A Bayesian and Analytic Hierarchy Process-Based Multilevel Community Resilience Evaluation Method and Application Study

Abstract

Cities are complex systems influenced by a multitude of factors, encompassing society, economy, culture, and environment. These factors make urban development highly vulnerable to various disturbances. Communities work as the fundamental building blocks of a city and directly impact both its social structure and spatial layout. Moreover, urban planning and policies play a crucial role in shaping the development trajectory of communities and the living environment for residents. This study aims to develop a Bayesian and analytic hierarchy process (BAHP)-based multilevel community resilience evaluation method to assess the ability of the community system to withstand disturbances and recover from them. First, the proposed method establishes a comprehensive assessment index system that can evaluate social and environmental resilience as well as institutional and managerial resilience at multiple levels. This system serves as a quantitative decision-making tool to elucidate the impact of various factors on community resilience. Furthermore, the “relative demand coefficient” (RDC) is proposed to compare different communities’ resilience by using Bayesian inference to determine its most probable value (MPV). To validate the applicability of the proposed method, an empirical study was conducted in the Dafapu community located in the Longgang District of Shenzhen. Meanwhile, a simulated virtual community is employed for comparison with the Dafapu community as an illustrative example showcasing the proposed method’s superior performance after integrating the RDC. The empirical study demonstrates that the proposed BAHP-based method can effectively and quantitatively highlight the recovery capabilities and limitations for different communities in various dimensions while providing a clear direction for enhancing urban community resilience. This research contributes new insights to the theory, provides a practical tool to quantify community resilience, and offers a viable path for the actual enhancement of community resilience.

1. Introduction

Cities are currently confronted with progressively intricate social, economic, cultural, and environmental challenges that may induce different kinds of disasters, such as wind, earthquake, fire, and flood disasters (see Figure 1). Globalization, urbanization, climate change, and other factors are intertwined, which introduce significant demands on the stability and sustainable development of cities. Simultaneously, cities have become the key areas of population aggregation, resource intensity, and risk diversity with the continuous progress of urbanization. In this case, the concept of resilience was introduced into urban research and planning to lead to the emergence of the “resilient city”. Urban resilience refers to the capability of a city to maintain its function and stability when facing various external shocks and changes [1,2]. Numerous studies have proved that improving urban resilience can reduce the risk of disasters, enhance social cohesion, and promote the sustainable development of cities [3,4]. Taking New York as an example, Glaeser puts forward the viewpoint that “resilience” is one of the necessary conditions for a successful city [5]. Moreover, civil structures play a pivotal role in safeguarding resilient cities against natural disasters [6,7]. Consequently, extensive research efforts [8,9,10,11,12,13,14,15,16,17,18,19,20,21] have been dedicated to the field of structural health monitoring and evaluation over the past few decades, aiming to enhance the safety and resilience of civil structures.

The research on urban resilience has garnered significant attention. The evaluation of urban resilience has a wide range of perspectives, so it is very important to establish a resilience research framework suitable for the target city. The Rockefeller Foundation [22] of the United States proposed, in 2015, a resilience evaluation framework for urban systems, including four dimensions of health and well-being, economy and society, infrastructure and ecology, and leadership and strategy, with a total of 12 goals and 52 indexes. The framework mainly evaluates the resilience of cities to potential pressures and future unknown risks, focusing on three areas: regional economic capacity, community population, and community connectivity. In 2009, UNU-EHS [23] put forward a research framework for the resilience of megacities in response to the global urbanization process and the increasing number of megacities. The framework takes into account the opposing attributes of the resilience and brittleness of megacities and includes three main aspects: the global and local impacts of megacities, the official and unofficial systems within cities, and the coupling of social and natural environments. More recently, Hudec et al. conducted a joint analysis in terms of resilience capacity and vulnerability in Slovak urban districts during the 2007–2014 economic crisis by using the Resilience Capacity Index (RCI). Recent research indicates that the enhancement of urban resilience is multi-dimensional, covering infrastructure safety, community response capabilities, land use, and environmental adaptability. These studies highlight the importance of assessing and improving urban resilience and the necessity of adopting comprehensive strategies in the face of natural disasters, socio-economic changes, and environmental challenges [24,25,26,27,28,29,30,31]. By thoroughly exploring the different dimensions of urban resilience, including community recovery after earthquakes [32], the role of nature-based solutions in urban adaptation [33], and the discussion of network resilience in the context of regional integration [34], the research offers a comprehensive perspective on understanding and strengthening the resilience of urban systems. Collectively, these studies emphasize the importance of understanding and implementing urban resilience measures at various levels, providing valuable guidance for urban planning and management.

With the increasing focus on urban resilience research, the concept of community resilience began to fall within people’s focus. As the basic unit of a city, community is an important part of urban resilience. The description of community resilience is an extension of individual resilience to community group resilience, which is the sum of all individual resilience in the community. At the same time, community resilience includes a collection of capabilities and strategies for effectively solving critical incidents and maintaining sustainable community development [35,36,37,38]. On one hand, the improvement of community resilience can better protect the life and property safety of residents in the face of challenges such as natural disasters, environmental changes, and economic fluctuations and improve the ability of post-disaster reconstruction and sustainable development of communities. On the other hand, it can establish a strong community network, form mutual assistance and support among residents, and provide a better living environment and social services. Therefore, it is of great practical significance to study and evaluate the methods and index systems of community resilience. In recent years, scholars in China and abroad have a more comprehensive definition of the community resilience evaluation framework. Rana et al. [39] established a community resilience evaluation index system based on the resilience dimensions of economy, system, social infrastructure, ecology, and residents’ psychology and then applied it to three communities in Pakistan. Houston et al. [40] adopted the views of media and communication (communication ecology, public relations, and strategic communication) to put forward a community resilience evaluation model, which is composed of four parts: communication system and resources, community relations, strategic communication process, and community attributes. Imperiale et al. [41] proposed the SIA (Social Impact Assessment) community action framework and applied it to a community in rural Italy, aiming to serve as an appropriate tool for community resilience development outcomes. Renschler et al. [42] established a comprehensive framework for measuring disaster resilience in communities at various scales. This framework identified seven dimensions, including population, environment, government services, infrastructure, lifestyle, economic development, and social–cultural capital. This framework can help planners assess and enhance community resilience. White et al. [43] discussed the projects and case studies conducted by the Community and Regional Resilience Institute to enhance community resilience. They presented a theoretical framework, reviewed practical work as well as collaborative development processes, and provided insights for other communities. Under the determined resilience framework, scholars have improved the quantitative evaluation method. Ceskavich et al. [44] believed that community resilience depends on the functions of building complexes and infrastructure systems and developed an engineering resilience evaluation method to solve the priority performance gap between the levels of community resilience so as to improve community resilience. Cohen et al. [45] highlighted the importance of community resilience in emergency preparedness and response and the lack of scientifically accepted measurement methods. They developed a tool called CCRAM to assess community resilience and identify its strengths and weaknesses. The plan was to conduct longitudinal studies to evaluate intervention plans for improving community resilience. Pfefferbaum et al. [46] developed a community resilience assessment tool called the CART survey. Through engaging communities in measurement, this tool aims to enhance community resilience to disasters. They underwent multiple field tests and identified four interrelated constructs: Connection and Caring, Resources, Transformative Potential, and Disaster Management. The utilization of the CART survey can offer valuable insights to organizations and communities seeking to evaluate the resilience of their community. Cutter et al. [47] introduced the BRIC, a resilience metric used to assess the inherent resilience of US counties. The research reveals that the resilience levels are highest in counties located in the US Midwest and Great Plains states, whereas counties situated along the US–Mexico border in the west and those along the Appalachian ridge in the east exhibit comparatively lower resilience. Additionally, inherent resilience is distinct from social vulnerability. The BRIC can help policymakers identify intervention strategies to improve resilience. Some scholars apply the community resilience evaluation system to specific backgrounds (such as disaster types, environment, and cities) and improve the resilience evaluation according to local conditions. Alshehri et al. [48] proposed a framework of “community resilience to disasters” based on the Delphi method and AHP method. Starting from this framework, he integrated quantitative and qualitative mixed assessment tools to measure community resilience in Saudi Arabia and other regions. Abenayake et al. [49] introduced a comprehensive environmental index to evaluate the flood resistance capacity of communities and conducted an empirical study in Colombo, Sri Lanka. Fox-Lent et al. [50] introduced a new method for assessing community resilience, which incorporates stakeholder-informed metrics aligned with the temporal stages of disaster resilience. They applied this method to assess coastal community resilience in Rockaway Peninsula, New York, demonstrating its flexibility. Arbon et al. [51,52] explored the creation of a practical toolkit that can be utilized by communities to assess the anticipated level of resilience in response to calamities. The toolkit adopts a comprehensive approach towards all potential hazards and assists local decision-makers in establishing priorities, allocating resources, and formulating emergency and disaster management initiatives aimed at enhancing resilience within the local community.

Due to variations in society, economy, culture, and environment across different countries or regions, the resilience index systems and evaluation methods established by different urban communities generally need to be modified and improved accordingly. This paper aims to study and discuss a new community resilience assessment method and index system for evaluating community resilience using a Chinese community in Shenzhen city as an example, demonstrating its practical application. In order to compare the resilience of communities with different backgrounds more effectively, the “relative demand coefficient” is innovatively introduced and derived through Bayesian inference in this paper. Additionally, the proposed resilience evaluation approach will be applied to both a real community and a simulated virtual community for comparison and further validation of its functionality. The main sections of this article are organized as follows: Section 2 outlines the typical urban resilience definitions, plans, and frameworks related to urban resilience; Section 3 introduces the methodology for the Bayesian and analytic hierarchy process based on multilevel community resilience evaluation; Section 4 presents case studies along with application demonstrations by using both a real community and a simulated virtual community. Finally, concluding remarks are provided in Section 5.

2. Typical Urban Resilience Definition, Plan and Framework

2.1. Resilience Definition

Resilience is demonstrated through the ability to withstand sustained pressure. In a broader context, resilience refers to the capacity exhibited by an affected entity in response to changes in the external environment. This capacity can be inherent or developed through optimization, encompassing various dimensions such as ecology, population, economy, and society. From a disaster management perspective, resilience entails the innate capability to adapt non-core elements in order to adjust to shocks or pressures and ensure survival in new environments. During the process of disaster response, resilience is manifested by transforming the negative impact of disasters into positive factors, thereby enhancing future disaster response capabilities. The concept of resilience, closely associated with urban resilience, was initially introduced into ecological studies by HOLLING in 1973 [53], describing ecosystems’ ability to recover and stabilize after disturbances. As depicted in Figure 2, researchers from different fields have subsequently supplemented and explained the concept of resilience separately.

2.2. Urban Resilience Plan

In developed countries, the concept of building resilient cities has been proposed earlier, and there are a large number of research results and a series of completed and ongoing resilient city construction plans. Many countries, including the United States, the United Kingdom, and Australia, have developed development policies based on the theory of resilience worldwide. However, most cities have developed resilience policies at the strategic planning level, lacking implementable action plans for major disasters. A notable shared aspect across these plans is the focus on enhancing the overall protective capacity of cities against impending climate risks, aiming to create cities that are secure, resilient, and conducive to quality living. Several representative resilient city construction plans from abroad [67] are summarized in

Table 1. Urban resilience planning for the seven most representative cities worldwide.

City	Plan Name	Risks	Published Time	Content
Chicago, USA	Climate Action Climate Plan	Hot summer, dense fog, flood, and rainstorm	2008.9	Goal: To establish a model of a harmonious living environment in large cities. Key areas: Establishing green buildings for stagnant rainwater, flood management, tree planting, and green roof projects.
Rotterdam, Netherlands	Rotterdam Climate Proof	Floods and rising sea levels	2008.12	Goal: To enhance environmental resilience, infrastructure resilience, and social resilience. Key areas: Green increment, roof greening, solar power generation, urban planning based on water, residential renewal plans, and equipping affordable housing.
Quito City, Ecuador	Quito Climate Change Strategy	Mudslides, floods, droughts, and glacier retreat	2009.10	Goal: To develop comprehensive policies that ensure the implementation of adequate, crosscutting, and equitable adaptation and mitigation measures to climate change. Key areas: Ecosystems and biodiversity, drinking water supply, public health, infrastructure and electricity production, climate risk management.
Durban, South Africa	Climate Change Adaptation Planning: For A Resilient City	Floods, rising sea levels, coastal erosion, etc.	2010.11	Goal: To become the most caring and livable city in Africa by 2020. Key areas: Water resources, health, and disaster management.
London, UK	Managing Risks and Increasing Resilience	Continuous floods, droughts, and extreme high temperatures	2011.10	Goal: To construct the “London Climate Change Public Private Partnership Mechanism” and introduce the “UK Climate Impact Plan”. Key areas: Flood risk management, increase in parks and greenery, poverty alleviation, climate protection standards, climate risk information, and resilience assessment guidelines.
New York, USA	A Stronger, More Resilience New York	Floods and storm surges	2013.6	Goal: To combine social, climate, organizational, and infrastructure resilience. Key areas: Poverty alleviation, hardening engineering with green ecology, climate protection standards, climate risk information, resilience assessment guidelines, transforming electricity, roads, water supply, and drainage.
Japan	Outline of Land Strengthening and Resilience Policy	Earthquakes and tsunamis	2013.12	Goal: To protect the people, maintain the functioning of the country and society, minimize the loss of national property and public facilities, and rapidly revitalize the economy. Key areas: Risk communication, human resource development, infrastructure upgrades, emergency preparedness, database and open data promotion, and community engagement.

