0.0318	0.0185	0.0238

. Among the primary indexes, the weights of response resilience and state resilience reached 0.4328 and 0.4071, respectively. They both significantly contributed to the resilience rating, but response resilience was slightly higher. At the same time, the index weight of pressure resilience was 0.1601, and its influence on the resilience rating was not apparent.

The standardized data and the combined weight were substituted into Formulas (7)–(12) to obtain the MRCI and response resilience index. The static measurement scores and trends from 2010–2019 are shown in

Table 5. Static time span and dynamic comprehensive resilience performance.

Region	${\mathit{y}}_{2010}$	${\mathit{y}}_{2011}$	${\mathit{y}}_{2012}$	${\mathit{y}}_{2013}$	${\mathit{y}}_{2014}$	${\mathit{y}}_{2015}$	${\mathit{y}}_{2016}$	${\mathit{y}}_{2017}$	${\mathit{y}}_{2018}$	${\mathit{y}}_{2019}$	${\mathit{h}}_{\mathit{I}}$	${\mathit{h}}_{{}^{{\mathit{D}}^2}}$	Rank
Zhengzhou	0.2763	0.2747	0.2952	0.3266	0.3701	0.4003	0.4494	0.5075	0.5418	0.5702	0.4685	0.4664	$\mathrm{IV}$
Nanjing	0.4202	0.4224	0.4740	0.4977	0.4848	0.5375	0.5650	0.5325	0.5318	0.6222	0.5427	0.5430	$\mathrm{IV}$
Dalian	0.3325	0.3433	0.3840	0.4022	0.3996	0.4171	0.4550	0.5223	0.5319	0.5008	0.4690	0.4692	$\mathrm{IV}$

and Figure 3.

Due to the emphasis on recent data, the experts recommended that time degree $\lambda =0.3$. Then, Formulas (18) and (19) were used to calculate the time weighting vector.

$$U_I={(0.0274,0.0347,0.0441,0.0559,0.0710,0.0901,0.1143,0.1450,0.1840,0.2336)}^{\mathrm{T}}$$

$$U_{D^2}={(0.0018,0.0236,0.0455,0.0673,0.0891,0.1109,0.1327,0.1545,0.1764,0.1982)}^{\mathrm{T}}$$

Next, the second weight came. The first evaluation results $y_i\left(t_p\right)$ were combined with the TOWA operator, and the dynamic evaluation result $h_i$ was computed using Formula (17).

Megacity resilience research has not yet established standardized evaluation criteria. Thus, we used the natural breakpoint method in Python to process the value domain of the sample data, resulting in the five resilience ratings, shown in

Table 6. Classification of resilience rating.

Resilience Rating	I	II	III	IV	V
Resilience index	[0.0000, 0.2834)	[0.2834, 0.3549)	[0.3549, 0.4585)	[0.4585, 0.6204)	[0.6204, 1.0000)
Rating meaning	lowest resilience	lower resilience	moderate resilience	higher resilience	highest resilience
Representational state	dangerous	vigilant	general	good	superior

.

The overall resilience level rose yearly, with there being occasional slight fluctuations in a few years. In Zhengzhou, the MRCI rose significantly (from 0.2763 in 2010 to 0.5702 in 2019), with the highest average annual growth rate being 7.51%. From its lowest level to its highest level, its resilience rating completed a transition. In Nanjing, the MRCI occupied the leading position over the study period. In 2012, it had already reached a high resilience performance. Likewise, it was the city that achieved the highest resilience rating. In Dalian, the MRCI was usually in the middle, but its average annual growth rate was the slowest, 3.50%, and it was outpaced by Zhengzhou from 2018 onwards. Compared to the dynamic evaluation result, Nanjing continued to lead, while Dalian and Zhengzhou switched places, with there being very little difference between them. All three cities were more resilient than average and obtained the same resilience rating. There was a difference, however: Nanjing was relatively stable in the middle of this resilience rating, while the rest had just entered this resilience rating and were yet to mature. Interestingly, response resilience closely resembled the overall resilience trend. In reality, several policies and implementation measures are tied to response resilience. Zhengzhou may benefit from combining defensive measures with ecological space [116]. The concept of resilience was introduced early to Nanjing, and its collaborative governance is constantly being updated. In addition, Dalian has failed to complete the strategy of moving from traditional engineering to recovery and adaptability.

