A Framework for Resilient City Governance in Response to Sudden Weather Disasters: A Perspective Based on Accident Causation Theories

Abstract

With climate change, urban resilience is becoming a critical concept for helping cities withstand disasters and accidents. However, current research often focuses on concept identification, leaving a gap between concept and implementation. This study aims to investigate the lack of urban resilience in the face of sudden weather disasters, with a focus on the inadequate capacity of urban systems to effectively govern such events. The Zhengzhou subway flooding accident on 20 July 2021, serves as a case study for this research, and the accident causation theories, such as the Swiss cheese model, Surry’s accident model, and trajectory intersection theory are used to conduct a comprehensive analysis of the accident’s causes. Through this analysis, the paper identifies vulnerabilities in the natural, technical, and man-made systems of the urban system, and reveals deficiencies in four aspects of urban resilience: natural, technological, institutional, and organizational. Based on this analysis, the study proposes a resilient city governance framework that integrates the “Natural-Technical-Man-made” systems, offers relevant recommendations for urban resilience governance, and discusses potential challenges to urban resilience implementation.

1. Introduction

1.1. Chinese Cities Will Face More Frequent Weather Disasters

Climate change has led to a significant increase in the number of weather, climate, and water disasters globally. According to the World Meteorological Organization (WMO), the number of weather, climate, and water disasters suffered globally has increased 1.8 times in the last 20 years (2000–2019) compared with the period of 1980–1999 [1]. These disasters have caused significant human casualties and economic losses, particularly from floods and storms.

China, the country most affected by weather, climate, and water disasters in Asia, is particularly vulnerable to these disasters due to its monsoon climate and topographic environment. Between 1979 and 2019, such disasters in China resulted in 90,624 deaths and nearly $600 billion in economic losses, accounting for almost half of all economic losses in Asia [1]. China will experience more frequent and intense climate hazards with climate change in the future [2].

China has also experienced rapid urbanization in the 21st century, with the urbanization rate increasing from 36.22% in 2000 to 64.72% in 2021 [3]. As an aggregate of ecological, social, and technological systems [4], cities are vulnerable to natural hazards, which are further compounded by increasing urban population density and climate change [5,6]. In the future, Chinese cities will need to cope with more frequent and severe climate hazards, and it will be necessary to improve the disaster governance capacity of urban systems in order to protect people and reduce economic losses from disasters.

1.2. China Introduced the Concept of Resilient City

In recent years, in response to the vulnerability of urban systems, the concept of resilience has been widely applied to various fields such as natural disasters, urban risks, and environmental risks. Resilient cities require cities to be able to absorb, adapt, and respond to changes in response to disasters or disruptions based on urban planning, policy development, and strategic directions. This allows cities to mitigate the damage and impact of disasters, maintain the basic functioning of cities, guarantee the provision of essential services and functions, recover quickly after disasters, and ensure sustainable development [7,8,9,10]. Thus, by improving urban resilience, cities can better withstand disasters and protect residents.

The construction and development of resilient cities have gained increasing international attention as a way to prepare for extreme disasters caused by climate change. Between 2010 and 2018, the United Nations and its organizations, such as UNISDR and UN-HABITAT, promoted the concept of resilient city in the terms of guidelines and relevant policy to enhance its role in urban disaster risk management, infrastructure development, and urban development [11,12,13]. The Chinese government has also embraced this concept and included “Building Resilient Cities” in its 14th Five-Year Plan in 2021, with the aim of improving the disaster resilience of Chinese cities.

However, building truly resilient cities remains a challenge as they are complex systems made up of multiple interconnected systems, such as the economy, society, institutions, environment, and nature [14,15]. Attention has been focused on identifying the concept of resilient cities, but not on implementing it in practice, leading to a disconnect between the concept and urban risks, threats, and vulnerabilities [16]. There is a lack of research on aspects of implementation such as policy development, urban resilience assessment, and urban disasters. The concept of urban resilience is still vague and requires different definitions across disciplines, such as ecological, social, and engineering [17] to form a suitable definition to guide resilient city construction in practice.

Despite this, policymakers have begun to apply the vague concept to city design, such as the Resilient City 100 Program (RC100) supported by the Rockefeller Foundation, but some scholars question whether this leads to “superficial resilience” that only offers cities a “brand” for urban development and tourism, rather than actual adaptation to changing environments [18,19]. More research is needed on both the concept and its implementation to bridge the gap between the concept and urban risks, threats, and vulnerabilities.

This study aims to investigate the lack of urban resilience in the face of sudden weather disasters, with a focus on the inadequate management capacity of urban systems to effectively govern such events. The Zhengzhou subway flooding accident on July 20, 2021, is used as a case study. Multiple accident causation theories are used to analyze the roles and responsibilities of stakeholders and identify challenges in implementing resilience measures. A resilient city governance framework is proposed based on the “Natural-Technological-Man-made” system for urban disaster resilience governance and recommendations for improvement are provided. This research adds to the existing literature on urban resilience implementation and can be used by urban systems to improve their disaster preparedness and response efforts.