.

In 2010, the United Nations initiated the global “Make Cities Resilient-My Cities Ready” campaign, delineating the roles of government bodies and outlining actionable strategies for authorities to bolster urban resilience. By 2014, at the Urban Resilience Summit, 35 cities were identified as models of resilience and were furnished with financial and technological assistance. Furthering these efforts, in 2015, the Earthquake Mitigation Initiative (EMI) crafted a comprehensive resilience blueprint tailored for developing countries. The evolution of resilient urban planning is progressively moving from broad overarching strategies to detailed, localized disaster preparedness measures, signifying a shift from macro-level planning to micro-level implementations.

2.3. Urban Resilience Framework

Urban resilience research encompasses a wide range of significant directions and methodologies, with six major frameworks (see

Table 2. Overview of key urban resilience frameworks.

Framework	Focus	Strength	Characteristics	Differences
Resilience Alliance: Urban resilience framework (early 21st century) [68]	Comprehensive resilience across a range of hazards	Holistic methodology considering various aspects of urban resilience	Collaborative network fostering knowledge exchange	Unique collaborative and network-based approach
UNU-EHS: The megacity resilience framework (2009) [23]	Linking environmental sustainability with human security	Emphasizes socio-ecological systems	Research-oriented with a theoretical foundation	Emphasis on human security and environmental sustainability
Rockefeller Foundation: City resilience framework (2014) [69]	Enhancing urban resilience against various shocks and stresses	Comprehensive support for global cities	Advocates for cross-sectoral collaboration and holistic urban systems	Emphasizes a global network and resource-sharing
EMI: Urban resilience master planning (2015) [70]	Earthquake mitigation and emergency preparedness	Specialized in addressing seismic risks	Action-oriented approach with a focus on response and recovery	A narrower focus on seismic activities
Japanese framework: Disaster Prevention 4.0 (2016) [71]	Advanced disaster management using IT and big data for efficient response	Real-time response capabilities and knowledge transfer	Focuses on community involvement and infrastructure resilience	Emphasizes technological innovation in disaster prediction and analysis
UNISDR: Trends in urban resilience (2017) [72]	Disaster risk reduction in urban planning	Aligns with international disaster risk reduction strategies	Policy and governance-oriented guidelines	Alignment with global disaster risk policies

) emerging as particularly representative: the urban resilience framework developed by the Resilience Alliance, the UNU-EHS’ resilience framework, the Japanese framework for urban and energy systems resilience, the Rockefeller Foundation’s city resilience framework, the EMI’s urban resilience master planning, and the UNISDR’s urban resilience framework. Each framework focuses on specific areas, ranging from earthquake mitigation to comprehensive considerations of environmental and human security and specialized studies on urban infrastructure and energy systems. By examining these frameworks, we can gain valuable insights into their guidance for cities in responding to and recovering from natural disasters, socio-economic changes, and environmental challenges. This offers significant perspectives for urban planning and management. The summarization and comparison of the aforementioned six frameworks in Table 2 not only elucidate the unique contributions of each framework but also facilitate interdisciplinary academic discourse and practical application, thereby further enhancing overall urban resilience. Moreover, the characteristics of these frameworks will be separately introduced in the following paragraphs.

To effectively evaluate and scientifically quantify urban resilience, various research institutions have established many framework systems for studying resilient cities from their respective fields. Since the establishment of the Resilience Alliance in the early 21st century, its urban resilience research framework has gradually become an important reference in the domains of urban planning and disaster management. This framework, adopting a multidisciplinary and multi-scale perspective, comprehensively considers aspects such as urban planning, infrastructure design, environmental management, community involvement, and governance, intending to enhance the city’s response and recovery capabilities against a variety of internal and external pressures, including climate change, economic fluctuations, social inequalities, and natural disasters. At its core lies an emphasis on adaptive management approaches that foster continuous learning and innovation while promoting participatory governance. The framework recognizes cities as complex networks of interactions between humans, nature, and built environments within socio-ecological systems. It advocates for flexible and dynamic management strategies that adapt after experiences while exploring novel solutions through collaborative engagement with a wide range of stakeholders to collectively build resilience. The practical application of this framework in cities worldwide has demonstrated its effectiveness by assisting these cities in improving their preparedness and response capabilities to various challenges while providing robust guidance and support for constructing sustainable and resilient urban environments.

The megacity resilience framework published by the United Nations University Institute for Environment and Human Security (UNU-EHS) aims to integrate the objectives of environmental protection and human security to promote sustainable development and enhance resilience in urban areas. Recognizing cities as hubs of economic and social activities, as well as frontlines facing multiple pressures, including natural disasters, climate change impacts, and social challenges, the framework emphasizes the crucial connection between environmental health and societal well-being. It proposes a range of interdisciplinary strategies, including ecosystem services protection and restoration, sustainable infrastructure design, disaster risk reduction, and community engagement. These strategies are designed to strengthen cities’ capacities to respond to both immediate and long-term challenges while ensuring the safety and well-being of all urban residents. In practice, the framework has guided planning processes and policy-making in various cities by implementing forward-thinking solutions that are inclusive in nature. It offers a comprehensive model for global urban resilience that encompasses both environmental sustainability and human security.

The “Urban Resilience Master Planning” was launched in 2015 by the Earthquake Mitigation Initiative (EMI), with a focus on developing countries. Its objective is to enhance its capabilities in earthquake response and recovery through a multidisciplinary and holistic approach. The framework emphasizes a comprehensive methodology that encompasses urban planning, infrastructure design, environmental management, community involvement, and governance, among other aspects. Its primary goal is to mitigate the potential impacts of earthquakes and related disasters, such as tsunamis and landslides, through preventative and preparatory measures while ensuring rapid and effective emergency responses and recovery processes. Core components of the framework include urban risk assessment, seismic planning and design, governance and policy development, community engagement, as well as emergency preparedness and response strategies, all aimed at creating safer, more sustainable, and resilient urban environments. This framework has been successfully applied in various urban planning and disaster risk management projects in developing countries, proving effective in enhancing these cities’ preparedness and response capabilities to earthquakes and other natural disasters.

In June 2016, the Japanese Cabinet released a proposal titled “Disaster Prevention 4.0-vision for a future disaster prevention system” by experts in disaster prevention after a long-term exploration. Following the devastating Typhoon Ise Bay in 1959, the Japanese government recognized that the absence of a unified disaster prevention system was one of the key factors contributing to significant losses. Consequently, over the next three years, they established “Disaster Prevention 1.0”, which served as Japan’s fundamental disaster prevention system. This system focused on government-led initiatives and planning until it was exposed by the shortcomings revealed during the 1995 Hanshin earthquake. To address these deficiencies, Japan introduced “Disaster Prevention 2.0”. This updated version mandated building earthquake-resistant structures and implemented new autonomous disaster prevention organizations and emergency disaster response headquarters to strengthen the emergency response mechanism. The subsequent events of the 2011 East Japan earthquake and Fukushima nuclear accident further tested Japan’s disaster prevention system, leading to its evolution into “Disaster Prevention 3.0”. This iteration incorporation considerations for handling unexpected disasters and regulating nuclear energy functions. The government actively sought input from experts and scholars while integrating measures for reducing disasters into their existing framework through revisions in relevant regulations. Now entering a new phase of management with “Disaster Prevention 4.0”, Japan aims to enhance speed, efficiency, and effectiveness in preventing, responding to, and recovering from disasters by leveraging information technology advancements alongside big data analysis capabilities. This concept emphasizes the importance of multi-party cooperation, including the collaborative involvement of government, enterprises, and academic institutions. By analyzing historical data and conducting regular emergency drills, such as annual training sessions, preventive measures, and emergency preparedness, have been strengthened. The core elements include community participation and enhancing the disaster prevention function of urban infrastructure. Additionally, the real-time sharing of disaster information and the application of innovative technologies like artificial intelligence have enhanced the efficiency of disaster management. The ultimate goal is to achieve global integration in disaster prevention management and collectively enhance global capabilities for disaster prevention and emergency response. The proposal for “Disaster Prevention 4.0” aims to fully instill risk awareness while rebuilding connections and network systems among various entities to improve overall societal resilience.

The “Trends in urban resilience”, published by the United Nations Disaster Risk Reduction (UNISDR), is a comprehensive guide designed to systematically identify, assess, and mitigate disaster risks in urban areas to enhance resilience. This framework emphasizes the importance of integrating disaster risk management into urban planning and development, promoting collaboration among policymakers, urban planners, and community stakeholders. By implementing risk assessments, disaster prevention strategies, emergency preparedness plans, and effective recovery mechanisms, this framework aims to reduce the potential impacts of natural and human-made disasters while ensuring sustainable urban development. The UNISDR framework particularly highlights building local capacities, raising public awareness and education, and encouraging information sharing and collaboration across sectors. Through the global promotion of this framework, UNISDR has empowered cities to develop more resilient socio-economic and environmental systems that enable them to effectively face future challenges.

From the aforementioned content, it is evident that the majority of existing studies have chosen cities as their research objects, encompassing social, economic, material, spatial, population, and other constituent elements. The investigation into the framework of resilient cities not only necessitates comprehensive coverage of all aspects pertaining to urban resilience but also entails an analysis of the role and impact of resilience within urban systems. The extant conceptual framework offers guidance for interdisciplinary research on resilient cities by delineating the interplay between disasters and resilience while establishing a robust foundation for future investigations in the realm of disaster prevention and mitigation.

2.4. Policy Development Process of Community Resilience in China

The research on the theory of resilient cities in China started relatively late. In recent years, scholars in relevant fields have conducted in-depth analyses of the concept of resilient cities by studying the existing renowned urban resilience frameworks. Although research on the theoretical framework, research methods, evaluation system, and evaluation methods of resilient cities is still at an early stage, some scholars have provided excellent insights into resilient urban planning, resilient underground spaces, and earthquake-resistant resilient cities.

In 2017, the Beijing Municipal Institute of City Planning and Design, in collaboration with Tsinghua University, the Institute of Geographic Sciences and Natural Resources Research of the Chinese Academy of Sciences, and Atlas Company, initiated a comprehensive research endeavor focused on resilient urban planning. This initiative was grounded in the fundamental principles of urban resilience, emphasizing thorough risk assessment and evaluation as its core. Through this approach, a robust theoretical and technical framework for resilient urban planning was established, outlining specific goals, strategies, and implementation pathways for enhancing Beijing’s resilience. The research particularly delved into thematic studies of flood and health risks to provide concrete decision-making support and planning guidelines for constructing a resilient city in Beijing. It innovatively integrated spatial epidemiology, geography, and urban–rural planning to explore proactive methods and approaches in urban and rural planning to influence public health. Based on the analysis of the spatiotemporal distribution characteristics of typical chronic and infectious diseases in Beijing, the research examined the spatial distribution of vulnerable populations and their major influencing factors, proposing resilience enhancement strategies in the public health domain.

In November 2022, Academician Xie Lili, a member of the Chinese Academy of Engineering (CAE), delivered an academic presentation at the International Forum on Resilience of Critical Infrastructure Systems. He pointed out that constructing resilient cities is not solely an engineering problem but rather a complex scientific problem that spans multiple fields and disciplines. He advocated for the development of a new discipline to integrate diverse resilience needs and scientifically analyze the constituent elements and their impact on urban resilience. Taking “earthquake-resistant and resilient cities” as an example, he highlighted the importance for cities to possess self-reliance in withstanding earthquakes and restoring functionality post-disasters. Xie Lili analyzes the seismic resilience of a city from six dimensions, including earthquake analysis and prediction capabilities, building and infrastructure levels, socio-economic conditions, disaster management capabilities, emergency response, and related non-engineering factors. He underscored that the construction of resilient cities can be graded based on their current status while drawing inspiration from biological self-healing abilities to establish a city’s “immune system”. Xie Lili foresaw the future emergence of more resilient urban and rural areas, communities, etc., emphasizing the “resilience path” to enhance overall disaster resistance capacity within systems. He believed that constructing resilient cities is an intricate yet promising systematic endeavor requiring a careful balance between investment costs, benefits gained, and economic development.