4.1.2. Process Resilience Correlation

After obtaining the MRCI, we examined how pressure, state, and response resilience were correlated and aimed to reveal the internal mutual feedback mechanism. For data sets that did not fit either a linear or normal distribution, the Spearman coefficient is appropriate here. As used in a similar article, the Spearman correlation coefficient requires the existence of monotonic relationships between variables [117]. After the data were processed, the larger the index, the better the evaluation result is. TOPSIS characteristics were met, where all relationships were positive, and the focus was on the correlation strength. The table was checked, and the critical value of 0.368 was obtained to examine the correlation (p < 0.05).

As shown in Figure 4, resilience performance strongly correlated with state and response resilience, but not significantly with pressure resilience. This result was in line with our expectations. Pressure resilience mainly pertains to natural disasters and system service problems. These disturbances are more uncertain and complex and have obvious spatiotemporal changes that are harder to measure. However, state resilience encompasses support capability, covering both physical infrastructure and human elements. These resource endowments, in a broad sense, are relatively stable, which makes resilience possible [118]. Moreover, response resilience involves recovery skills and learning abilities. With this as its core, multi-party collaborative governance has shown promising development trends [119]. Further, a weak correlation was found between pressure resilience and state resilience. Sometimes, state resilience is seen as a characteristic of cities that gradually emerge from the process of disaster resistance. Additionally, state and response resilience were coupled to some degree. Response resilience links the soft and hard component state with transformative capacities over time [120].

4.2. Megacity Resilience Obstacle Diagnosis

4.2.1. Obstacle Factors at the Criterion Layer

The major barriers at the criterion and index levels were used to detect vulnerabilities and enhance megacity resilience. Using the obstacle diagnosis model (20), they were examined. Due to the more general nature of the criterion layer compared to the index layer, a time series analysis of the barrier diagnosis of the criterion layer was carried out. Figure 5 depicted their changes.

Regarding the compositional changes, the obstacle degree of pressure resilience was much lower than that of the state and response resilience, and the gap between the latter two was even smaller. The obstacle degree of pressure resilience varied greatly in different places and was usually irregular. It relates to the characteristics of pressure resilience itself. Local natural disasters fluctuate in frequency and intensity, and urban system services may also experience intermittent problems. Almost everywhere, the obstacle degree of state resilience is relatively stable. There is a need for the reconstruction or expansion of entities. However, it is hard to change entities in a short time. Additionally, it takes time to significantly change agents, such as the optimization of a population’s age structure and the allocation of reserve forces. However, barriers to state resilience were found to have changed in percentage terms. The total resilience identified was a prerequisite, and the increase in state resilience was because of the decrease in response resilience. Overall, response resilience is improving as impediments to it are reduced. As the most energetic kind of resilience, it plays a subtle role in the linkage between normalcy and emergency.

Considering the strength of the barrier degree, we examined the obstacle degree of response resilience further. Zhengzhou showed a continuous downward trend, with a total drop of 10%. There is no doubt that Zhengzhou achieved high performance in response resilience, which is crucial to securing overall resilience. Nanjing saw little change from 2010 to 2015. From 2015 to 2019, its decline was rapid, with the obstacle degree of response resilience reaching a historic low of 41%. Since Nanjing enjoys a high degree of openness and good urbanization development, the barrier degree of response resilience drops only if its response resilience is significantly greater than the state resilience level. In Dalian, the obstacle degree of response resilience remained roughly the same and still comprised more than half of the total. Clearly, response resilience is a major factor behind slow growth in overall resilience. As response resilience changes, green, development, and coordination ideas are more likely to be effective in governance. Zhengzhou put into practice integrated energy transformation projects in industrial parks and low-carbon transformations in public buildings, such as sponge cities. Recently, the Nanjing government has been successfully integrating several strategies to manage urban refinement governance in a highly efficient manner. However, Dalian has failed to avoid traditional industrial cities’ path dependence, and adaptive improvements to the city are not yet complete.

4.2.2. Obstacle Factors in the Index Layer

Differences in the barrier degree among sub-indicators may be obscured by the criterion layer. Therefore, we analyzed the index layer data after the time weight vector. When added up, the top five obstacle factors accounted for almost 40% of the total obstacle factors, meaning that they had a noticeable influence on the 28 basic indicators. Figure 6 displayed the degree of obstruction factors in the index layer.

