A Study on Multi-Dimensional Analysis and Spatial Differentiation of the Resilience of Folk Cultural Spaces on Xiamen Island, China

Abstract

Amid rapid global urbanization, folk cultural spaces are facing a pronounced “resilience crisis.” Existing studies primarily emphasize material preservation while lacking a holistic assessment of cultural spaces. Using Xiamen Island as a case study, this research integrates GIS-based spatial analysis, questionnaire surveys, and statistical modeling to develop a resilience assessment framework for folk cultural spaces, encompassing four key dimensions: connectivity, modularity, diversity, and adaptability. The study systematically identifies spatial differentiation, formation mechanisms, and typological patterns of these spaces. The main findings are as follows: First, the resilience of folk cultural spaces on Xiamen Island exhibits a spatial pattern characterized by “dual-core leadership, corridor transition, and marginal vulnerability.” High-resilience areas are mainly concentrated in Siming Old Town and the Wuyuanwan district, representing two typical development trajectories—“organic evolution” and “planned intervention.” Second, the influencing mechanisms of each resilience dimension show pronounced spatial heterogeneity, reflecting the coupled effects of structural characteristics, social processes, and governance logics across different urban contexts. Third, three resilience zones are identified through K-means clustering, providing a typological basis for developing differentiated strategies for protection and renewal. This study provides theoretical insights and methodological references for the “living” preservation and adaptive governance of folk cultural spaces.

1. Introduction

Amid the tide of rapid global urbanization, urban spaces are undergoing unprecedented restructuring. At the national scale, for example, China’s urbanization rate increased from 36.2% in 2000 to 66.2% in 2023, with an average annual growth of about 1.3 percentage points, indicating a prolonged phase of rapid urbanization [1]. At the city scale, Xiamen City, where this study is located, has experienced particularly intense urban growth: the urban built-up area expanded from 38.5 km² in 1985 to 348.23 km² in 2017, and the urbanization rate of the permanent population reached 89.2% by 2019 [2]. As vital carriers of local knowledge, collective memory, and traditional cultural practices, folk cultural spaces are facing severe existential challenges [3]. These challenges are not limited to the physical disappearance of spaces but manifest as a complex “resilience crisis.” Under the dual pressures of external urban development and internal social restructuring, cultural functions shrink, community identity weakens, and adaptability declines, ultimately leading to the disruption of cultural inheritance chains [4]. In many cities, such “shrinking” cultural areas are closely associated with degraded residential environments, unattractive or low-end business storefronts, and concentrations of unemployed or underemployed populations, which jointly erode everyday vitality and perceived desirability and often precede functional re-districting and gentrification [4,5,6]. Traditional cultural heritage protection models have primarily emphasized the static “preservation” of material entities, while relatively neglecting the dynamic capacities for “adaptation” and “evolution” that cultural spaces, as complex socio-ecological systems, should possess [7,8]. In this context, resilience theory, which originated in ecology, provides a novel perspective for re-examining and evaluating the development trajectories of folk cultural spaces. Unlike robustness, which stresses the ability to withstand disturbances while maintaining an almost unchanged structure, resilience is no longer understood merely as the capacity to resist and “return to the original state,” but as the evolutionary ability of a system to maintain its core functions, reorganize, learn, and innovate in response to disturbances [9]. However, current studies on the resilience of folk cultural spaces have largely focused on conceptual interpretation, with an urgent need to develop practical and operable methodologies [10]. Against this backdrop, three core research questions arise:

RQ1: What is the spatial differentiation pattern of resilience in folk cultural spaces on Xiamen Island?

RQ2: How can a multidimensional and operable framework be constructed to quantitatively assess the resilience level of folk cultural spaces?

RQ3: What are the key dimensions that influence the formation and differentiation of resilience in folk cultural spaces, and what are their underlying mechanisms?

Addressing these questions will not only deepen the application of resilience theory in the field of cultural geography but also provide a scientific basis and practical pathway for the “living” preservation and sustainable development of cultural heritage in the process of urbanization.