This paper is organized into six sections. Section 2 provides a brief review of the current research on resilient cities and urban disaster risk, and explores sponge city programs in China. Section 3 presents the research methods and information of the case accident. Section 4 presents the findings of the case using multiple accident causation theories. Section 5 discusses the results, proposes a framework and recommendations for urban resilience governance, identifies the impediments to implementing urban resilience policies. Finally, Section 6 summarizes the paper’s conclusions and presents limitations and future research perspectives.

2. Literature Review

2.1. Urban Resilience and Urban Disaster Risk

Recent research on urban resilience has covered topics such as climate change, urban planning, urban hazards and disaster risks, urban sustainability and green infrastructure, the adaptation of urban resilience, and urban resilience and smart cities [20]. These topics are all revolve around urban hazards and disaster risks. In this field, researchers and practitioners have focused on natural and man-made disasters in urban areas, such as floods, earthquakes, and landslides, and have worked on identifying risks, developing strategies, and discussing post-disaster urban development [20]. Climate change has increased the number of uncontrollable urban emergencies and disasters, and green infrastructure, sound urban planning, and smart city systems can help cities improve their resilience and reduce disaster risks, some researchers also emphasized the importance of urban disaster resilience for city sustainability [21].

The government is a key player in urban resilience and disaster governance. Silva, et al. [22] found that when urban resilience increases, the government places more emphasis on the social and urban planning sectors, as both are necessary to maintain a broad perspective and analyze urban resilience holistically in addition to building safe infrastructure. However, communication issues between government departments and lack of capacity building can be problematic. A spatial planning process framework developed by Orsetti, et al. [23] addresses these issues by engaging more stakeholders and providing a comprehensive tool for assessing vulnerabilities and urban spaces. This framework can help the government improve communication and capacity between urban planning and health departments.

Therefore, many scholars argue that involving more stakeholders in urban resilience development and disaster governance is important. Ke, et al. [24] argued that cities cannot rely solely on public power to resist disasters and that individuals, schools, businesses, non-profit organizations, and other groups should participate in urban disaster resilience efforts. Pirlone, et al. [25] argued that recent studies have focused too much on natural disasters and not enough on man-made disasters. Governance for man-made disasters requires the cooperation of more stakeholders, such as government, research institutions, businesses, citizens, etc. This study also highlighted the role of the Urban Emergency Plan in the emergency and prevention phase. Lu and Li [26] argued that government-led resilient city programs are more focused on disaster prevention than on the needs of affected people. Therefore, post-disaster development requires the involvement of non-governmental organizations (NGOs), they highlighted the contributions of NGOs to post-disaster development, such as the construction of infrastructure, capacity, mechanisms, and culture. This study provided support for future government-NGO collaboration in disaster governance.

Banai [27] argued that the COVID-19 pandemic exposed the vulnerability of urban systems but also serves as a change agent for the planning of resilient cities and regions globally. Similarly, Wang, et al. [28] argued that the COVID-19 pandemic poses a major challenge to city and community resilience and emphasizes the important role of community medical facilities in improving resilient community construction during large-scale public health emergencies.

In addition, some scholars also construct disaster resilience model to defend against specific disasters like flood or earthquake based on resilience model. For example, Saikia, et al. [29] and Xu, et al. [30] developed tools to improve the assessment and governance of urban water resilience; Maroufi and Borhani [31] focused on community resilience, constructing a framework to evaluate the ability of communities to cope with earthquakes and enhance their seismic resilience.

2.2. Resilient Cities and Sponge Cities

Before incorporating the concept of resilient city, China has been actively exploring its own path to construct resilient cities and govern urban disasters. In 2015, the General Office of the State Council launched the “Guidance on Promoting Sponge City (SPC) Construction” program to improve cities’ resilience to flood. Then 30 Chinese cities have been selected as pilot cities for this program, and over 130 cities (including Zhengzhou) have formulated SPC construction plans under the guidance of this program.

Many countries and regions are exploring water management systems to mitigate flooding disasters and manage resources, discharge, and develop rainwater, while restoring urban ecosystems (in residential areas, parks, etc.). Yin, et al. [32] indicated SPC is similar to Low-impact development practices (LID) and sustainable approaches developed for urban drainage systems, but as an integrated scheme for comprehensive urban water problems, it has a broader spatial dimension and addresses wider issues. The concept is based on green infrastructures (green roofs, rain gardens, etc.) and a mid-way drainage grid optimization system (considering water quality, quantity, etc.) addressing water security and ecology. Through building a regionalized flood control system, protecting biodiversity, restoring habitats, and creating green tourism networks, sponge cities can address urban floods and droughts.

It has been seven years since the SPC policy was implemented in pilot cities and research has developed rapidly. Scholars are debating the future direction, highlighting issues in policy implementation, and evaluating SPC infrastructure schemes. SPC is seen as a challenge and an opportunity, but obstacles in its implementation must not be ignored. Chan, et al. [33] argued it will have a revolutionary impact on land use, water resources management, urban flood hazards and climate/environmental risks, ecology, and social welfare in China. Wang, et al. [34] argued it also has implications for developing countries undergoing urbanization. Challenges and practical problems, such as technical complexity, management awareness, knowledge and learning capacity, participatory and integrated governance, investment and financing, implementation pathways, planning and organization, project evaluation systems, and public perception and support, have been identified [33,35]. Investment and financing, insufficient funding, and uncertainty in SCP planning are severe issues that could lead to failure [36]. Global climate change and urbanization’s potential impact on SCP programs are additional challenges [34]. Scholars believe the future of SPC is toward smarter, greener, and more harmonious cities [32,34].