In Zhengzhou, R6 had the highest barrier, a sign that the city did not take the appropriate measures to cultivate a resilient culture. As an active and accessible distribution channel, public library collections assist individual resilience cognition. This implementation is still insufficient and has failed to enable citizens to engage in city co-construction. The S3 obstacle factor ranked close behind. Zhengzhou is a railway and highway transportation hub, and its road density standards are naturally higher than other cities. This indicates that the traffic support capacity of Zhengzhou is not high enough, thereby losing some functions. In addition, P4 demonstrates that the public safety situation is not optimistic, and security services are under strain. A lack of security inspections and disputes in the management and control process may be to blame for this. As P3 and S11 show, it has always been challenging for Zhengzhou to manage its population. Because of its historical foundations and customs, its population density is extremely high, which puts much pressure on the availability of basic services. In terms of its age structure, the proportion of young adults offers the city no advantages. One reason for this may be labor turnover due to unattractive recruiting resources.

In Nanjing, surprisingly, the obstacle degree of R9 ranked first. Funds allocated to R9 are generally divided into three categories. Comparatively, the proportion of funding for basic research was the lowest, and applied research did not advance, but rather retreated. Despite the apparent advantages of local technological resources, the funding structure still needs to be optimized, especially for applied research. Similar to Zhengzhou, R6 also had a higher barrier. Collections in public libraries have not yet reached the point where residents can easily access books, and resilience guidance is not provided to residents enough. The S12 barrier was ranked third. Being part of a public administration organization results from shared values and life orientations. Currently, the proportion of such practitioners is low, which undermines the strength of the emergency reserve. S11 appeared in Nanjing, but was unlike that in Zhengzhou. This is due mainly to Nanjing’s increased vulnerability regarding its aging population. R5 denotes a large, but not strong insurance industry. Families sometimes do not take out insurance in anticipation of disaster assistance, which may explain this.

In Dalian, R2 was found to be an obvious obstacle. Frequently, a limited spatial distribution and insufficient density prevent social organizations from accumulating resources via collective activity, which hinders sustainability. R1 concerns the individual. A high education level and access to resources is a straightforward way to promote self-help. Unfortunately, the educational backgrounds of residents are still not equipped enough to deal with this threat. In addition, the R7 barrier implies a lack of infrastructure renovation. Dalian failed to learn from experience and optimize the layout and expansion of its urban infrastructure. Concerning S5, a contradiction was found between the efficacy of the drainage network and the mismatched needs, partly due to the delay in rebuilding municipal infrastructure. The obstacle degree of R5 was the same as that of Nanjing. Insurance penetration promotes resilience through the diffusion of responsibility, underwriting a return to the diversity of livelihoods. Currently, insurance penetration in Dalian does not complement a well-institutionalized environment for mitigating the adverse effects of natural disasters.

Specifically, megacities share similarities concerning the popularity of resilience culture and civic awareness, but are unique in other aspects. Each one differs in the strategies it employs for guaranteed services, the capacity it has for R&D construction, and the way it applies the development concept.

5. Discussion

5.1. Regional and Process Differences in the MRCI

The resilience levels of the three megacities studied have not developed at the same pace. Under the same size, the differences may stem from these cities’ regions. Regional differences exist in the functional positioning and development stages. This finding is consistent with city resilience in the eastern region of China being significantly higher than in the central and northeastern regions [76]. Judging from the MRCI subsystem, response resilience is found to play an essential role in narrowing the resilience level gap. Several reports have shown that resilience governance is thought to adopt a self-organization and participatory approach, which is precisely what response resilience stresses [64,121]. The structure for gauging megacity resilience compels us to consider processes. Weak pressure resilience involves more uncertainty. Consistently, megacities are prone to compounding disturbances, such as the mix of climate change and human activity [122]. In this study, state and response resilience were coupled to some degree. State resilience in Nanjing is related to its good resource base of entities and agents. The accompanying higher response resilience adapts configurations of entities and agents to changing conditions [58].

In our institutional context, agent collaboration needs to be facilitated. The government forms a situation where resources are allocated reasonably and information is communicated effectively. In addition, organizations and citizens brainstorm ideas to develop explicitly operational definitions and concrete baselines, facilitating decentralization and a “bottom-up” collective resilience management system. Resilience process assessment should be addressed. The lifecycle should be designed to minimize pressure’s adverse consequences, ensure the state’s robustness, and strengthen the response subject. Specifically, dynamic process assessment should be carried out, the construction effect should be tested, and the planning content and plans should be adjusted in time. Moreover, we should clarify the functions and tasks of various factors at stages to set policy priorities. We should be up-to-date on the latest research findings, such as disaster prevention and emergency management expenses, to dynamically reflect advances.