The conceptual framework of resilience theory has undergone a profound evolution—from “engineering resilience,” to “ecological resilience,” and ultimately to “evolutionary resilience.” Initially, resilience theory—often framed as engineering resilience—emphasized a system’s capacity to resist external disturbances and return to a single equilibrium state [11], a view that is conceptually closer to robustness than to contemporary understandings of resilience. Subsequent work on ecological resilience recognized that systems may possess multiple stable states, and the core connotation of resilience has thus expanded to refer to a system’s ability to absorb disturbances and achieve orderly transitions between these states without losing its essential functions [12].

In recent years, resilience research has shifted from the notion of “returning to the original state” toward an evolutionary perspective characterized by “adaptation–learning–transformation.” Within this view, systems are conceptualized as complex adaptive systems whose resilience is expressed through continuous learning and reorganization amid disturbances, thereby facilitating pathway renewal and directional transformation [13]. In urban and regional studies, this evolutionary perspective underscores that adaptability is not merely a reactive response to disturbances but also the capacity to foster long-term self-organization and innovation [14].

The measurement framework has likewise evolved from a focus on “recovery time and magnitude” to a three-dimensional evaluation encompassing “persistence, adaptability, and transformation” [15]. Furthermore, studies in geographic information science and spatial analysis have incorporated the evolutionary perspective into spatiotemporal frameworks, emphasizing the temporal dynamics, spatial spillover effects, and network-structural characteristics of resilience [16].

Recent interdisciplinary reviews highlight that resilience is not a single attribute but is shaped by multiple mechanisms and trade-offs, emphasizing the refinement and transferability of both conceptual and measurement frameworks [17]. Furthermore, resilience is conceptualized as a continuous process through which a system sustains development by learning, adapting, and self-organizing amid ongoing change.

This theoretical paradigm has been widely applied in urban studies, giving rise to a burgeoning body of research on “urban resilience.” Meerow et al. [18] systematically reviewed the conceptualizations of urban resilience, emphasizing that it represents a dynamic process through which urban systems and their constituent communities maintain or rapidly restore essential functions when confronted with multiple disturbances, while simultaneously enhancing future capacities through adaptation, learning, and transformation.

Current research has primarily concentrated on “hard” domains such as climate change adaptation and disaster prevention, yet an increasing number of scholars have called for greater attention to “soft” resilience in the social, economic, and cultural spheres [19]. However, when translating resilience theory into specific evaluation indicators, many existing frameworks remain biased toward static structural measures while neglecting process-oriented indicators that capture the system’s dynamic adaptability. These process-oriented indicators are particularly sensitive to exogenous demographic and socio-economic dynamics, including out-migration, changes in birth and mortality rates, shifts in age structure, unemployment and income restructuring, and fluctuations in tourism and consumption patterns, all of which can profoundly reshape the vitality and long-term trajectories of cultural spaces. Therefore, this study incorporates quantified questionnaire data into the evaluation framework to develop a comprehensive resilience assessment system that balances process-oriented and structural indicators.

Folk cultural spaces are conceptualized in this study as a specific type of traditional cultural space, referring to sites where traditional cultural activities are practiced in accordance with local customs at specific times and places [20]. They serve as vital carriers of cultural continuity and social life, embodying rich historical memories and collective cultural identity. However, with the proliferation of the Internet and smartphones, lifestyles have undergone a rapid transformation, leading to the marginalization of local traditional cultures and customary practices [21]. These phenomena have accelerated the decline in folk culture, causing the cultural spaces that sustain these practices to lose their vitality.

Research on cultural spaces has undergone a paradigmatic shift from a focus on material entities to an emphasis on social construction [22,23]. Influenced by the concept of material heritage preservation, early studies primarily examined architectural forms, historical value, and spatial distribution [24,25]. Subsequently, the research perspective shifted toward understanding cultural spaces as venues for social processes and the production of meaning [26]. Cultural spaces are not merely physical containers but also active stages for community identity formation, the continuity of cultural practices, and everyday social interaction [27,28].