The lack of research on SPC assessment systems has led to an increase in studies on the subject of infrastructure assessment in recent years. Many scholars have developed various comprehensive assessment frameworks for green infrastructure based on the Storm-water Management Model (SWMM) [37,38,39,40]. These studies help decision-makers evaluate sponge city infrastructure and select the appropriate facilities or framework to be applied in SPC construction.

The construction of SPC does not conflict with the concept of urban resilience. It is one of China’s exploration paths for new urban construction, highly overlapping with urban resilience and thus improving cities’ ability to withstand flood disasters. SPC mainly focuses on physical city systems, improving infrastructure. But resilient cities require more than physical systems; they need an urban management system with the capacity to cope with disasters. This is a necessity in current urban systems. Therefore, Chan, et al. [41] argued that future urban construction in China should combine the concepts of Resilient Cities and Sponge Cities to build Resilient Sponge Cities.

3. Research Methods and Accident Replay

3.1. Research Methods

This paper uses qualitative research methods to comprehensively analyze the causes of the case, identify the subjects involved in each dimension of urban resilience and their responsibilities, and explore implementation problems of resilient cities. The paper then proposes recommendations for resilient city governance.

Accident causation models, as tools to explain accident mechanisms and assess risks, can help urban systems explore accident causation and support the construction of resilient cities. The Domino Model, proposed by Heinrich, et al. [42], is one of the earliest accident causation models. It suggests that accidents are caused by a sequence of events, but removing one of the key factors, such as unsafe conditions or behaviors, can prevent a chain reaction of accidents. This model laid the foundation for later chain accident causation models. As the field of accident causation modeling has developed, other models have been proposed, such as the Swiss cheese model [43], the Surry’s model [44], the “2-4” model of behavioral safety [45], and the trajectory crossover theory [46] model were proposed. These models have evolved from simple chains to complex chains and to systemic networks, becoming a more comprehensive model system.

The Zhengzhou subway flooding accident (referred to as Zhengzhou Metro accident hereinafter) was caused by a combination of factors. The complexity of the urban rail transit system [47] and the meteorological disaster created a dynamic and nonlinear interaction of multiple systems, such as human behaviors, urban technical facilities, and organizations, influenced by the natural environment. This interaction ultimately led to the accident. To better understand the causes of the accident, this paper uses several models, including the Swiss cheese model, Surry’s model, and trajectory crossover theory to conduct a comprehensive analysis. These models allow this study to gain a more realistic understanding of the factors that contributed to the accident and identify any weaknesses or vulnerabilities in the various aspects of the urban system in response to this weather disaster.

3.2. A Brief Introduction to Zhengzhou

The accident occurred in Zhengzhou, in Henan Province, China. Henan is in the middle of China and is known for its transportation hub. The terrain is mostly mountainous in the west, and lower in the east. Its northern, western and southern borders are bordered by the Taihang Mountains, Funiu Mountains, Tongbai Mountains, and Dabie Mountains, and it is crossed by the four rivers (Hai, Yellow, Huai, and Yangtze). Zhengzhou City is north-central, at the junction of the Yellow River’s middle and lower reaches. It has high elevations in the southwest and low in the northeast, and its plains divide into eastern and western regions. The eastern plain, where the accident occurred, is home to Zhengzhou, Zhongmou and Xinzheng districts and counties. Its climate is monsoon-influenced temperate continental, with hot summers and concentrated precipitation from June to August. It also has 124 rivers, including the Yellow and Huai Rivers.

By 2021, Henan Province had a population of 98.83 million [48] and Zhengzhou City had a population of 12.74 million [49]. At present, Zhengzhou has 8 metro lines, with the busiest being Line 5—a 40.7km ring around the city center with Yuejigongyuan Station as its first and last station [50]. It carried an estimated 300,000 passengers daily, followed by Lines 1 and 2. The operational map of Zhengzhou metro is shown in Figure 1.

3.3. Accident Reply

The Zhengzhou Metro accident was a serious accident in the urban rail system, caused by extreme rainstorms and urban flooding. This accident aggravated the major impact of flooding on the city residents. It demonstrated the challenges of effective risk governance in urban systems during meteorological disasters. This incident highlighted the need for flexible and adaptive strategies to address the complexity and dynamism of weather-related disasters and their effects on urban systems.