In June 2023, Academician Chen Xiangsheng, a member of CAE, delivered a keynote speech at the China–Japan Engineering Academician Forum titled “Key Scientific Issues on the Resilience of Deep Underground Spaces in Megacities”. In his research on resilient cities, Chen Xiangsheng emphasizes the pivotal role of underground space development in mitigating urban carbon emissions and enhancing urban resilience. He advocates for relocating facilities and infrastructure that impact the landscape underground to decrease carbon emissions and achieve negative carbon emissions through surface greening. Furthermore, Chen explores the multidimensional concept of urban safety, encompassing infrastructure, ecology, water systems, and social scenarios, while highlighting the significance of comprehensive perception and intelligent management in constructing resilient cities. He points out that improving the human environment of underground spaces and achieving the integrated planning of surface and underground spaces are current challenges. Chen’s concept of “resilient underground space” underscores the importance of safeguarding critical lifelines during disasters while advocating for design principles that consider cultural heritage, regional characteristics, and rapid recovery capabilities. This provides significant guidance for building sustainable and resilient cities.

The community serves as not only the fundamental unit of social management but also the forefront of urban disaster prevention and mitigation. The development of community environments and the provision of facilities are directly related to the quality of daily life and assurance of their safety. Additionally, communities bear the responsibility of initiating immediate disaster relief efforts following a calamity. Simultaneously, as the most grassroots level of social organization, communities play a pivotal role in the mobilization and dissemination of information during the process of disaster prevention. Commencing disaster prevention and reduction from the community level, using the community as the basic unit to promote these efforts, fully harnessing diverse community-based disaster prevention resources, formulating and implementing community-specific plane for disaster prevention, and enhancing the comprehensive capabilities for both preventing and mitigating disaster within communities have become issues that garner widespread attention domestically and internationally.

As shown in Figure 3, many policies for community disaster prevention and reduction have been conducted in mainland China. In January 2006, China officially promulgated the “National General Emergency Response Plan for Public Events”, which clarified the need to strengthen “public emergency capability building based on towns and communities, and to leverage their important role in responding to sudden public events”. In August 2007, the General Office of the State Council issued the “National Comprehensive Disaster Reduction ‘11th Five-Year’ Plan”, proposing to strengthen the disaster reduction capacity building of urban and rural communities, promote grassroots disaster reduction work, and carry out comprehensive disaster reduction demonstration community activities. In May 2010, the office of the National Disaster Reduction Commission issued the “National Standards for Comprehensive Disaster Reduction Demonstration Communities”, encouraging localities to actively create comprehensive disaster reduction demonstration communities, continuously improve community disaster prevention and mitigation capabilities and emergency management levels, enhance urban and rural community residents’ awareness of disaster prevention and mitigation and their ability to evade disasters and self-rescue, and effectively safeguard the safety of people’s lives and property, thus promoting the construction of a harmonious socialist society. In March 2021, Shenzhen issued the “Shenzhen Comprehensive Disaster Reduction Community Creation Implementation Plan (2021–2023)”, which set higher requirements based on the “National Comprehensive Disaster Reduction Demonstration Community” and planned to achieve 100% coverage of “Comprehensive Disaster Reduction Communities” across the city by 2023. As of 1 September 2022, the Shenzhen local standard “Comprehensive Disaster Reduction Community Creation Guide” has been officially implemented. The “Guide” specifies 13 requirements for the creation of comprehensive disaster reduction communities, guiding the work of community creation. Recently, the implementation plan for constructing a safe and resilient city was launched by Shenzhen in 2024, with the objective of establishing an internationally recognized and pioneering benchmark city for safety and resilience. This initiative aims to significantly enhance cities’ capacity to withstand catastrophic events and swiftly recover from disturbances.

Currently, China has established 6397 national comprehensive disaster reduction demonstration communities, designated 13 pilot units as the initial batch of national comprehensive disaster reduction demonstration counties, constructed 12 national-level fire safety science education museums, and systematically advanced the construction of a public platform for disaster prevention and mitigation science education network. The general public’s awareness of disaster prevention and mitigation, as well as their proficiency in self-rescue and mutual aid, have witnessed significant enhancement.

3. Methodology of Multilevel Community Resilience Evaluation

3.1. Definition and Grading of the Multilevel Resilience Index System

Community resilience reflects the ability of a city to maintain or quickly restore its characteristics and operating modes to the pre-disaster state in the event of a disaster, relying on its functions. In other words, community resilience is the response and adaptability of the urban system to uncertain factors, which can afford, adapt, and restore rapidly in the inverted environment. As depicted in Figure 4, illustrating changes in community resilience following disasters, this ability serves as a model for stable urban community development and acts as a precautionary measure for communities to effectively cope with major risk incidents such as natural disasters, production safety issues, and public health emergencies. Currently, communities only exhibit partial aspects of resilience, and continuous improvement of overall community resilience is crucial. To comprehensively enhance the resilience of existing communities, it is imperative firstly to establish an evaluation system based on a community resilience index that reflects the capacity of the community system to withstand multiple disturbances while also considering diverse absorptive capabilities and adaptability, which dynamically change alongside community construction indexes. Constructing a comprehensive index system for evaluating community resilience holds significant importance in understanding this concept better, identifying current deficiencies in terms of community toughness, and subsequently improving resilience accordingly.

The urban community resilience index should align with the characteristics of goal orientation and characteristics of the city attributes while reflecting the ability of the urban system’s capacity to withstand multiple disturbances and diversified absorptions. To effectively represent urban community resilience, the selection and establishment of the index system should adhere to certain principles (refer to Figure 5):

(1)

Independence principle: Each indicator of the index system possesses unique characteristics and fully explicates its connotation. This implies that each indicator does not duplicate others or share content, thereby enabling a comprehensive reflection of the actual state of the corresponding index classification from diverse perspectives and levels.

(2)

Representativeness principle: The selected city evaluation indicators should more comprehensively reflect or represent the overall resilience level of a city within a specific period and a specific environment while effectively capturing the distinctive features of the city.

(3)

Systematicness principle: Evaluating urban community resilience is a holistic and systematic task, encompassing multiple dimensions such as society, economy, system, urban infrastructure, and ecological environment. Indexes at all levels embody interactions and coordination among various aspects involved in urban development, thus necessitating careful planning and consideration.

(4)

Feasibility principle: The selection and hierarchical definition of urban resilience indexes should consider the actual acquisition of relevant information and data that are quantifiable, easy to calculate/process, and sourced from credible sources to ensure authenticity and effectiveness in evaluating urban resilience.

(5)

Scientificity principle: The selection and design of urban resilience indicators should be based on scientific references that accurately reflect the level of urban resilience objectively while identifying the weak links in the process of urban construction, thereby providing scientific guidance for the construction of resilient city development.

Figure 5. Principles of index system construction for community resilience evaluation.

Figure 5. Principles of index system construction for community resilience evaluation.

The applicability of the resilience index system depends on the scale and characteristics of the evaluation region, encompassing various levels such as counties, cities, towns, and communities. Indicators at each level should possess sensitivity and representativeness towards the concerned region being evaluated. The community serves as the smallest cell of urban management, acting as both a home for its residents and playing a crucial role in disaster prevention and mitigation efforts. Consequently, the composition of the community resilience index system differs significantly from that at the municipal level. After carefully screening, modification, and grading, boundary adjustment of the index items listed in the resilience index database is conducted based on the conceptual framework of community resilience. This paper adopts a hybrid assessment method combining qualitative and quantitative approaches to initiate a dynamic assessment model for community resilience. A multilevel resilience assessment index system designed for communities’ evaluation is conducted herein, as illustrated in Figure 6. The comprehensive index evaluation method is employed to assess indicators at each level. In qualitative evaluations, evaluators determine indicator values according to the average levels derived from relevant scenario analyses. In quantitative evaluations, target values for urban community resilience indices are comprehensively considered while obtaining indicator scores through weighted statistical analysis. The proposed index system simultaneously considers the guidelines for safety resilience evaluation issued by China, the practical situation, the feasibility of the evaluation indicators, and data availability. Through field research and the Delphi method (expert empowerment method and feedback anonymous letter inquiry method), the proposed index items listed in

Table 3. Multilevel evaluation index system for community resilience.

Primary Index	Secondary Index	Tertiary Index
Social and Environmental Resilience (A1)	Building Construction (B1)	Building fortification standard (ratio of pre-1990 to post-2002 floor space) (C1); proportion of self-built houses (C2); proportion of container-type mobile houses (C3)
	Population Characteristics (B2)	Gini coefficient (C4); employment rate (C5); average years of schooling (C6); percentage of elderly population (C7); percentage of underage population (C8); percentage of the population with physical or mental disabilities (C9); gender ratio (C10); percentage of floating population (C11); incidence rate of criminal cases per 100 people (C12); number of food safety incidents (C13); Number of rehabilitation treatment points for disabled individuals in the community (C14)
	Infrastructure (B3)	Per capita emergency shelter area (C15); Internet penetration rate (C16); per capita daily water consumption (C17); per capita daily electricity consumption (C18); natural gas penetration rate(C19); level of inspection, monitoring, and maintenance of infrastructure (C20); urban and rural sewage collection rate (C21); garbage sorting and recycling capacity (C22); number of hazardous chemical enterprises (C23); number of beds in community medical and health institutions (C24); capacity of community nursing institutions (C25); number of rehabilitation centers for persons with disabilities in the community (C26)
	Transportation Capacity (B4)	Number of transportation hubs (C27); traffic mileage of urban rail and main road (C28);
	Ecological Environment (B5)	Percentage of forest cover (C29); per capita public green space (C30); percentage of undeveloped land (C31); air quality compliance rate (C32); water quality compliance rate (C33); acid Rain Frequency (C34)
Institutional and Managerial Resilience (A2)	Emergency Management Capacity (B6)	Leadership level (C35); compilation of comprehensive disaster prevention and reduction plans (C36); preparation of special plans (C37); situation of emergency drills (C38); percentage of emergency service personnel (emergency management personnel, firefighter, policeman, expert teams, rescue teams) in the employed population (C39); number of volunteers (C40); goods and materials reserves for emergency as well prevention of drought, flood and wind (C41); construction of comprehensive disaster reduction demonstration communities (C42); number of science popularization demonstration schools for earthquake prevention and disaster reduction (C43)
	Economic Resilience (B7)	Public financial strength (C44); ratio of emergency funds to fiscal expenditures (C45); commercial insurance coverage rate (C46); Social Insurance coverage rate (C47)
	Disaster Risk Assessment (B8)	Completion of the natural disaster risk general survey (C48); compilation of disaster risk maps (C49); hidden danger investigation (C50)
	Disaster Monitoring and Early Warning (B9)	Video surveillance rate (C51); coverage of disaster monitoring, early warning, and emergency command information platform (C52); epidemic monitoring and control situation (C53)

have been carefully screened, modified, and graded to ensure independence, representativeness, systematicness, and feasibility. This evaluation index system will be applied to Shenzhen’s community as a demonstration consisting of two primary indexes (social and environmental resilience, institutional and managerial resilience) along with nine secondary indexes corresponding to the first-level indicators. Each secondary index comprises some fine-sorted tertiary indexes, resulting in a total of 53 tertiary indexes for detailed community resilience evaluation (refer to Table 3).

The present study combines two dimensions in terms of social and environmental resilience as well as institutional and managerial resilience to analyze community resilience. Subsequently, a BAHP-based multilevel evaluation approach is adopted for assessing community resilience. The calculation of the resilience index score is based on its definition (refer to

Table 4. Grading and definition of all tertiary indexes corresponding to social and environmental resilience.