In the Chinese context, folk cultural spaces—particularly ancestral halls and folk belief venues—are widely regarded as the social “adhesive” that binds local communities together and sustains collective cultural identity. Ancestral halls function as clan-based ritual buildings used for lineage worship, communal decision-making, and the organization of festivals, while folk belief venues encompass temples, shrines, and altars associated with locally venerated deities and ritual practices. Wu [29] noted that these spaces constitute the core arenas for the preservation and transmission of intangible cultural heritage.

In recent years, scholars have increasingly focused on how folk and historic cultural spaces are transformed under the broader dynamics of urbanization, heritage-led regeneration, and tourism development. Representative cases from China and other regions show that renewal strategies often strengthen the physical fabric and economic functions of cultural spaces, while simultaneously driving commercialization, functional restructuring, and social reconfiguration [30,31,32]. Typical examples are summarized in

Table 1. Representative cases of cultural spaces under urban transformation.

City/Region	Type of Cultural Space	Main Transformation Drivers	Key Impacts on Cultural Spaces	Ref.
Xi’an and Shanghai, China	Historic districts, heritage streets	Heritage-led renewal; tourism development	Improved physical fabric and tourism economy; increased commercialization and partial displacement of everyday uses	[30]
Jinan, China (Baihuazhou)	Historic–cultural block	Spatial regeneration; cultural tourism	Living spaces converted into display- and consumption-oriented spaces; risk of homogenization of cultural practices	[31]
Fez, Morocco (World Heritage Medina)	Historic medina, traditional markets	Heritage tourism; modernization	Economic revitalization and upgraded infrastructure; commercialization and reconfiguration of spatial–social structures	[32]

.

Beyond the Asian and North African cases, Caribbean cruise-oriented duty-free shopping zones provide another instructive configuration of cultural spaces under intense tourism pressure [33,34]. In port cities such as Charlotte Amalie (St. Thomas), Philipsburg (Sint Maarten), St. John’s (Antigua), and Old San Juan (Puerto Rico), historic waterfronts and colonial cores have been partially rebranded as duty-free retail corridors for cruise passengers, with heritage architecture serving as a scenic backdrop for high-turnover shopping and entertainment [33]. Critical studies show that this development model often distorts the original social uses of historic streetscapes, concentrates external capital, and amplifies socio-spatial inequalities, a process described as a shift from “heritage” to “feritage” in Caribbean cruise destinations [33]. At the same time, regional and international policy initiatives—such as the Caribbean Sustainable Tourism Policy Framework 2020 and post-pandemic diversification strategies—seek to enhance the long-term resilience of these port-city cultural spaces by linking visitor-oriented commercialization with the preservation of urban form, local knowledge, community participation and economic diversification [35,36].

Despite the growing body of research, most existing studies remain confined to qualitative descriptions or isolated quantitative analyses, lacking a systematic framework that integrates the material attributes of space with its social and humanistic processes to assess the overall health and developmental potential of cultural spaces.

Building upon these gaps, this study takes Xiamen Island as a case study and adopts cultural space resilience as the core indicator for evaluating the vitality and sustainability of folk cultural spaces. Guided by resilience theory and the intrinsic attributes of cultural spaces, this study identifies a set of factors closely associated with the formation of resilience. Using GIS-based spatial analysis, questionnaire surveys, and the Analytic Hierarchy Process (AHP), a comprehensive quantitative evaluation model is constructed to reveal the spatial differentiation patterns of each resilience dimension and their underlying driving mechanisms.

Subsequently, cluster analysis is performed based on the comprehensive resilience index to identify differentiated levels and spatial typologies of resilience. Finally, targeted protection and planning strategies are proposed for areas with different resilience typologies. This study provides theoretical and methodological support for developing a scientific diagnostic framework for folk cultural spaces, thereby promoting their “living” preservation and adaptive development.