From 17–23 July 2021, Henan Province was hit by severe rainfall, resulting in extensive urban flooding. The disaster resulted in 398 reported deaths, with 380 in Zhengzhou. On 20 July, Zhengzhou suffered major casualties and economic losses, including the Zhengzhou Metro accident that caused 14 fatalities. Between 17:00 and 18:00, Metro Line 5 was traveling from Haitansi Station to Shakoulu Station when floodwater rushed into the metro tunnel from the Wulongkou parking lot, a stop and maintenance area for metro trains. This caused over 900 commuters to be trapped, with more than 400 of them detained for over three hours. A map of the accident site and the Wulongkou parking lot is shown in Figure 2, and a summary of the accident is provided in Figure 3. The information is derived from the “Investigation Report on the ‘7·20’ Extraordinary Rainstorm Disaster in Zhengzhou, Henan Province” published by the State Council Disaster Investigation Team of the Ministry of Emergency Management of the People’s Republic of China (MEM) [51], as well as relevant information compiled by The Paper News [52].

4. Results Analysis

By summarizing and organizing the information about the Zhengzhou Metro accident, it is possible to identify three main systems involved in the incident: the Natural system, the Technical (equipment) system, and the Man-made system.

4.1. The Swiss Cheese Model

The Swiss cheese model [43] divides the safety system into four aspects: environmental factors, management factors, security risk factors, and unsafe behavior factors, and compares it to layers of cheese, in which the “holes” are the defects in the defense system, and when all the “holes” are in alignment, the safety system fails, and the accident or disaster is like a light that passes through the holes in the layers of the defense system and causes serious consequences.

The Swiss cheese model was applied to analyze the causes for the Zhengzhou Metro accident, as shown in Figure 4.

• Environmental factors (Natural system)

Henan Province is susceptible to long periods of heavy rainfall due to its geography, influenced by Taihang Mountains, the Funiu Mountains, and the atmospheric pressure and westerly wind belt. In August 1978, Henan also experienced extreme flooding. From 17 July 2021, Henan was hit by record-breaking extreme heavy precipitations influenced by a typhoon. Before the Zhengzhou Metro accident, from 16:00 to 17:00, the average hourly rainfall reached 201.9 mm, a record for hourly meteorological observation of rainfall in mainland China. The extreme rainfall directly led to the accident and hindered rescue.

2.

Management factors (Man-made system)

• The government of Zhengzhou failed to recognize the unique characteristics of the region, lacked effective disaster prevention measures, and had weak emergency response and management capabilities. Furthermore, disaster prevention deployment was incomplete, and warning and response actions were delayed. Consequently, many citizens were unaware of the danger posed by the heavy rainfall and chose to return home by public transportation without caution.
• Zhengzhou Metro Group Co. has inadequate emergency response capabilities, evidenced by a lack of understanding of the maximum risk subway equipment can withstand. This has caused poor decision-making and inadequate risk screening. For instance, despite red warnings of heavy rainfall from 19–20 July, Zhengzhou Metro Group Co. failed to strengthen inspections or recognize the potential danger of flooding. Furthermore, it was not until people were trapped in the metro for over an hour that the Zhengzhou Metro Branch reported to Zhengzhou Metro Group Co.
• The inadequate resources and coordination caused a delayed response to the disaster. Trapped passengers sent out distress signals via social media and emergency hotlines, but had to wait several hours before the rescue teams arrived. Flooded tunnels further hindered the progress of the rescue teams, leading to a delayed response.

3.

Security risk factors (Technical system)

• Unauthorized changes to the railway line design by Zhengzhou Metro Group Co. without approval caused subsidence of 1.973 m near the Wulongkou parking lot, violating Metro Design Code and not reported to regulatory authorities. This depression was prone to waterlogging, resulting in a breach of the waterproof wall and flooding of the subway entrance. The train was forced to stop between Haitansi Station and Shakoulu Station.
• The waterproof walls in the Wulongkou parking lot have no ability to withstand flooding. Some waterproof walls were not high enough due to miscalculations; others were of poor quality because of not built according to the original drawings; and others were replaced with temporary construction fences. Overall, the inadequate supervision by Zhengzhou Metro Group Co. led to the wall’s collapse under the weight of the heavy rain.
• The drainage ditch near the Wulongkou parking lot was not functioning properly due to ineffective management and infractions of regulations, including the installation of a 58-m-long cover plate and the accumulation of construction waste. These factors led to the waterlogging and flooding that occurred in the parking lot during the rainstorm.
• The accident highlighted the lack of safety features in Zhengzhou subway tunnel and vehicles, notably the lack of escape doors and inadequate emergency resources. This caused passengers and staff to pull open cab and carriage doors to exit and impeded air circulation. And the flooding impeded rescue efforts by pushing the train off the tracks, creating a distance of more than 2 m between the train and escape path.

4.

Unsafe behavior factors (Human system)

• The Zhengzhou government’s response to severe weather warnings was inadequate; despite the Central Weather Station issuing consecutive red warnings of heavy rainfall, the government failed to suspend work and school activities in a timely manner. On 20 July, the second red warning was issued at 6 a.m., yet no measures were taken to protect citizens from potential flooding and other hazards. This negligence exposed the public to unnecessary risks.
• The government and transport authorities’ inadequate response to the emergency was highlighted, with Metro Line 5 remaining open when other lines shut down at 17:00 p.m., despite severe weather. The Transport Commission and Emergency Management Bureau failed to consider safety hazards of rail transportation in flooding, disregarding the metro operating company’s request for an emergency shutdown. Prioritizing convenience and accessibility over safety exposed passengers to risks, resulting in the accident.
• The Metro Operating Control Center’s (OCC) poor judgement to emergency situations was revealed by the accident. Their inadequate assessment and response to the emergencies led to mistakes in instructions, such as allowing the train to continue on its route without understanding the risks of flooding in the tunnel, and misdirecting the train to back off 30 m. This resulted in the train losing power and stopping in a lower area.