Tertiary Index	Index Score (Score the Index According to the Index Interpretation, and Divide the Score into 1 to 5 Points)	Index Interpretation
C1	5--≤0.5 4-->0.5, ≤1.0 3-->1.0, ≤1.5 2-->1.5, ≤2.0 1-->2.0, ≤2.5	According to the statistics of construction years, the proportion of floor space from 1990 to 2001 and the floor space after 2002, considering the difference between the 1989 code and the 2001 code of seismic design of buildings and the application of new seismic technology.
C2	5--≤5% 4-->5%, ≤10% 3-->10%, ≤15% 2-->15%, ≤20% 1-->20%	The proportion of houses with limited property rights and farmers’ self-built houses in the total number of houses.
C3	5-->30%, ≤100% 4-->25%, ≤30% 3-->15%, ≤25% 2-->5%, ≤15% 1--≤5%	The proportion of the number of houses made of special container boards and processes in the total number of houses in the evaluation area.
C4	5--Reasonable. 4--Relatively reasonable. 3--Moderately reasonable. 2--Significant disparity. 1--Stark disparity.	The statistical data obtained (reflecting the income difference of national or regional residents) 0.3–0.4 is relatively reasonable, 0.4 and 0.5 is a significant disparity, and above 0.5 is a stark disparity, and the development trend is judged in turn.
C5	5-->95%, ≤100% 4-->90%, ≤95% 3-->85%, ≤90% 2-->80%, ≤85% 1--≤80%	The percentage of employed persons in the sum of employed persons and unemployed persons reflects the employment degree of the labor force.
C6	5-->16, ≤27 4-->12, ≤16 3-->9, ≤12 2-->6, ≤9 1--≤6	The average of the total number of years of academic education (including adult academic education, excluding various academic training) received by a certain population group in a certain period and a certain region. General classification criteria: more than 16 years of junior college, 12 years of senior high school, 9 years of junior high school, 6 years of primary school, and 0 years of illiteracy.
C7	5--≤15% 4-->15%, ≤25% 3-->25%, ≤40% 2-->40%, ≤50% 1-->50%, ≤100%	The proportion of the population aged 60 and above in the total population of the evaluation area is an index of the aging process.
C8	5-->25% 4-->20%, ≤25% 3-->15%, ≤20% 2-->10%, ≤15% 1--≤10%	Percentage of the population under the age of 18 in the total population in the evaluation area.
C9	5--≤5% 4-->5%, ≤15% 3-->15%, ≤25% 2-->25%, ≤30% 1-->30%, ≤100%	Refers to the percentage of the total population in the evaluation area who have lost a certain organization or function or perform abnormally in a certain organization or function and totally or partially lost the ability to engage in certain activities in a normal way in terms of psychology, physiology, and human structure.
C10	5-->100, ≤108 4-->108, ≤110 3-->110, ≤112 2-->112, ≤114 1-->114, ≤116	The gender ratio is a ratio of the number of men and women in society, basically calculated based on the number of men per 100 women. The value between 108 and 110 is normal.
C11	5--≤10% 4-->10%, ≤20% 3-->20%, ≤30% 2-->30%, ≤40% 1-->40%, ≤100%	The proportion of the floating population in the total population of the evaluation area refers to the population who engage in various economic activities such as working, doing business, and providing social services outside the permanent residence without changing their original registered permanent residence.
C12	5--≤10% 4-->10%, ≤15% 3-->15%, ≤20% 2-->25%, ≤30% 1-->30%, ≤100%	The ratio of the number of criminal cases in a certain period in the evaluation area to the population in the same period is expressed as a percentage.
C13	5--≤10 4-->10, ≤15 3-->15, ≤20 2-->25, ≤30 1-->30, ≤35	Statistics on the number of accidents harmful to human health caused by food problems like food-borne and food pollution.
C14	5-->80%, ≤100% 4-->70%, ≤80% 3-->60%, ≤70% 2-->50%, ≤60% 1--≤50%	The ratio of the number of participants in disaster emergency training to the total number of volunteers in the evaluation area.
C15	5-->1.5, ≤1.8 4-->1.2, ≤1.5 3-->0.9, ≤1.2 2-->0.5, ≤0.8 1--≤0.5	The per capita emergency shelter area in the evaluation area is calculated by regarding 70% of the permanent population as the number of refugees.
C16	5-->80%, ≤100% 4-->60%, ≤80% 3-->40%, ≤60% 2-->20%, ≤40% 1--≤20%	Expressed by the number of broadband access ports, the number of Internet users, and the penetration rate of mobile phones.
C17	5-->400, ≤500 4-->300, ≤400 3-->200, ≤300 2-->100, ≤200 1--≤100	The ratio of the total annual water supply to total population (unit: liters).
C18	5-->10,000, ≤15,000 4-->5000, ≤10,000 3-->2000, ≤5000 2-->0, ≤2000 1:0	The ratio of the total annual electricity supply to the total population (unit: KWH).
C19	5-->90%, ≤100% 4-->80%, ≤90% 3-->70%, ≤80% 2-->60%, ≤70% 1--≤60%	The proportion of the total population using natural gas in the total population in the evaluation area.
C20	5-->100, ≤120 4-->80, ≤100 3-->60, ≤80 2-->40, ≤60 1--≤40	The annual urban infrastructure maintenance and construction funds of the community (unit: 10,000 CNY).
C21	5-->95%, ≤100% 4-->90%, ≤95% 3-->85%, ≤90% 2-->80%, ≤85% 1--≤80%	The proportion of the treated domestic sewage and industrial wastewater in the total sewage discharge in the evaluation area.
C22	5-->30%, ≤100% 4-->25%, ≤30% 3-->15%, ≤25% 2-->5%, ≤15% 1--≤5%	Statistics of urban waste classification and recycling capacity.
C23	5--≤50 4-->50, ≤200 3-->200, ≤350 2-->350, ≤500 1-->500, ≤1000	The number of product manufacturing enterprises engaged in the production, storage, use, operation, and transportation of products involving physical and chemical hazards, health hazards, and environmental hazards of the nature of poisoning, corrosion, and combustion.
C24	5-->11, ≤15 4-->8, ≤11 3-->5, ≤8 2-->3, ≤5 1--≤3	Number of beds in community health institutions
C25	5-->200, ≤250 4-->150, ≤200 3-->100, ≤150 2-->50, ≤100 1--≤50	The total number of people accommodated by community nursing institutions (unit: person).
C26	5--4 4--3 3--2 2--1 1--0	The number of rehabilitation centers for persons with disabilities in the community.
C27	5--5 4--4 3--3 2--2 1--1	Statistics on the number of bus stations, subway stations, docks, and high-speed rail stations.
C28	5-->20, ≤25 4-->15, ≤20 3-->10, ≤15 2-->5, ≤10 1--≤5	The statistics of the total traffic mileage of urban rail and main road (unit: km).
C29	5-->80%, ≤100% 4-->60%, ≤80% 3-->40%, ≤60% 2-->20%, ≤40% 1--≤20%	The ratio of forest area to the total land area of the evaluation area reflects the actual level of forest resources and forest land occupation in the evaluation area.
C30	5-->20, ≤25 4-->16, ≤20 3-->12, ≤16 2-->8, ≤12 1--≤8	An important index reflecting the living environment and life quality of urban residents is the per capita public green space, including parks, amusement parks, street green squares, botanical gardens, zoos, and special parks (unit: square meter)
C31	5-->50%, ≤100% 4-->40%, ≤50% 3-->20%, ≤40% 2-->0%, ≤20% 1:0	Percentage of undeveloped land
C32	5--≥90%, <100% 4--≥80%, <90% 3--≥70%, <80% 2--≥60%, <70% 1--<60%	The proportion of days with an air pollution index (API) less than or equal to 100 in the total number of days is reflected by the excellent rate of ambient air quality.
C33	5-->90%, ≤100% 4-->70%, ≤90% 3-->40%, ≤70% 2-->20%, ≤40% 1--≤20%	The quality of drinking water reflects the physical, chemical, and biological characteristics and composition of the water body.
C34	5--≤10% 4-->10%, ≤20% 3-->20%, ≤40% 2-->40%, ≤50% 1-->50%, ≤100%	Acid rain is acidic wet precipitation, which is rain, snow, or other forms of precipitation with a pH of less than 5.6, mainly caused by man-made emissions of large amounts of acidic substances into the atmosphere.

), wherein statistical data and expert feedback information are utilized step by step to assign scores. The weighted average method is then applied to calculate and obtain the score value by multiplying each indicator’s score at a given level with its corresponding weight coefficient before summing them up for the subsequent level of indicators. Finally, similar procedures are followed to determine the primary index score as well as their total scores. The weight value of each indicator is determined through a comparison of the importance of the indicator using the Delphi method. The weighted average calculation equation for obtaining the comprehensive index score can be expressed.

$$S={\displaystyle {\sum }_{i=1}^n{w}_i}{s}_i;i=1,2,3,\dots ,n$$

(1)

where $w_i$ represents the weight of the ith indicator in three different levels; $s_i$ represents the corresponding score of the ith indicator; and $S$ represents the comprehensive score of an indicator corresponding to the higher level. The flow chart of the multilevel community resilience assessment is depicted in Figure 7.

Each tertiary index with the resilience index system is assessed on a scale of five grades, with a maximum score of 5 points. The qualitative evaluation of the resilience grade corresponding to 1 to 5 points for each grade ranges from very poor to very good. The units of the indicators are determined by their definition, such as percentage, ratio, actual quantity, qualitative grading, and qualitative determination (refer to Table 4 and

Table 5. Grading and definition of all tertiary indexes corresponding to system and management resilience.

Tertiary Index	Index Score (Score the Index According to the Index Interpretation and Divide the Score into 1 to 5 Points)	Index Interpretation
C35	5--For all emergencies, the command department can mobilize all relevant rescue forces and required materials. 4--For most emergencies, the command department can mobilize all relevant rescue forces and required materials. 3--For general emergencies, the command department can mobilize all relevant rescue forces and required materials. 2--For a few emergencies, the command department can mobilize all relevant rescue forces and required materials. 1--No command department.	For emergencies, the command department can mobilize the command of all relevant rescue forces and required materials. Coordination ability of multiple stakeholders, government coordination capacity, decision-making ability, emergency response capacity, and coordination capacity.
C36	5--Complete disaster prevention and reduction planning. 4--Relatively complete disaster prevention and reduction-related planning. 3--General disaster prevention and reduction planning. 2--Incomplete disaster prevention and reduction planning. 1--No relevant content.	The comprehensive disaster prevention and reduction system includes fire control planning, earthquake prevention and mitigation planning, flood control planning, civil air defense planning, and public safety planning and also requires hazardous chemical safety planning when involving hazardous chemicals.
C37	5--Complete special plans. 4--Relatively complete special plans. 3--General special plans. 2--Incomplete special plans. 1--No relevant content.	The emergency plans formulated by the relevant departments and units of the local people’s government, according to their division of responsibilities deal with a certain type of public emergency with significant impact, which are aimed at the specific accident categories.
C38	5--Scientifically formulate emergency preparedness drill plans and regularly conduct comprehensive and special drills with strong emergency response-ability. 4--Has formulated emergency preparedness drill plans and has conducted comprehensive and special drills with relatively strong emergency response-ability. 3--Has incomplete emergency preparedness drill plans, irregularly conduct emergency drills, with general emergency response-ability. 2--No emergency plan exercise plan, seldom conducts emergency drills, with poor emergency response-ability. 1--No relevant content.	The purpose of carrying out comprehensive and special drills with strong practicality and wide participation of the masses across industries and departments is to test the effectiveness of the plan, run in the working mechanism, train the emergency team, and improve the emergency response ability.
C39	5-->10%, ≤100% 4-->7%, ≤10% 3-->4%, ≤7% 2-->1%, ≤4% 1--≤1%	The proportion of emergency service personnel in the total employed population is calculated based on statistical data.
C40	5-->200, ≤300 4-->150, ≤200 3-->100, ≤150 2-->50, ≤100 1--≤50	Total number of volunteers.
C41	5--Complete. 4--Relatively complete. 3--Basically complete. 2--Generally complete. 1--No.	Qualitative description of the reserve situation of relief goods and materials, living goods and materials, three protection goods and materials, etc.
C42	5--National comprehensive disaster reduction demonstration community. 4--Province-level comprehensive disaster reduction demonstration community. 3--City-level comprehensive disaster reduction demonstration community. 2--District-level comprehensive disaster reduction demonstration community. 1--No.	Whether it is a comprehensive disaster reduction demonstration community reflects the disaster response capacity and resilience level at the community level.
C43	5--4 4--3 3--2 2--1 1--0	It achieves the purpose of educating a child to affect a family, drives the whole society to improve the awareness of earthquake prevention and disaster reduction, and has a far-reaching and positive impact on ensuring social stability and the smooth progress of economic construction.
C44	5-->200, ≤250 4-->150, ≤200 3-->100, ≤150 2-->50, ≤100 1--≤50	The government concentrates part of social resources to provide public services for the market, and the financial strength to meet the public needs of the society, including tax revenue, public debt, and non-tax revenue, largely determines the scale of public expenditure (unit: ten thousand yuan).
C45	5-->20%, ≤100% 4-->15%, ≤20% 3-->5%, ≤15% 2-->1%, ≤5% 1--≤1%	Statistics calculate specific numbers, such as the share of public security and defense spending.
C46	5-->80%, ≤100% 4-->60%, ≤80% 3-->30%, ≤60% 2-->10%, ≤30% 1--≤10%	Can be measured by the value of commercial insurance company insurance, divided into property insurance, life insurance, and health insurance.
C47	5-->80%, ≤100% 4-->60%, ≤80% 3-->40%, ≤60% 2-->20%, ≤40% 1--≤20%	It is an important part of the social security system, including pension insurance, medical insurance, unemployment insurance, industrial and commercial insurance, and maternity insurance.
C48	5--Comprehensive evaluations have existed within the last 3 years and have been reviewed by a third party. Extreme disaster conditions scenarios and conventional disaster conditions scenarios are widely accepted. 4--Comprehensive evaluations exist, with minor problems with update time. 3--Comprehensive evaluations exist, with serious problems with update time. 2--Some evaluations exist but are not comprehensive enough, or the assessment time has exceeded 3 years. 1--No evaluation.	Including earthquake disasters, geological disasters, meteorological disasters, flood and drought disasters, marine disasters, forest and grassland fires, and other major natural disaster investigation and assessment of qualitative evaluation.
C49	5--Disaster risk maps have been compiled and are updated regularly. 4--Most disaster risk maps have been compiled and are updated irregularly. 3--Part of disaster risk maps have been compiled and are not updated. 2--Only a small part of disaster risk maps has been compiled. 1--Not compiled.	It includes the qualitative evaluation of various scale disaster risk zoning maps for different types of disasters, comprehensive risk zoning maps for all types of natural disasters, and specialized disaster risk zoning maps for each type of disaster.
C50	5--Regularly carry out and update investigations. 4--Has conducted investigations and has minor problems with update time. 3--Has conducted investigations and has serious problems with update time. 2--Has conducted some investigations, but the coverage is insufficient, or the investigation time has exceeded 3 years. 1--No investigation.	To investigate and evaluate the qualitative evaluation of the key hazards caused by various natural disasters.
C51	5-->95%, ≤100% 4-->90%, ≤95% 3-->85%, ≤90% 2-->80%, ≤85% 1--≤80%	The coverage rate of video surveillance in key public areas.
C52	5-->80%, ≤100% 4-->60%, ≤80% 3-->40%, ≤60% 2-->20%, ≤40% 1--≤20%	The coverage rate of disaster monitoring, early warning, emergency command, and information platforms in key public areas.
C53	5--Comprehensive monitoring, accurate judgment, and orderly organization. 4--Relatively comprehensive monitoring, relatively accurate judgment, relatively orderly organization. 3--Partial monitoring of the area, general judgment, and organization. 2--Low coverage of monitoring area, low judgment, disorderly organization. 1--No monitoring.	Determine the epidemic of infectious diseases according to the actual situation (prediction model, organization, process, mechanism, etc.).