5.2. Resilience Improvement Paths Based on Key Obstacle Factors

Strategies tailored to common barriers: In general, strengthening response resilience is the key to improving the MRCI going forward. Specifically, promoting a resilient culture and citizenship must be urgent. A common approach for megacities is to build a resilient community. Participating in community planning and reconstruction enables citizens to cultivate their value identity of megacity resilience culture. In addition, group space preparation can create a communal network effect in a resilient community. Disaster preventive science guides and evacuation exercises teach inhabitants about resilience and proper conduct.

Follow characteristics “point to the area”: In Zhengzhou, the focus is on the urban population’s vulnerability and the ensuing potential conflict between dialect and perception. We should take a pluralistic and inclusive approach to fix issues, such as strengthening quarter-hour convenient living circles and regulating the age structure of communities. In Nanjing, the investment in applied research and the implementation of Big Data should be strengthened to gain access to intelligence information. Using these data, an integrated platform for disaster risk prediction and scenario simulations will naturally develop. In Dalian, a progressive and adaptive development model should be created to reduce the adverse effects generated by its rough industrial development. In order to undertake resilience construction, defensive measures in conjunction with ecological space must be used, such as green corridors.

5.3. Applicability, Limitations, and Future Research

Resilience-related findings can validate the research results of this paper. For example, policies such as urban ecological management and collaborative regional governance have contributed to the resilience of Zhengzhou in recent years [123]. This result accords with our earlier observations in Zhengzhou, which showed a quick resilience growth ratio. Nanjing has been proven to have the highest urban flood resilience in the Yangtze River Delta [124]. The slowest growth and the most mature resilience level in Nanjing are tested using the Theil index, Morans I, and the spatial Durbin model [125]. These studies further support our findings of resilience in Nanjing. In addition, the resilience improvement critical factors of connectivity, multi-stakeholder coalitions, and civic awareness have been confirmed by previous studies on the megacities of New Zealand [126], African countries [127], and Australia [128]. Our content on response resilience has a close connection to this idea. As a result, this paper’s research framework and policy implications can serve as a reference for domestic and foreign megacities.

In this study, we adopted a megacity resilience index system and integrated three solution sets from typical megacities. However, there were several limitations, including uncertainty pressure in the more precise metrics, criteria weighting method expansibility with more accuracy, and competing priorities regarding improved resilience. The critical next steps are scaling up the megacity resilience index system, optimizing measuring methods, and ensuring the solutions are complementary. Further experiments using a broader range of megacity cases could shed more light on the general case. Introducing an RBF-WINGS approach (combining artificial neural networks and an improved decision-making test and evaluation laboratory) could be useful [129]. VIKOR could also be included. This takes into account the relative importance of positive and negative ideal solutions and incorporates group utility maximization and individual regret minimization [130]. Furthermore, the experience mining approach could be applied to identify the most similar backgrounds between the target cases for solving problems.

6. Conclusions

This paper proposes a conceptual framework based on the PSR framework, mapping the megacity resilience of process classification and specific connotations. Process attributes comprehensively reflect the formation, status, and feedback of the MRCI. State resilience and response resilience are strongly related to megacity resilience, with the latter being even more associated with this and exhibiting comparable resilience tendencies. When pressure resilience is coupled with state resilience, state resilience is coupled with response resilience. When any change in one or more pressure occurs, state and response components will complement the overall structure or affect an orientation, manifesting as an open spiral circular chain relationship.

A megacity resilience index system designed following applicability was verified using empirical research. It identified resilience improvement paths based on key obstacle factors. Despite there being better response resilience in Zhengzhou and Nanjing, Dalian still had a high barrier to response resilience in the criterion layer. In the index layer, common barriers reflected the popularity of resilience culture and civic awareness. Characteristic obstacles were weak security services caused by population pressure in Zhengzhou, insufficient technology due to unevenly applied research budgets in Nanjing, and the lack of ecological application resulting from industrial path dependency in Dalian.

There are theoretical and practical contributions. Theoretically, this study enriches the theoretical resilience level both in scope and content. It expands the application of PSR theory from single-hazard scenarios and single-component resilience to comprehensive megacity resilience. Moreover, it also contributes to establishing short-term and long-term linkages and grasping the intricate nature of process resilience. Practically, it condenses the complexity of the research problem into a compact and manageable amount of information and provides effective content and reasonable quantities of indicators for operability. Further, the integrated resilience assessment and obstacle judgment method are highly accurate, replicable, and generalizable, which can serve as reference tools for similar megacities in developed or developing countries.