4.2. Surry’s Model

Surry’s accident model [44] is a useful tool for analyzing the underlying causes of risk formation in complex systems. It divides the accident process into two stages: danger formation and danger release. Human information processing steps—such as sensing, cognition, and behavioral responses—can prevent or mitigate hazards in the danger formation stage. However, incorrect steps can lead to exposure. In the danger release stage, correctly executed steps can avoid or mitigate the danger; if not, the danger may become harm or damage. The Zhengzhou Metro accident involved an interacting railway safety system that incorporated “Human-Machine-Environment” systems. The “Human” system referred to passengers, staff, and rescue personnel responsible for safety, emergency response, etc. The “Machine” system referred to subway vehicles, signaling systems, etc. The “Environment” system covered the train’s internal environments, tunnel environment, etc. Figure 5 illustrates the model applied to the Zhengzhou Metro accident.

• The danger formation stage analysis of subway flood

The inadequate response to danger signals and flaws in the urban metro management system at the danger formation stage are key factors that contributed to the accident. The safety system was designed to detect and respond to signals that may indicate the formation of a hazard; however, many signals were released (e.g., reports of flooding in the Metro Line 5 tunnels, the shutdown of Metro Line 1 and 2, and the flooding of the Wulongkou parking lot) and not correctly interpreted and acted upon, which increased the likelihood of an accident.

The accident was due to the failure of relevant metro departments to address risks promptly. They failed to spot the potential sources of danger in time and to take necessary action when alerted. These shortcomings point to flaws in urban metro management, e.g., inadequate mechanisms for dealing with accidents, inadequate emergency plans, and weak risk decision-making.

2.

The danger release stage analysis of subway flood

After the train lost power, the tunnel’s water level rose, prompting metro staff and passengers to quickly seek emergency shelter. Metro staff assessed the situation and gathered passengers at the front of the train, opening the cab and first carriage doors to evacuate some passengers to the side escape path. These passengers reached the Shakoulu Station, escaping the danger.

However, the rising water level in the tunnel made the flow more rapid, and most passengers were unable to evacuate. They returned to the carriage and waited for the rescue team. The Zhengzhou Metro Branch didn’t report the situation to the Zhengzhou Metro Group Co. until 19:48, delaying the rescue time. The remaining passengers were trapped for over an hour, and the water level continued to rise, entering the carriages and rising to a man’s (175 cm) chest. The train was pushed off the track, making it two meters away from the side escape path. By 22:00, all passengers were evacuated, but they had been in the water for over three hours. Some died due to lack of oxygen, loss of temperature, and other factors, even after resuscitation.

4.3. Trajectory Crossover Theory

The trajectory crossover theory investigates the causes of accidents [46]. This theory posits that two factors, human and material, are in separate trajectories but can influence and transform each other. When the trajectories of these two event chains—human unsafe behavior and material unsafe conditions—intersect in time and space, accidents can occur. Inadequate management, lack of inspection, and safety equipment imperfections can lead to material unsafe conditions, while operator lack of training, improper operation, and physical reasons can cause human unsafe behaviors.

The theoretical model figure of trajectory crossover theory for the Zhengzhou Metro accident is shown in Figure 6.

• Fundamental causes analysis (Natural system)

Starting on 17 July, multiple regions in Henan Province experienced unprecedented heavy precipitation events, with the maximum single-day rainfall in Zhengzhou exceeding 500 mm and 720 mm accumulated over three consecutive days.

2.

Indirect causes analysis (Technical system and Man-made system)

The Wulongkou parking lot suffers from multiple known issues, such as low terrain due to metro design changes and ineffective waterproof walls. In addition, the government’s disaster emergency publicity and education are inadequate, resulting in residents lack disaster resilience. The government’s limited capacity to execute emergency plans makes them delay time in decision-making and allocating resources during critical moments. Insufficient escape drills and limited emergency management capacity of metro-related departments pose further challenges.

3.

Direct causes analysis

First, analysis of human unsafe behaviors revealed four issues: (1) Metro Line 5 not stopped promptly; (2) OCC commanders made errors in operations; (3) insufficient rescue resources for metro staff to transfer all passengers; (4) Zhengzhou Metro Branch delayed rescue time. At 18:47, water level in the carriage was low, but insufficient resources prevented all passengers from being transferred safely. At 19:48, the branch reported the danger, but the passengers had already been trapped for an hour, and water had risen to the chest of a 175 cm passenger, putting them at risk.

Second, four issues are revealed in material unsafe conditions: (1) no escape doors in metro vehicles; (2) power loss in a low-lying area; (3) high water level in tunnel; (4) risk of electrical leakage. Passengers were unable to break windows due to high water level and felt suffocated and helpless. Turbulent water flow pushed the train off the track, two meters from the side escape path, forcing passengers to step on rescuers’ shoulders through the flooding water. Electrical leakage posed a risk, passengers were afraid of moving.