2. Materials and Methods

2.1. Study Area

This study focuses on Xiamen Island, located within Xiamen City of Fujian Province, southeastern China. The island serves as a major urban center, seaport, and tourist destination along China’s southeastern coast. As a Special Economic Zone, Xiamen Island has experienced rapid and high-intensity urbanization, characterized by substantial spatial restructuring and frequent alternation between old and new urban functions. Against this backdrop, traditional folk cultural spaces face an evident “squeezing–adapting–transforming” dynamic that highlights their struggle for survival, but also reveals their potential resilience attributes in terms of persistence, adaptation, and functional transformation under sustained disturbance.

Meanwhile, Xiamen Island constitutes the cultural core of the Minnan region, featuring a high density and diversity of folk cultural spaces such as ancestral halls, temples, and ancestral residences. These spaces serve as living carriers of local historical memory, community identity, and traditional cultural practices. From a resilience perspective, the study area displays a set of characteristic resilience attributes. First, long-established lineage and belief networks anchored in ancestral halls and temples, together with fine-grained street–alley fabrics in historic neighborhoods, provide strong social cohesion, everyday usage, and spatial continuity, which enhance the persistence and adaptive capacity of folk cultural spaces. Second, the island’s position as a Special Economic Zone and major tourist destination has stimulated continuous functional renewal, adaptive reuse of historic buildings, and investment in public space and infrastructure, creating opportunities for the transformation and innovation of cultural spaces rather than their simple disappearance. These resilience attributes were identified through a combination of document and policy review (including local chronicles, heritage and religious site lists, and planning documents), GIS-based mapping of registered folk cultural sites (Figure 1), and reconnaissance fieldwork conducted during the preparatory phase of this study. Within this context, the spatial pattern of Xiamen Island exhibits the typical characteristics of “collage” and “coexistence.” On the one hand, Siming District in the southwest preserves a well-integrated network of traditional streets, alleys, and densely distributed cultural spaces; on the other hand, Huli District in the northeast and the eastern coastal areas represent modern urban zones characterized by industrial parks and high-intensity development. This spatial heterogeneity provides an ideal comparative context for examining the differentiation patterns of resilience among folk cultural spaces across distinct urban environments. Therefore, this study selects the folk cultural spaces within Xiamen Island as its primary objects of analysis.

2.2. Research Framework and Methodological Steps

To evaluate the resilience levels of folk cultural spaces on Xiamen Island, this study integrates the island’s distinctive Minnan cultural characteristics and resilience attributes to construct an assessment framework centered on four key dimensions: connectivity, modularity, diversity, and adaptability (Figure 2). The research process comprises five major stages: (1) data collection and preprocessing; (2) indicator quantification; (3) weight determination and integration; (4) spatial correlation analysis; (5) resilience zoning and strategic formulation.

First, web-based map extraction techniques were employed to identify the spatial distribution of folk cultural spaces on Xiamen Island. Point of Interest (POI) data were collected from major online map platforms (e.g., Gaode Maps) using keywords such as “ancestral hall,” “temple,” “ancestral residence,” and related folk belief sites, which were then exported as candidate points. These POI data were subsequently cleaned to remove duplicates and obvious misclassifications before being overlaid onto high-resolution base maps in ArcGIS Pro 3.4. Based on the web-derived candidate points, reconnaissance field surveys were conducted to verify the actual existence, current usage, and accessibility of each site, excluding those that had been demolished, completely abandoned, or converted to non-cultural uses. To ensure spatial and functional representativeness, Xiamen Island was stratified into contrasting urban environmental types (traditional core areas, mixed residential-commercial areas, and modern high-intensity development zones). Folk cultural spaces were then sampled across different categories, surrounding land-use intensities, and levels of accessibility to public transport and commercial services. A total of 288 valid sample points were retained.