). Quantitative calculations based on investigation data are used to determine the resilience index value. For Tertiary index C, indicator scores are calculated according to definitions and statistical data in two cases. When the score value involves percentage, ratio, or actual quantity falling within the scoring range (a, b] (score is 1–5 points) or when involves qualitative grading or judgment, the data are calculated according to the text description in Table 4 and Table 5.

3.2. Bayesian Inference-Based Relative Demand Coefficient of Different Community

To facilitate the comparison of different communities, the concept of a “relative demand coefficient” (RDC) was developed to quantitatively measure the relative necessity level that a community has for the ith indicator listed in Table 4 and Table 5. The scores of tertiary indices can be expressed by incorporating the RDCs in this paper by the following equation.

$$\{\begin{array}{l}{r}_i^j=\max \left(\min \left(\frac{s_i^j}{{\alpha }_i^j},5\right),1\right);\\ \mathrm{s}.\mathrm{t}.\hspace{0.17em}i=1,2,3,\dots ,p;j=1,2,3,\dots ,q;\hspace{0.17em}{r}_i^j\in [1,\hspace{0.17em}5]\end{array}\hspace{0.17em}\hspace{0.17em}$$

(2)

where $s_i^j$ and ${\alpha }_i^j$, respectively, represent the score and its RDC of the ith tertiary indicator in the $j\mathrm{th}$ community, while the $r_i^j$ is the corresponding RDC-adjusted score; $p$ denotes the number of tertiary indices for which the RDC needs to be considered and $q$ denotes the number of communities; the operator $\max (a,b)$ and $\min (a,b)$ represent taking the maximum and minimum values between $a$ and $b$, respectively, thus the value range of $r_i^j$ is confined between 1 and 5.

The inclusion of RDC fulfills a normalizing function, facilitating the comparison of resilience between different communities. This is attributed to the fact that communities vary in their demand for specific indicators, influenced by factors such as geographical location, development level, and socio-economic status. Consequently, it becomes challenging to rely solely on scores for measuring their resilience. For instance, a thriving community situated in an urban core and a picturesque resort located in a suburban area would inevitably exhibit distinct requirements for transportation capabilities (Indicator B₄). Even if they achieve identical scores on indicators C₂₇ and C₂₈, it cannot be definitively concluded that they possess an equivalent level of resilience concerning transportation capabilities.

The random nature of RDCs is considered in this paper, and Bayesian inference is employed to estimate their most probable values. The posterior probability density function (PDF) of an RDC ${\alpha }_i^j$ can be obtained through the Bayesian formulation:

$$p\left({\alpha }_i^j|\mathcal{D}\right)=\frac{p\left(\mathcal{D}|{\alpha }_i^j\right)p\left({\alpha }_i^j\right)}{p\left(\mathcal{D}\right)};i=1,2,3,\dots ,p;\hspace{0.17em}j=1,2,3,\dots ,q$$

(3)

where $\mathcal{D}$ represents the score of a set of demand-related tertiary indices from the universal set of tertiary indices. The corresponding likelihood $p\left(\mathcal{D}|{\alpha }_i^j\right)$ can be expressed in the following form:

$$p\left(\mathcal{D}|{\alpha }_i^j\right)=\frac{1}{{(2\pi {\sigma }^2)}^{1/2}}\exp (-\frac{1}{2{\sigma }^2}{\Vert {\tilde{\alpha }}_i^j-{\alpha }_i^j\Vert }_2^2);i=1,2,3,\dots ,p;\hspace{0.17em}j=1,2,3,\dots ,q$$

(4)

The symbol ${\tilde{\alpha }}_i^j$ represents the normalized observed RDC, which follows a Gaussian distribution with the expectation ${\alpha }_i^j$ and variance ${\sigma }^2$. The normalized observed RDC ${\tilde{\alpha }}_i^j$ in Equation (4) is obtained using the subsequent formulas:

$$\{\begin{array}{l}{\tilde{\alpha }}_i^j=\frac{{\overline{\alpha }}_i^j-\inf ({\overline{\alpha }}_i^j)}{\sup ({\overline{\alpha }}_i^j)-\inf ({\overline{\alpha }}_i^j)}+0.5;\\ \mathrm{s}.\mathrm{t}.i=1,2,3,\dots ,p;j=1,2,3,\dots ,q;{\tilde{\alpha }}_i^j\in \left[0.5,\hspace{0.17em}1.5\right]\end{array}$$

(5)

with

$$\{\begin{array}{l}{\overline{\alpha }}_i^j={\displaystyle \sum _m{\kappa }_{i,m}{s}_m^j}+{\displaystyle \sum _n{\kappa }_{i,n}(6-s_n^j)};\\ \mathrm{s}.\mathrm{t}.\hspace{0.17em}i=1,2,3,\dots ,p;j=1,2,3,\dots ,q;\inf ({\overline{\alpha }}_i^j)=1;\hspace{0.17em}\sup ({\overline{\alpha }}_i^j)=5\hspace{1em}\hspace{0.17em}\end{array}$$

(6)

where ${\overline{\alpha }}_i^j$ is defined as the demand influenced score (DIS); $s_m^j$ and ${\kappa }_{i,m}$, respectively, indicate the score and its weighting of the positively influenced $m\mathrm{th}$ tertiary indicator from the $j\mathrm{th}$ community, while $s_n^j$ and ${\kappa }_{i,n}$, respectively, denote the score and its weighting of the negatively influenced $n\mathrm{th}$ tertiary indicator from the $j\mathrm{th}$ community; the symbols $\inf (·)$ and $\sup (·)$ represent the operation of supremum and infimum, respectively.

Note that the DIS ${\overline{\alpha }}_i^j$ in Equation (6) comprises two components of the weighted sum of scores for tertiary indices, where the weights are determined according to expert knowledge through the Delphi method. The first component aggregates scores that have a positive influence with ${\alpha }_i^j$, while the second component sums those with a negative influence. For instance, when there is a higher score for employment rate (C₅), it indicates greater demand for transportation hubs (C₂₇), resulting in a positive influence from $s_5$ to ${\overline{\alpha }}_{27}$. Conversely, when there is a higher score for the percentage of the floating population (C₁₁), it implies a smaller floating population, leading to a negative influence from $s_{11}$ to ${\overline{\alpha }}_{27}$. The range of values of scores is [1,5], so scores from negatively influenced indicators are subtracted from 6 to maintain their value within [1,5]. After calculating the DIS for all $q$ communities using Equation (6), the normalization process described in Equation (5) is applied to maintain the normalized observed RDC variating between 0.5 and 1.5.

The prior $p\left({\alpha }_i^j\right)$ in Equation (3) represents the evaluation of the demand level for a specific tertiary index based on empirical judgment, assuming to follow a Gaussian distribution in terms of ${\alpha }_i^j\sim \mathcal{N}\left(c_i^j,{\epsilon }^2\right)$, and can be mathematically expressed as

$$\{\begin{array}{l}p\left({\alpha }_i^j\right)=\frac{1}{{(2\pi {\epsilon }^2)}^{1/2}}\exp (-\frac{1}{2{\epsilon }^2}{\Vert {\alpha }_i^j-c_i^j\Vert }_2^2);\\ \mathrm{s}.\mathrm{t}.\hspace{0.17em}i=1,2,3,\dots ,p;\hspace{0.17em}j=1,2,3,\dots ,q;\hspace{0.17em}{c}_i^j\in \left[0.5,\hspace{0.17em}1.5\right]\end{array}$$

(7)

where the constant $c_i^j$ represents the expectation of the prior RDC with a preset value assumed to be determined between 0.5 and 1.5 based on expert knowledge or practical experience; ${\epsilon }^2$ denotes the variance of the prior RDC, coming from expert knowledge or experience.

Additionally, $p\left(\mathcal{D}\right)$ in Equation (3) is the evidence and is generally not involved in the estimation of the most probable value of ${\alpha }_i^j$. Consequently, the maximum a posterior (MAP) of ${\alpha }_i^j$ estimation can be conducted as follows:

$$\begin{array}{cc}\hfill {\left({\alpha }_i^j\right)}^{\mathrm{MAP}}& =\arg \underset{{\alpha }_i^j}{\max }p\left({\alpha }_i^j|\mathcal{D}\right)\hfill \\ & =\arg \underset{{\alpha }_i^j}{\min }-\ln p\left({\alpha }_i^j|\mathcal{D}\right)\hfill \\ & =\arg \underset{{\alpha }_i^j}{\min }\left\{-\ln p\left(\mathcal{D}|{\alpha }_i^j\right)-\ln p\left({\alpha }_i^j\right)\right\}\hfill \\ & =\arg \underset{{\alpha }_i^j}{\min }\left\{\frac{1}{{\sigma }^2}{\Vert {\tilde{\alpha }}_i^j-{\alpha }_i^j\Vert }_2^2+\frac{1}{{\epsilon }^2}{\Vert {\alpha }_i^j-c_i^j\Vert }_2^2\right\}\hfill \end{array}$$

(8)

By taking the derivative of Equation (8) with respect to ${\alpha }_i^j$ and setting it equal to zero as

$$-\frac{2}{{\sigma }^2}({\tilde{\alpha }}_i^j-{\alpha }_i^j)+\frac{2}{{\epsilon }^2}({\alpha }_i^j-c_i^j)=0$$

(9)

one can determine that

$${\left({\alpha }_i^j\right)}^{MAP}=\frac{{\epsilon }^2{\tilde{\alpha }}_i^j+{\sigma }^2{c}_i^j}{{\sigma }^2+{\epsilon }^2}$$

(10)

where ${\left({\alpha }_i^j\right)}^{MAP}$ in Equation (10) represents the most probable value (MPV) of ${\alpha }_i^j$. Subsequently, these values are utilized in Equation (2) to calculate the RDC-adjusted scores $r_i^j$ of tertiary indices.

3.3. Implementation Procedure of the Multilevel Resilience Evaluation

The definition and grading of the multilevel resilience index system for the community are developed in Section 3.1. Subsequently, the Bayesian inference approach is proposed to calculate the adjusted scores of tertiary indices considering the RDCs. According to the flow chart of the multilevel community resilience assessment depicted in Figure 7, the detailed implementation procedure of the proposed BAHP-based multilevel resilience evaluation is described as follows:

Step 1: Data collection is initially conducted through field research, questionnaire interviews, existing statistical reports from relevant government agencies, etc., according to the definition and grading of all tertiary indexes corresponding to social and environmental resilience as well as institutional and managerial resilience (refer to Table 4 and Table 5).