4.4. Comprehensive Analysis

The models and theories discussed above can be grouped into three categories: Natural system, Technical (Equipment) system and Man-made system, as follows.

• Natural system

Henan Province’s unique geography and environment make it susceptible to steady heavy rainfall, and typhoons exacerbate the situation, leading to unprecedented precipitation events starting on 17 July. This caused urban flooding then directly triggered the accident and hindered rescue efforts.

2.

Technical (Equipment) system

Unauthorized design changes to the Wulongkou parking lot caused it to be in a low-lying area with serious waterlogging. Piled construction waste reduced the ditches’ drainage capacity and illegal cover plates further decreased water collection. The waterproof walls around the parking lot were substandard—first, they used temporary construction walls, and second, they did not build the foundation according to the drawings—led to flooding water overrunning the walls and reaching the subway entrance.

The train lacked escape doors but had a side escape path, necessitating passengers to pull open the doors with metro staff, affecting escape efficiency. The situation was compounded by the train being in a lower area with deep flooding, leading to a decrease in oxygen levels and increased risk of hypothermia. The flooding eventually pushed the train off the track and away from the side escape path, making rescue even more challenging.

3.

Man-made system

The city government bore major responsibility for the accident due to lack of preventive deployment, unsound flood prevention plans, inadequate disaster warnings, and weak emergency response and disposal capabilities. Most citizens did not recognize the risk of the extreme rainstorm disaster. The lack of coordination between official and unofficial rescue organizations resulted in uncoordinated deployment of resources, delaying the rescue of the trapped passengers.

The Zhengzhou Metro Group Co. failed to shut down Metro Line 5 and dispatch workers for safety inspections when urban flooding had already formed, and the OCC commanders released it from Haitansi Station without investigating the safety of the surrounding area, instructing the train to move back 30 m without understanding the conditions in the tunnel. This resulted in the train losing power and stopping in the water. The subway company lacked escape drills and had insufficient rescue resources, leading to a delayed report of the dangerous situation and hindered rescue efforts.

5. Discussion

5.1. Resilient City Governance Framework

Table 1. Urban Resilience Governance Factors and Dimensions.

System	Governance Factors	Dimensions of Urban Resilience	Factors Contributing to the Accident
Natural system	Geographical location	Natural resilience	1. Stable and strong rainfall in Henan Province due to its geographical location and air pressure
	Historic disasters		2. Previous extreme flood events in 1975
	Extreme weathers		3. Record-breaking heavy precipitation events due to typhoon
	Urban disasters		4. Extreme rainstorms cause severe urban flooding.
Technical (Equipment) system	Violations in facility construction	Technological resilience	5. Violations in facility construction (e.g., unauthorized changes to design) and inadequate infrastructure (e.g., substandard water retaining fences, insufficient drainage capacity)
	Urban infrastructure
	Transportation stations
	Escape plan design
Man-made system	Disaster warning and emergency knowledge	Institutional resilience	6. Lack of disaster warning and emergency knowledge among citizens
	Emergency response plan		7. Weak emergency response plan and capacity of the government
	Management of rescue organization and resource		8. Lack of coordination between official and unofficial rescue organizations
	Emergency management	Organizational resilience	9. Mistakes in emergency response actions (e.g., failure to shut down subway line in time, inadequate instructions to train operators) 10. Delays in reporting dangerous situations and inadequate rescue resources

reveals the 12 governance factors exhibited in the case incident and corresponds to the deficiencies of Chinese cities in natural resilience, technological resilience, institutional resilience, and organizational resilience. For these 12 governance factors, 3 governance bodies are shown in Figure 7—government, enterprises and residents. Among them, government is the most important governance body, governing 10 factors. Enterprises govern 5 factors, while residents, as a vulnerable group in disasters, are responsible for 1 factor.

Figure 7 illustrates that the natural system corresponds to natural resilience, which is governed by geographic location, historical disasters, extreme weathers, and urban disasters. The technical system corresponds to technological resilience, which is governed by violations in facility construction, urban infrastructure, transportation stations, and escape plan design. The man-made system corresponds to institutional resilience and organizational resilience, where institutional resilience is governed by disaster warning and emergency knowledge, emergency response plan, and management of rescue organization and resource. Organizational resilience is governed by emergency management.

Figure 7 indicates that accidents are often a result of interactions between the natural, technological, and man-made systems, due to governance deficiencies. To improve resilience in sudden weather disasters, there is a need to identify governing body factors, anticipate disasters, and strengthen collaborative governance while reducing defects.

5.2. Resilient City Governance Recommendations

Ribeiro and Gonçalves [15] summarized 11 resilience characteristics in

Table 2. Main characteristics of urban resilience [15].

	Characteristics	Description
1	Redundancy	Existence of several functionally similar components, so that the system does not fail when one of the components fails.
2	Diversity	Existence of several functionally different components to protect the system against the various threats. The more diversity the system possesses, the better the ability to adapt to a wide range of diverse circumstances.
3	Efficiency	Positive relationship between the functioning of a static urban system in relation to the operation of a dynamic system.
4	Robustness	Ability to resists to attacks or other external forces. Robust design anticipates potential system failures, ensuring that failures are predictable, secure and not disproportionate to the cause.
5	Connectivity	Connected system components for support and mutual interaction.
6	Adaptation	Ability to learn from experience and be flexible in the face of change.
7	Resources	Existence of resources that can be rapidly displaced to respond to disruptions and their effects.
8	Independence	Ability to operate for a continuous post-disaster period without relying on external physical intervention.
9	Innovation	Ability to quickly find different ways to achieve goals or meet their needs during a shock, or when a system is under stress. Innovation is critical to developing a city’s ability to restore the functionality of critical systems under severely limited conditions.
10	Inclusion	Development of broad consultation and involvement of communities, particularly of the most vulnerable groups in the development of processes and plans. An inclusive approach contributed to a joint vision to build the city’s resilience.
11	Integration	Integration and alignment between urban systems promotes stronger decision-making and ensures that all users/components mutually support each other for a common outcome. The exchange of information between systems allows them to function collectively and respond quickly through shorter response cycles across the city.

through a review of urban resilience literature, underpinning all resilience dimensions. To promote a resilient urban system, resilience characteristics must be combined to generate reliable, structured, socially robust and implementable scientific knowledge [15]. Therefore, this study will also make relevant recommendations based on the 11 characteristics in terms of natural, technological, organizational, and institutional resilience.

Natural resilience is usually understood as the urban ecosystem [15], such as maintaining the ecosystem elements like water and soil. However, ecosystems have linkage with socioeconomic systems at spatial, temporal, and complexity scales [57], making it difficult for urban system to cope with unexpected meteorological hazards solely rely on ecological resilience governance. Lacking recognition of regional geography, environment conditions and historic hazards can lead city government into a one-size-fits-all model of urban resilience governance dilemma [58]. Extreme weather hazards and new type of urban disasters due to climate change will further put pressure on urban resilience governance. To respond, the government as the leading governance body [22,23], must understand the special geographical location of cities, historic disasters, be alert to regional extreme weather, and study new types of urban hazards. Differentiated urban resilience policies and specific facilities should be developed to provide resources for different cities to deal with varied urban vulnerabilities. Figure 7 also shows that the governance of natural resilience affects technological, institutional, and organizational resilience governance, this is a phenomenon observed from this case event, requiring further research for support.

Technological resilience and organizational resilience are important resilience dimensions of Critical Infrastructure Resilience (CIR) in cities [59,60]. The case event involves the urban metro system, which is closely linked to CIR. The government, Critical Infrastructure (CI) enterprises, and other social resources together form the governance body of CIR. Technical resilience refers mainly to the physical aspects of CI, which need to withstand shocks while minimizing functional losses and recovering quickly from unwanted impacts during disasters [60]. Physically, the green infrastructure implemented in China’s Sponge City program can withstand urban flooding to some extent [37], but in the case of this record-breaking extreme rainstorm in Henan Province, the ability of green infrastructure to mitigate flooding is unclear and depend more on the emergency management and decision making of urban system. The government and organizations should prioritize the inspection of urban infrastructure security risks (drains, underground pipelines, utility poles, etc.), especially during critical disaster preparedness periods, to prevent accidents and enhance the robustness of the city. Enterprises must design escape plans, check escape devices and facilities, increase rescue resources and conduct escape drills regularly to ensure disaster relief efficiency.

Organizational resilience is the ability of organizations that operate and manage CI to withstand shocks, including processes such as capacity, planning, leadership, and training [60]. This case shows that the governance of organizational resilience relies on CI organizations themselves. Urban metro and other infrastructure enterprises need to enhance emergency management before hazards by inspecting risks, developing plans, designing escape routes and allocating emergency resources. For example, the Zhengzhou Metro Group Co. should develop emergency plans and conduct drills to train staff in rescue knowledge and skills. Additionally, cooperation between the branch company and group company should be strengthened to increase integration and efficiency of the urban system.

Institutional resilience encompasses governance and mitigation policies or strategies for urban resilience [15], which is a very complex topic because government resilience policies are closely related to the “who, what, when, where, why” [61]. It is difficult to be all-encompassing in its formulation, and thus will lead to widespread interpretation and criticism. In institutional resilience, this study combines this event with previous urban resilience research, and focuses on disaster warning and emergency knowledge [6,62], emergency response plan [62], which are often mentioned in emergency services policies, and management of rescue organization and resources, to make more specific recommendations.

Governments and residents work together in disaster warning and emergency knowledge aspect. The government is primarily responsible for warning of disasters, while residents are vulnerable. The government should emphasize the pre-disaster prevention and deployment stage to implement timely disaster prevention and mitigation deployment. The disaster warning and emergency knowledge should be released through multiple channels, including community outreach. The community also takes the responsibility to help residents prevent disaster. This joint effort improves urban resilience by increasing diversity, robustness, connectivity, inclusion, and integration.