Step 2: For all the target communities, expectations for prior RDC ($c_i^j$) and variances for prior RDC (${\epsilon }^2=0.1c_i^j$) are preset based on expert knowledge or practical experience.

Step 3: By using Equation (6), the DIS (${\overline{\alpha }}_i^j$) is computed for all the target communities based on the scores of the ith tertiary indicator in the jth community along with corresponding weights of positively and negatively related indicators to ${\alpha }_i^j$. After obtaining DIS, normalization is performed using Equation (5) to compute the normalized observed RDC (${\tilde{\alpha }}_i^j$) for all the target communities. Then, variances for likelihood (${\sigma }^2=0.1{\tilde{\alpha }}_i^j$) are determined by expert knowledge or practical experience.

Step 4: Calculate the MPV of RDC, ${\left({\alpha }_i^j\right)}^{MAP}$, for all target communities using Equation (10). Then, substitute ${\left({\alpha }_i^j\right)}^{MAP}$ into Equation (2) to compute the relative demand coefficient-adjusted scores $r_i^j$ for all target communities.

Step 5: After the previous data collection and determination of the weights of resilience evaluation index elements at each level by the Delphi method based on expert knowledge or experience, scoring computation for the tertiary indexes is conducted. According to the BAHP approach, the weighted average method is, respectively, adopted to calculate scores of resilience indexes for all three levels using Equation (1).

Step 6: Compare resilience indices at all three levels among different communities to identify their weaknesses and develop corresponding strategies to enhance community resilience capacity.

Moreover, the algorithmic details of the Bayesian inference for RDC-adjusted scores calculation are summarized and presented in Algorithm 1.

Algorithm 1. Computation process of Bayesian inference for RDC-adjusted scores of tertiary indices

Inputs:  - Scores $s_i^j$ of tertiary indicators;  - The expectation of the prior RDC, $c_i^j$, coming from expert knowledge/experience;  - The variance of the prior RDC determined by ${\epsilon }^2=0.1c_i^j$.

Solution process: for $i=1:p$   for $j=1:q$   - Calculate the DIS ${\overline{\alpha }}_i^j$ for all communities using Equation (6)   end   - Obtain the normalized observed RDC, ${\tilde{\alpha }}_i^j$, for all communities using Equation (5)   - Determine the variance of the likelihood by ${\sigma }^2=0.1{\tilde{\alpha }}_i^j$   - Obtain the MPV of RDC, ${\left({\alpha }_i^j\right)}^{MAP}$, for all communities using Equation (10)   - Substituting ${\left({\alpha }_i^j\right)}^{MAP}$ into Equation (2) to calculate the RDC-adjusted score of the tertiary index $r_i^j$ end

Outputs: RDC-adjusted scores of tertiary indices $r_i^j$ for $i=1:p$, $j=1:q$

4. Case Study and Application Demonstration

4.1. Profile Description of the Dafapu Community and the Virtual Community

The proposed BAHP-based multilevel community resilience evaluation method and index system are applied in typical communities in Shenzhen city to validate its effectiveness and provide guidance for local government in enhancing community resilience. Shenzhen, as the first special economic zone established after China’s reform and open policy, is not only one of the four super first-tier cities but also a hub for economy, culture, science, and technology. It is situated on the southern coast of China, east of the Pearl River Estuary, in proximity to Hong Kong. To the east lie Daya Bay and Dapeng Bay, while to the west are the Pearl River Estuary and Lingding Bay. In the south lies the Shenzhen River connecting to Hong Kong, while Dongguan and Huizhou border it from the north. Covering an area of about 1997.47 km², Shenzhen lies south of the Tropic of Cancer, experiencing a warm subtropical monsoon climate with an average annual temperature of 23.0 degrees Celsius. Due to the monsoon climate influence throughout the year, average annual rainfall reaches 1935.8 mm with significant variations between seasons; summer experiences higher precipitation compared to winter and spring months. The rainfall varies greatly in different seasons, with a significant increase in summer compared to winter and spring. Owing to the terrain, the spatiotemporal distribution of rainfall in Shenzhen is uneven, showing a trend of more in the southeast and less in the northwest, with an average annual rainy day of 144 days.

In recent years, Shenzhen has been prioritizing the development of urban disaster prevention and mitigation, aiming to enhance safety resilience. This includes (1) highlighting the crucial role of prevention, monitoring, and early warning in emergency management; (2) emphasizing the significance of disaster prevention engineering system as a protective barrier for urban safety; (3) focusing on the fundamental achievements of comprehensive community construction for disaster reduction; (4) recognizing the important impact of emergency shelters in urban emergency response; (5) maximizing the guarantee provided by emergency goods and materials in rescue operations and disaster relief efforts; (6) pioneering nationwide implementation of a catastrophe insurance system. Simultaneously, due to its complex urban system, unique geographical location, and significant economic influence, research and analysis regarding its community resilience evaluation are representative and valuable.

The Dafapu community, located in the Longgang district of Shenzhen city, is a typical comprehensive disaster reduction demonstration community. This community advances in organizational management and work system, risk assessment and hidden danger management, infrastructure construction and emergency plan exercise, emergency support materials and rescue forces, as well as publicity and education, etc. Therefore, we chose the Dafapu community to study Shenzhen’s community-level resilience evaluation. The Dafapu community was established in December 2006 and is situated on Bantian street. It falls under the jurisdiction of Wuhe Avenue to the east, Pingnan railway to the south, Meiguan expressway to the west, and Beier Road to the north. The total area covered by this jurisdiction is 1.1 km². The community has a population of 60,077 people (including 46,240 permanent residents and floating population of 13,837), with a total of 941 houses along with 372 industrial and mining enterprises. Additionally, there are 913 small venues (including small stalls, workshops, and entertainment venues). Within the Dafapu community exist three villages (Dafapu Village, Lishi Pai Village, Shuidoukeng Village) along with five garden apartments (Jiayu Hao Garden, Yulong Hao Garden, the One Apartment, Chuangxing Times, Jiahua Link Hui). The investigation area for resilient community evaluation in this paper is demonstrated in Figure 8.

Through conducting a field investigation on the resilience construction of the community, which involved interviews with key personnel from the community emergency management department, community emergency management staff, and community residents, as well as data collection, we obtained a comprehensive understanding of the working environment related to the resilience construction in this community. This understanding was then integrated with the unique characteristics of the community for a holistic analysis. Additionally, we utilized scores from various indices for a hypothetical virtual community evaluated using the same method to compare and elaborate on the proposed method in detail. The simulated virtual community is assumed to be situated in the Longgang District suburb in Shenzhen, and it is primarily characterized by a dominant local population, significant aging demographics, and lower levels of development.

4.2. Computation of the Bayesian Inference-Based RDCs for Two Target Communities

To demonstrate the feasibility and effectiveness of the proposed BAHP-based multilevel community resilience evaluation method, this section conducts a resilience evaluation and comparison between two communities with different RDCs. The Dafapu community and a virtual community will be used as examples to provide a detailed explanation of the solution process for the proposed RDC-adjusted scores of tertiary indices. Initially, three groups of demand-related indicators (C_24,25; C_27,28; C_36~43,48~52) are identified for the target communities that require consideration of RDC in this paper. These groups have distinct demand-influencing indicators, influence direction, and weights. Since demand-related indicators within each group share common data or information mentioned above, their corresponding RDCs are also shared. The specific solution process for these three groups of demand-related indicators will be elaborated below.

(a)

The Process for Calculating the RDC of C₂₄ and C₂₅

Parameters for demand-related indicators to Dafapu and virtual communities are listed in

Table 6. Parameters for demand-related indicators to the Dafapu and the virtual communities.

Demand-Related Indicators	Demand-Influencing Indicators	Influence Direction	Weights $\mathit{\kappa }$
C24,25	C7	negative	0.3
	C8	positive	0.2
	C9	negative	0.1
	C13	negative	0.1
	C32	negative	0.1
	C33	negative	0.1
	C53	negative	0.1
C27,28	C5	positive	0.3
	C8	positive	0.1
	C11	negative	0.1
	C29	negative	0.05
	C30	negative	0.05
	C44	positive	0.4
C36~43 C48~52	C1	negative	0.2
	C2	negative	0.2
	C3	negative	0.2
	C20	negative	0.1
	C23	negative	0.1
	C34	negative	0.1

, which includes the demand-influencing indicators, influence direction, and weights. The RDC for demand-related indicators in terms of the number of beds in community medical and health institutions (C₂₄) and capacity of community nursing institutions (C₂₅) are associated with several demographic and environmental factors, including percentage of elderly population (C₇), percentage of underage population (C₈), percentage of the population with physical or mental disabilities (C₉), number of food safety incidents (C₁₃), air quality compliance rate (C₃₂), water quality compliance rate (C₃₃), and epidemic monitoring and control situation (C₅₃). Among these factors, C₇ has the strongest influence on C₂₄ and C₂₅ as a higher percentage of elderly individuals increases the demand for beds and nursing capacity, whose weight is 0.3. Therefore, the weight $\kappa$ assigned to C₇ is 0.3, indicating its significant impact. It should be noted that there is only one positive influencing indicator with $m=8$ for both C₂₄ and C₂₅, while all other influencing indicators have negative effects. Considering prior information about the three hospitals within a 4 km radius of Dafapu community, its prior demand for beds and nursing capacity is not considered high when initially set at $c_i^1=1.0$ for $i=24,25$. In contrast, since the virtual community is located in a suburban area without hospitals within a 4 km radius, its prior demand for these facilities is considered higher initially set at $c_i^2=1.5$ for $i=24,25$, which exceeds $c_i^1$. Then, variances of the prior RDC are obtained by ${\left({\epsilon }^2\right)}_i^j=0.1c_i^j$ as ${\left({\epsilon }^2\right)}_i^1=0.1$ and ${\left({\epsilon }^2\right)}_i^2=0.15$ for $i=24,25$. The DIS of C₂₄ and C₂₅ can be subsequently obtained using Equation (6) as ${\overline{\alpha }}_i^1=1.9$ and ${\overline{\alpha }}_i^2=2.5$ for $i=24,25$, where the score of the corresponding indicator $s_{i,j}$ can be found in the third column from the right in

Table 7. Evaluation results of social and environmental resilience of the Dafapu community.

Primary Index	Weight Coefficient	Score	RDC Adjusted Score	Secondary Index	Weight Coefficient	Score	RDC Adjusted Score	Tertiary Index	Weight Coefficient	Score	RDC	RDC Adjusted Score
A1	0.5	3.740	3.588	B1	0.2	2.6	2.6	C1	0.60	3	1.0	3
								C2	0.30	1	1.0	1
								C3	0.10	5	1.0	5
				B2	0.2	4.2	4.2	C4	0.15	3	1.0	3
								C5	0.10	5	1.0	5
								C6	0.10	4	1.0	4
								C7	0.05	5	1.0	5
								C8	0.05	3	1.0	3
								C9	0.05	5	1.0	5
								C10	0.10	4	1.0	4
								C11	0.10	3	1.0	3
								C12	0.10	5	1.0	5
								C13	0.05	5	1.0	5
								C14	0.15	5	1.0	5
				B3	0.2	4.3	4.385	C15	0.10	5	1.0	5
								C16	0.10	5	1.0	5
								C17	0.10	5	1.0	5
								C18	0.10	4	1.0	4
								C19	0.10	5	1.0	5
								C20	0.05	4	1.0	4
								C21	0.10	3	1.0	3
								C22	0.10	5	1.0	5
								C23	0.05	5	1.0	5
								C24	0.10	3	0.841	3.567
								C25	0.05	3	0.841	3.567
								C26	0.05	4	1.0	4
				B4	0.2	4.0	3.155	C27	0.50	4	1.268	3.155
								C28	0.50	4	1.268	3.155
				B5	0.2	3.6	3.6	C29	0.20	4	1.0	4
								C30	0.10	3	1.0	3
								C31	0.20	1	1.0	1
								C32	0.20	4	1.0	4
								C33	0.20	5	1.0	5
								C34	0.10	5	1.0	5

,

Table 8. Evaluation results of social and environmental resilience of the virtual community.