In terms of emergency response plans, the government, as the main governance body, should actively upgrade emergency plans to include new types of hazards, such as urban infrastructure accidents caused by extreme weather. These types of incidents are becoming a significant threat to citizens but are often neglected by government [63]. The government should prioritize prevention and quick deployment of resources to minimize casualties of such incidents. In addition, the timeliness of responding to emergency plans should be re-emphasized.

In terms of “management of rescue organizations and resources”, it is necessary for the government to strengthen cooperation between official and non-governmental rescue organizations, to enhance cooperation with infrastructure organizations, to increase corporate resources for disaster prevention and relief, and to judge and deal with hazardous situations in a timely manner. Enhance the resources, redundancy, innovation, integration and efficiency that the city system may lack in terms of management of rescue organizations and resources.

In summary, although we have included only the government, enterprises and residents among the governance bodies, the government, as the dominant governance body, needs to work closely with various stakeholders, including residents, enterprises, civil society organizations, and community to develop tailored solutions that address the specific vulnerabilities and risks of each city, forming a collaborative governance.

5.3. Obstacles to Resilient City Implementation

During the process of policy enactment to policy implementation, many obstacles will affect policy implementers of resilience policies, resulting in inconsistencies between the outcome and expectations. Shamsuddin [16] defined the obstacles as Resilience Resistance, and several aspects of resistance were highlighted, including fatigue, complacency, and overconfidence.

Government staff, partner organizations and members of the public can become fatigued by repeated policy discussions and this can slow down decision-making and responses. This was the case with rainstorm preparedness, causing delays in the issuance of new initiatives and directives. To prevent fatigue, varying communication format and involving more stakeholders may be beneficial.

Complacency can occur in policy implementation, especially when it is deemed successful, leaving room for potential problems. This was the case in Zhengzhou when the State Council and provincial government met and deployed response plans, but the Zhengzhou Municipal Government may have become too complacent in their belief that they had prepared for the rainstorm and it would not be an issue. To avoid this, policy implementation should be regularly evaluated and potential risks remain vigilant.

Policy implementers may feel overconfidence in their “effective” disaster response plans. This will lead citizens believe in their disaster response policies without vigilance. For example, the government’s delay in shutting down Metro Line 5 during a rainstorm may have led people to believe they were safe to ride it. To avoid overconfidence, it is important for the government to properly assess and manage their policies, and to communicate clearly with the public about potential risks and precautions.

Several resilience resistances were present in the accident, emphasizing the need for the government to remain vigilant in the policy implementation process to reduce the chances of policy failure. Identifying and addressing resilience resistances can improve policy success and better protect the public from disasters.

6. Conclusions

6.1. Main Conclusions

This paper adds research on the implementation aspects of urban resilience based on actual case studies, and obtains a specific urban resilience governance framework for sudden weather disasters, governance recommendations, and obstacles in the governance process. The conclusions are as follows:

(1)

The cause of the Zhengzhou Metro accident can be attributed to the interaction of natural, technical (equipment), and man-made systems. In the natural system, the accident was triggered by the combination of the city’s special geographical location, extreme weather conditions, and urban flooding. In the technical (equipment) system, the main causes included illegal changes to the railroad design, lack of inspection of facilities, inadequate investigation of security risks at traffic stations, and unreasonable design of escape plans. In the man-made system, the government and enterprises were both involved. On the government side, there were unsound pre-disaster response measures, inadequate disaster warnings, untimely emergency response, an imperfect emergency plan, and rigid rescue resource management and cooperation. On the enterprise side, the main cause was a lack of emergency management, manifesting in insufficient consideration and preparation of emergency supplies before the disaster, lack of direction and dispatch during the disaster, and delay in rescue timing.

(2)

From the perspective of resilient cities, China’s cities exhibit risk governance rigidity in the face of sudden weather disasters and emergencies, and there are deficiencies in natural, technological, institutional, and organizational resilience. The Zhengzhou Metro accident illustrates the fact that China still has a long way to go in the construction of resilient cities. In the process of building resilient cities, China should not adopt a one-size-fits-all approach, but rather should take into account the unique characteristics of each city and harness the collective resources of society to build differentiated, resilient cities. This will require the city government to work closely with various stakeholders, including residents, enterprises, and society organizations, to develop tailored solutions that address the specific vulnerabilities and risks of each city.

(3)

The accident highlighted the vulnerability of urban infrastructure to natural disasters, emphasizing the need for urban systems to prioritize disaster preparedness. To improve city resilience in the face of similar disasters, city governments can refer to the resilient city governance framework and recommendations outlined in this paper. Additionally, policy implementation should take into account the potential for complacency, overconfidence, and fatigue to hinder the effectiveness of resilience efforts.

6.2. Limitations and Prospects

The limitation of this paper is that the results are based on the analysis of a single event, which may not be representative of broader contexts. Climate change is expected to increase the uncertainty and potential hazards facing cities. As urban populations continue to grow, it is important for city governments to develop more comprehensive and in-depth approaches to disaster governance in order to protect residents and enhance the resilience of urban systems. This will require a synthesis of theory and practice, with a focus on exploring urban theories in practice, and exploring practical experiences in the context of theoretical frameworks. By forming a cycle of theory and practice, city governments can improve their understanding of urban disaster governance and develop more effective strategies.