Primary Index	Weight Coefficient	Score	RDC Adjusted Score	Secondary Index	Weight Coefficient	Score	RDC Adjusted Score	Tertiary Index	Weight Coefficient	Score	RDC	RDC Adjusted Score
A1	0.5	3.330	3.558	B1	0.2	2.9	2.9	C1	0.60	3	1.0	3
								C2	0.30	2	1.0	2
								C3	0.10	5	1.0	5
				B2	0.2	3.7	3.7	C4	0.15	4	1.0	4
								C5	0.10	2	1.0	2
								C6	0.10	3	1.0	3
								C7	0.05	2	1.0	2
								C8	0.05	1	1.0	1
								C9	0.05	5	1.0	5
								C10	0.10	4	1.0	4
								C11	0.10	5	1.0	5
								C12	0.10	5	1.0	5
								C13	0.05	4	1.0	4
								C14	0.15	4	1.0	4
				B3	0.2	3.45	3.441	C15	0.10	4	1.0	4
								C16	0.10	4	1.0	4
								C17	0.10	4	1.0	4
								C18	0.10	3	1.0	3
								C19	0.10	5	1.0	5
								C20	0.05	3	1.0	3
								C21	0.10	3	1.0	3
								C22	0.10	5	1.0	5
								C23	0.05	5	1.0	5
								C24	0.10	1	1.105	1
								C25	0.05	2	1.105	1.810
								C26	0.05	1	1.0	1
				B4	0.2	2.0	3.150	C27	0.50	2	0.635	3.150
								C28	0.50	2	0.635	3.150
				B5	0.2	4.6	4.6	C29	0.20	5	1.0	5
								C30	0.10	5	1.0	5
								C31	0.20	3	1.0	3
								C32	0.20	5	1.0	5
								C33	0.20	5	1.0	5
								C34	0.10	5	-	5

,

Table 9. Evaluation results of institutional and managerial resilience of the Dafapu community.

Primary Index	Weight Coefficient	Score	RDC Adjusted Score	Secondary Index	Weight Coefficient	Score	RDC Adjusted Score	Tertiary Index	Weight Coefficient	Score	RDC	RDC Adjusted Score
A2	0.5	4.625	4.766	B6	0.25	4.30	4.863	C35	0.10	4	1.0	4
								C36	0.10	5	$0.615$	5
								C37	0.10	5	$0.615$	5
								C38	0.10	5	$0.615$	5
								C39	0.20	3	$0.615$	4.878
								C40	0.10	5	$0.615$	5
								C41	0.10	5	$0.615$	5
								C42	0.10	5	$0.615$	5
								C43	0.10	3	$0.615$	4.878
				B7	0.25	4.20	4.20	C44	0.30	4	1.0	4
								C45	0.30	4	1.0	4
								C46	0.20	4	1.0	4
								C47	0.20	5	1.0	5
				B8	0.25	5.0	5.0	C48	0.40	5	$0.615$	5
								C49	0.30	5	$0.615$	5
								C50	0.30	5	$0.615$	5
				B9	0.25	5.0	5.0	C51	0.30	5	$0.615$	5
								C52	0.40	5	$0.615$	5
								C53	0.30	5	1.0	5

and

Table 10. Evaluation results of institutional and managerial resilience of the virtual community.

Primary Index	Weight Coefficient	Score	RDC Adjusted Score	Secondary Index	Weight Coefficient	Score	RDC Adjusted Score	Tertiary Index	Weight Coefficient	Score	RDC	RDC Adjusted Score
A2	0.5	2.975	2.930	B6	0.25	3.1	3.044	C35	0.10	4	1.0	4
								C36	0.10	3	1.022	2.935
								C37	0.10	3	1.022	2.935
								C38	0.10	3	1.022	2.935
								C39	0.20	3	1.022	2.935
								C40	0.10	3	1.022	2.935
								C41	0.10	5	1.022	4.892
								C42	0.10	3	1.022	2.935
								C43	0.10	1	1.022	1
				B7	0.25	2.4	2.4	C44	0.30	2	1.0	2
								C45	0.30	2	1.0	2
								C46	0.20	2	1.0	2
								C47	0.20	4	1.0	4
				B8	0.25	3.7	3.620	C48	0.40	4	1.022	3.914
								C49	0.30	3	1.022	2.935
								C50	0.30	4	1.022	3.914
				B9	0.25	2.7	2.655	C51	0.30	3	1.022	2.935
								C52	0.40	3	1.022	2.935
								C53	0.30	2	1.0	2

. After substituting the calculated DIS ($\overline{\alpha }$) into Equation (5), the normalized observed RDC can be determined as ${\tilde{\alpha }}_i^1=0.725$ and ${\tilde{\alpha }}_i^2=0.875$ for $i=24,25$. By setting ${\left({\sigma }^2\right)}_i^j=0.1{\tilde{\alpha }}_i^j$ as ${\left({\sigma }^2\right)}_i^1=0.0725$ and ${{\sigma }^2}_i^2=0.0875$, the Bayesian inference can then be applied to obtain MPV of RDC for all communities by incorporating prior information using Equation (10), resulting in ${\left({\alpha }_{24}^1\right)}^{\mathrm{MAP}}={\left({\alpha }_{25}^1\right)}^{\mathrm{MAP}}=0.841$ and ${\left({\alpha }_{24}^2\right)}^{\mathrm{MAP}}={\left({\alpha }_{25}^2\right)}^{\mathrm{MAP}}=1.105$. The complete calculation process and specific values of priors are summarized in

Table 11. The preset parameters and the computed MPV of RDCs for the Dafapu and the virtual communities.

Parameters	Demand-Related Indicators
	Dafapu Community			Virtual Community
	C24,25	C27,28	C36~43 C48~52	C24,25	C27,28	C36~43 C48~52
Expectation of the prior RDC ($c_i^j$)	1	1.3	0.5	1.5	0.6	1.5
Variance of the prior RDC (${\epsilon }^2$)	0.1	0.13	0.05	0.15	0.06	0.15
DIS value (${\overline{\alpha }}_i^j$)	1.9	3.95	2.2	2.5	1.7	2.1
Normalized observed RDC (${\tilde{\alpha }}_i^j$)	0.725	1.2375	0.8	0.875	0.675	0.775
Variance of the likelihood (${\sigma }^2$)	0.0725	0.12375	0.08	0.0875	0.0675	0.0775
MPV of RDC ${\left({\alpha }_i^j\right)}^{MAP}$	0.841	1.268	0.615	1.105	0.635	1.022

. Finally, RDC-adjusted scores of C₂₄ and C₂₅ for both communities can be obtained by applying Equation (2), which are listed in the last column of Table 7 and Table 8, respectively.

(b)

The Process for Calculating the RDC of C₂₇ and C₂₈

As indicated in Table 6, the RDC for demand-related indicators, specifically the number of transportation hubs (C₂₇) and traffic mileage of urban rail and main road (C₂₈), is associated with the employment rate (C₅), percentage of underage population (C₈), percentage of floating population (C₁₁), percentage of forest cover (C₂₉), per capita public green space (C₃₀), and public financial strength (C₄₄). The most significant influences are observed from C₅ and C₄₄, whose weights are, respectively, 0.3 and 0.4, as the daily commute demands a substantial transportation capacity from the employed population, while it would be inefficient for financially undeveloped areas to construct an extensive network of transportation hubs. The number of indicators positively influencing C₂₇ and C₂₈ is $m=5,8,44$, whereas there are indicators with n = 11, 29, 30 exerting negative influence. Considering prior information, the Dafapu community’s strategic location as a transit point for all surrounding districts heading towards Shenzhen city center necessitates high transportation capacity; therefore, it is assigned a prior demand coefficient of $c_i^1=1.3$ for $i=27,28$. In contrast, the virtual community situated in a suburban area does not necessitate such high transportation capacity. Therefore, a prior demand coefficient of $c_i^2=0.6$ is assigned for $i=27,28$. Subsequently, variances of the prior RDC are obtained through ${\epsilon }_i^j=0.1c_i^j$ as ${\left({\epsilon }^2\right)}_i^1=0.13$ and ${\left({\epsilon }^2\right)}_i^2=0.06$ for $i=27,28$. By utilizing Equation (6), one can calculate DIS values as ${\overline{\alpha }}_i^1=3.95$, ${\overline{\alpha }}_i^2=1.7$ for $i=27,28$. By substituting the DIS values into Equation (5), we can obtain that ${\tilde{\alpha }}_i^1=1.2375$ and ${\tilde{\alpha }}_i^2=0.675$ for $i=27,28$. Setting ${\left({\sigma }^2\right)}_i^j=0.1{\tilde{\alpha }}_i^j$ as ${\left({\sigma }^2\right)}_i^1=0.12375$ and ${\left({\sigma }^2\right)}_i^2=0.0675$, MPV of RDC can be calculated using Equation (10), resulting in ${\left({\alpha }_i^1\right)}^{\mathrm{MAP}}=1.268$ and ${\left({\alpha }_i^2\right)}^{\mathrm{MAP}}=0.635$ for i = 27, 28. The complete calculation process, along with specific values of priors for C₂₇ and C₂₈, are listed in Table 7. RDC-adjusted scores of C₂₇ and C₂₈ for both communities are derived by applying Equation (2), which are presented in the last column of Table 7 and Table 8, respectively.

(c)

The Process for Calculating RDC for C_36~43 and C_48~52

As shown in Table 6, the RDCs for demand-related indicators C_36~43 and C_48~52 are associated with encompassing building fortification standard (C₁), proportion of self-built houses (C₂), and proportion of container-type mobile houses (C₃), level of inspection, monitoring, and maintenance of infrastructure (C₂₀), number of hazardous chemical enterprises (C₂₃), and acid rain frequency (C₃₄). The most significant impact is observed from C_1~3 due to non-standardized building constructions usually suffering a higher risk of damage during disasters, and their weights are 0.2. It should be noted that all the demand-influencing indicators have a negative influence on C_36~43 and C_48~52. Historically, the Dafapu community has experienced fewer natural disasters; thus, the expectation of prior RDC is chosen as $c_i^1=0.5$ for $i=36~43,48~52$. On the other hand, considering the unique topography leading to a higher risk of land subsidence in the virtual community’s location, its expectation of prior RDC is set at $c_i^2=1.5$ for i = 36~43, 48~52. The variances of the prior RDC are obtained by ${\epsilon }_i^j=0.01c_i^j$ as ${\left({\epsilon }^2\right)}_i^1=0.05$ and ${\left({\epsilon }^2\right)}_i^2=0.15$ for i = 36~43, 48~52. By utilizing Equation (6), one can derive DIS values as ${\overline{\alpha }}_i^1=2.2$ and ${\overline{\alpha }}_i^2=2.1$ for $i=36~43,48~52$. Substituting the DIS values into Equation (5), one can obtain the corresponding normalized observed RDCs as ${\tilde{\alpha }}_i^1=0.8$ and ${\tilde{\alpha }}_i^2=0.775$ for $i=36~43,48~52$. Finally, by setting ${\left({\sigma }^2\right)}_i^j=0.1{\tilde{\alpha }}_i^j$ as ${\left({\sigma }^2\right)}_i^1=0.08$ and ${\left({\sigma }^2\right)}_i^2=0.0775$, Equation (10) can be employed to solve MPV of RDC as ${\left({\alpha }_i^1\right)}^{\mathrm{MAP}}=0.615$ and ${\left({\alpha }_i^2\right)}^{\mathrm{MAP}}=1.022$ for $i=36~43,48~52$. The complete calculation process and specific values of priors for C_36~43 and C _48~52 are listed in Table 11. The RDC-adjusted scores of C_36~43 and C_48~52 for both communities can be determined using Equation (2), which are presented in the last column of Table 7 and Table 8, respectively.

4.3. Resilience Evaluation and Comparison of Two Communities with Different RDC

After obtaining the RDCs in Section 4.2, the scores and RDC-adjusted scores for all indicators at the three levels of the Dafapu community and virtual community are computed according to the implementation procedure introduced in Section 3.3. The corresponding scores are then listed in Table 7, Table 8, Table 9 and Table 10, respectively. It should be noted that except for the three groups of tertiary indicators mentioned in Table 6, which were considered for RDC calculation, other tertiary indicators had a uniform value of 1 assigned to their RDCs. Radar maps are utilized to visually represent the score distribution of indicators from the proposed multilevel community resilience index system (listed in Table 3) for these target communities.

4.3.1. The Case without Considering RDC

Firstly, the distributions of the scores of the tertiary indexes corresponding to the primary index of social and environmental resilience of the two communities are analyzed and compared in Figure 9 without considering RDC. In Figure 9, the red line represents a score of 3, which is considered the threshold value for assessing resilience in tertiary indexes. Generally, a score equal to or greater than 3 indicates satisfactory resilience, while a score less than 3 suggests insufficient resilience. Figure 9a illustrates the distribution of scores for tertiary indexes related to social and environmental resilience as the primary index. For the Dafapu community, most tertiary indexes scored above 3 except for the proportion of self-built houses (C₂) and percentage of undeveloped land (C₃₁), both receiving a score of 1. The virtual community has several tertiary indexes scoring below 3, including proportion of self-built houses (C₂), employment rate (C₅), percentage of elderly population (C₇), percentage of underage population (C₈), number of beds in community medical and health institutions (C₂₄), capacity of community nursing institutions (C₂₅), number of rehabilitation centers for persons with disabilities in the community (C₂₆), number of transportation hubs (C₂₇), and traffic mileage of urban rail and main road (C₂₈). The score distribution of tertiary indexes associated with institutional and managerial resilience as the primary index is presented in Figure 9b. The Dafapu community demonstrates exceptional performance in this aspect, with all tertiary indexes scoring above 3; in fact, most tertiary indexes achieve a score of 5. Conversely, the virtual community demonstrates significantly lower scores compared to Dafapu, with most tertiary indexes scoring an average of around 3. Among them, the number of science popularization demonstration schools for earthquake prevention and disaster reduction (C₄₃) only attains a score of 1, while public financial strength (C₄₄), ratio of emergency funds to fiscal expenditures (C₄₅), commercial insurance coverage rate (C₄₆), and epidemic monitoring and control situation (C₅₃) receive a score of 2. In comparison, the virtual community outperforms the Dafapu community on indicators C₄, C₁₁, C₂₉, C₃₀, and C₃₂. However, it either matches or falls behind the Dafapu community on all other tertiary indicator scores, particularly on indicators C_24~28.

Subsequently, the analysis of scores without RDC of the secondary indexes will be further conducted and compared. The scores of the Dafapu community’s secondary index corresponding to the primary index of social and environmental resilience are shown in Figure 10, where B₃ > B₄ > B₂ > B₅ > B₁. This reflects that infrastructure (B₃), transportation capacity (B₄), and population characteristics (B₂) indicators are all in good condition, with corresponding resilience values are 4.3, 4.2, and 4, respectively. The ecological environment index (B₅) exhibits a relatively low resilience value of 3.6 due to an extremely low score on the undeveloped land percentage index (C₃₁). This suggests that spatial layout, utilization, and safety aspects related to undeveloped land need comprehensive consideration and enhancement. The building construction index (B₁) exhibits the lowest resilience value of 2.6, which notably stands as the sole index to fall below a score of 3. This low score primarily results from the overall poor quality of building construction and the prevalence of self-built houses. Specifically, the score for the building fortification standard index (C₁) is 3, while the proportion of self-built houses index (C₂) receives a score of 1. More efforts should be made to improve this aspect, particularly focusing on enhancing construction safety performance for self-built houses. Simultaneously, Figure 10 illustrates the scores of the secondary index corresponding to the primary index of social and environmental resilience of the virtual community. It can be observed that B₅ > B₂> B₃> B₁> B₄, with their corresponding scores being 4.6, 3.7, 3.45, 2.9, and 2. The ecological environment index (B₅) demonstrates excellent conditions, while the transportation capacity index (B₄) exhibits the lowest level of resilience. Additionally, the score of the building construction index (B₁) is less than 3, indicating an inadequate resilience level. The comparison of the two communities is presented in Figure 10 regarding the distribution curves of scores without RDC for secondary indexes between the Dafapu community and the virtual community. Notably, compared to the Dafapu community, the virtual community excels in building construction (B₁) and ecological environment (B₅), performs similarly on population characteristics (B₂), but lags on infrastructure (B₃) and transportation capacity (B₄). The disparities mainly arise from geographical differences. Specifically, the virtual community is in a suburban area with a smaller transient population and a more conducive environment for living. However, its infrastructure and urban transportation are not as developed as those of the Dafapu community, which is situated in an urban area.

Similarly, the distributions of the scores of the tertiary indexes corresponding to the primary index of institutional and managerial resilience of the two communities, without considering RDC, are analyzed and compared. The scores of the secondary index corresponding to the primary index of institutional and managerial resilience of Dafapu are illustrated in Figure 10, where B₈ = B₉ > B₆ > B₇. This indicates that indicators for disaster risk assessment (B₈) and disaster monitoring and early warning (B₉) are essentially stable with a resilience value of 5. The scores of all four tertiary indexes are above 3, indicating that the Dafapu community possesses substantial institutional and managerial resilience. However, it is still necessary to continue improving according to national and provincial requirements. The emergency management capability (B₆) in the Dafapu community is relatively low at 4.3 due to the limited numbers of emergency service personnel and disaster prevention and mitigation demonstration schools, resulting in lower resilience scores for C₃₉ and C₄₃ at 3.0. Based on these findings, relevant departments such as government bodies, communities, and local authorities may need to enhance public efforts while strengthening education on disaster prevention and mitigation knowledge along with increasing professional technical personnel for emergency services. The score of economic resilience (B₇) is the lowest at 4.2. Upon closer examination of the corresponding tertiary indicators, it is evident that the scores of the public financial strength index (C₄₄), the ratio of emergency funds to fiscal expenditure index (C₄₅), and the commercial insurance coverage rate (C₄₆) are 4, which serve as the main restricting factors. Therefore, it is necessary for relevant government agencies and units to increase emergency fund allocations based on actual circumstances, allocate public finance income and expenditure rationally, and provide economic security to effectively deal with uncertain emergencies. In summary, as a disaster prevention and mitigation demonstration community, the Dafapu community excels in institutional and managerial resilience. Conversely, as depicted in Figure 10, the virtual community performed poorly in terms of institutional and managerial resilience with scores of 3.1 for B₆, 2.4 for B₇, 3.7 for B₈, and 2.7 for B₉, respectively, where half of the tertiary indexes score less than 3. Notably, the indicator for the number of science popularization demonstration schools for earthquake prevention and disaster reduction (C₄₃) only received a score of 1, while public financial strength (C₄₄), the ratio of emergency funds to fiscal expenditures (C₄₅), commercial insurance coverage rate (C₄₆), and epidemic monitoring and control situation (C₅₃) each receive a score of 2 points. The virtual community’s scores on B₆~B₉ are significantly lower than those achieved by the Dafapu community. In conclusion, the virtual community exhibits an institutional and managerial resilience index value of merely 2.975, which falls well below Dafapu’s score of 4.625.

Figure 10. The distribution of the scores without RDC of the secondary indexes of the Dafapu community and virtual community. (the red dotted line is the threshold warning level for assessing resilience)

Figure 10. The distribution of the scores without RDC of the secondary indexes of the Dafapu community and virtual community. (the red dotted line is the threshold warning level for assessing resilience)

Furthermore, Figure 11 presents and compares the distributions of the primary indexes’ scores for both the Dafapu community and the virtual community without considering RDC. It is worth noting that the resilience level value of the social and environmental resilience index (A₁) in the Dafapu community is 3.74, whereas it is 3.330 in the virtual community. Similarly, for the institutional and managerial resilience index (A₂), the resilience level value in the Dafapu community is 4.625 compared to 2.975 in the virtual community. Consequently, by utilizing Equation (1), we can derive overall resilience scores for both communities as follows: 4.1825 for Dafapu and 3.1525 for virtual communities with equal weights assigned to A₁ and A₂, respectively. This indicates that not only does the Dafapu community exhibit superior performance in terms of social, environmental, institutional, and managerial resilience when compared to its virtual counterpart, but it also highlights a potential drawback of solely comparing their respective indicators’ scores without considering their distinct demands. Therefore, subsequent discussions will further compare resilience assessments between these two communities while taking into account RDC.

4.3.2. The Case with Consideration of RDC

The distributions of the scores considering the RDC of the tertiary indexes are demonstrated in Figure 12, and the red line in the radar map represents a score of 3, which is the threshold value for assessing resilience in tertiary indexes. Firstly, the analysis results corresponding to the primary index of social and environmental resilience of both communities are shown in Figure 12a. After incorporating RDC, the Dafapu community for tertiary indicators C₂₄ and C₂₅ increased from 3 to 3.567, while its scores for C₂₇ and C₂₈ decreased from 4 to 3.155. Meanwhile, the virtual community’s score for C₂₅ decreased from 2 to 1.81, but its scores for C₂₇ and C₂₈ increased from 2 to 3.15. Additionally, there was a significant disparity between the two communities in their scores for C₂₇ and C₂₈, as depicted in Figure 9a; however, this gap is nearly eliminated after considering the RDC for transportation capacity. Additionally, after taking into account RDC, the number of tertiary indexes scoring below 3 in the virtual community has reduced from 9 to 7. The distributions of the scores considering RDC for the tertiary indexes corresponding to the institutional and management resilience primary index of both communities are presented in Figure 12b. For most indicators in C_36~43 and C_48~52, the scores of the Dafapu community achieved full marks before introducing RDC (except for indices C₃₉ and C₄₃, their values have also increased to 4.878 from 3 after introducing the RDC), indicating that it can adequately meet its demands and performs exceptionally well comparatively in these aspects. However, the resilience scores of the virtual community exhibit a further decline when considering its higher demand levels, resulting in an even wider disparity between the two communities on these indicators. This is evident from all tertiary indexes of C_36~43 and C_48~52. Upon considering RDC, several tertiary indexes that initially scored 3 have dropped to 2.935, including C_36~40, C₄₂, C₄₉, C₅₁, and C₅₂.

The distribution of the scores considering RDC of the secondary indexes of both communities is further demonstrated and compared in Figure 13. For the scores of indicators B₃, B₆, and B_8–9, the gap between the two communities has widened, while the gap in scores for B₄ has almost disappeared. Overall, after considering RDC, all secondary indicators in the Dafapu community except for B₁ score above 3. In the virtual community, the number of secondary indicators scoring below 3 has decreased from four to three. Regarding primary indicators in Figure 14, the score of the social and environmental resilience index (A₁) in the Dafapu community is 3.588, while that in the virtual community is 3.558, indicating that there has been a significant reduction in the original gap between these two communities on A₁. However, the gap on A₂ has become larger: scores of the institutional and managerial resilience index (A₂) for the Dafapu and virtual communities are 4.766 and 2.93, respectively. Consequently, by utilizing Equation (1), we can derive overall resilience scores for both communities by conducting a weighted combination with equal weights assigned to A₁ and A₂, respectively, resulting in a resilience level of 4.177 for the Dafapu community and 3.244 for the virtual community. The analysis result also indicates that the Dafapu community exhibits superior performance compared to its virtual counterpart when it comes to social, environmental, institutional, and managerial resilience as evident from their respective indicators’ scores. It should be noted that this analysis result becomes more reasonable to meet the actual situation in practice when considering the distinct demands of both communities.

The comparison of the multilevel RDC-adjusted scores between Dafapu and virtual communities indicates that the resilience of the Dafapu community is currently stable and satisfactory. However, there is still room for improvement in the future. Based on the scores of resilience evaluation indicators at all levels of Dafapu community, the following suggestions are put forward to enhance its resilience of the community: (1) enhance overall construction quality of community buildings; (2) further improve transportation capacity; (3) ensure future land layout of community avoids high-risk areas through safety risk assessment, while also strengthening ecological management and restoration in geological disaster-prone regions; (4) increase sewage collection rate and garbage recovery rate, strengthen service capacity of community medical and health institutions as well as community elderly care facilities, with a focus on enhancing infrastructure resilience; (5) strengthen investigation and regulation of construction land surrounding hazardous sources such as hazardous chemicals, natural gas pipelines, high-voltage power lines corridors; (6) intensify efforts to promote disaster prevention and mitigation science demonstration schools, innovate education models, and enhance the safety awareness throughout the entire community; (7) expand emergency refuge places coverage area per capita by increasing their number and improving service blind spots.

Figure 14. The distribution of the RDC-adjusted scores of the primary indexes of the Dafapu community and virtual community.

Figure 14. The distribution of the RDC-adjusted scores of the primary indexes of the Dafapu community and virtual community.

5. Conclusions

The resilience index systems and evaluation methods established by different urban communities generally need to be modified and improved according to variations in society, economy, culture, and environment across different countries or regions. Learning from the comprehensive literature and absorbing experience already available around the world, this paper has developed a novel community resilience index system and assessment method for evaluating community resilience using a Chinese community in Shenzhen city as an example, thereby demonstrating its practical application. Specifically, a Bayesian and analytic hierarchy process (BAHP)-based multilevel community resilience evaluation method has been developed to assess the ability of a community system to withstand disturbances and recover from them. The proposed method establishes a comprehensive assessment index system that can evaluate social and environmental resilience as well as institutional and managerial resilience at multiple levels. This system serves as a quantitative decision-making tool to elucidate the impact of various factors on community resilience. Furthermore, the “relative demand coefficient” (RDC) is proposed to compare different communities’ resilience by using Bayesian inference to determine its most probable value (MPV). The feasibility and effectiveness of the proposed method have been examined and validated through case studies and application demonstrations in both a real community and a simulated community. The significance of this study lies in gaining a deep understanding of key elements and influencing factors within existing research on community resilience index systems and evaluation methods while providing policymakers with a more effective tool for promoting improvements in community resilience. Through a comprehensive examination of community resilience, we can establish a more scientifically grounded foundation for urban planning and management while also providing valuable insights and experiences for the development of resilient systems in similar cities to some extent. Additionally, this study contributes meaningfully to the development of community resilience theory and its practical application.