For each sample point, circular buffers with radii of 500 m were delineated. OpenStreetMap road network data and POI datasets of bus and subway stations were used to calculate ‘telemetric’ indicators of street connectivity, service density, and public transport accessibility. These indicators (e.g., density of road intersections, number of commercial POIs, and number of bus/subway stops within the buffers) served as proxies for local pedestrian flow, business composition, and everyday functional intensity in subsequent analyses.

At each sample point, a structured questionnaire survey was administered to obtain process-oriented indicators of folk cultural space resilience. The instrument utilized a five-point Likert scale and covered four resilience dimensions: Connectivity, Modularity, Diversity, and Adaptability (Table S1). The final questionnaire consisted of 12 core assessment items organized into 3 main sections: (1) Basic respondent information; (2) Perceptual assessment across the four resilience dimensions; (3) Open-ended feedback. All items were adapted from established urban/community resilience and place-attachment scales and were modified to fit the context of Minnan folk cultural spaces [37,38]. The questionnaire was pre-tested with 30 respondents in June 2024, and the wording and item sequence were refined based on their feedback.

The formal survey was conducted between July and August 2024 at the 288 sample points. Respondents were adult users of the cultural spaces (local residents, worshippers, and repeat visitors) approached using a convenience sampling strategy during periods of typical activity (e.g., weekends, market days, festival-related occasions). The questionnaires were primarily self-administered, with researchers providing clarification when necessary. Each questionnaire took approximately 8–10 min to complete. On average, about 4–5 valid responses were collected per sample point (1272 valid questionnaires in total). The survey was administered by trained postgraduate students from Fuzhou University. No monetary incentives were provided; respondents were informed about the academic purpose of the study, assured of anonymity, and provided verbal consent prior to participation.

To address the potential issue of common method bias arising from the use of a single self-report instrument, Harman’s single-factor test was conducted by loading all measurement items into an unrotated exploratory factor analysis. An unrotated principal component analysis was performed in SPSS 27.0.1 to examine the variance explained by the first factor. The first unrotated factor accounted for 28.5% of the total variance, which is below the commonly used threshold of 40%, indicating that common method bias is unlikely to threaten the validity of the results.

Second, specific quantification methods were applied for each indicator (

Table 2. Table of Resilience Assessment Indicator System.

Dimension	Specific Indicators	Data Resource	Calculation Method
Connectivity	Network Accessibility	OSM Road Network, Bus/Metro Station POIs	Space Syntax Network Analysis (sDNA)
	Ease of use	Questionnaire survey	Average Score of 5-Point Scale
Modularity	Spatial agglomeration degree	Field Visits, Map POI	Kernel Density Estimation (K = 1000 m)
	Community participation	Questionnaire survey	Average Score of 5-Point Scale
Diversity	Functional Diversity	Field Visits	Functional Type Count
	Spatial Business Format Richness	Map POI	Shannon Diversity Index
Adaptability	Environmental compatibility	Field Visits	Expert Scoring Based on AHP
	Spatial disturbance resistance	Questionnaire survey	Average Score of 5-Point Scale

). Multimodal transportation accessibility was calculated using spatial network analysis; spatial modularity was measured through kernel density estimation; and environmental adaptability was assessed through expert scoring based on the Analytic Hierarchy Process (AHP). In parallel, the questionnaire data underwent reliability and validity testing (Table S2), followed by coding and aggregation to quantify indicators such as accessibility, ease of use, and community participation.

Third, all quantified indicators were standardized. The entropy weight method was employed to objectively assign weights to the four dimension-level indices. A weighted summation model was used to integrate and calculate the comprehensive resilience index for each cultural space, after which the natural breaks method was applied to categorize the results into three resilience levels—high, medium, and low. Sensitivity analyses were conducted using alternative weighting and classification methods.

Subsequently, Global Moran’s I was applied to analyze the spatial autocorrelation of the comprehensive resilience index and to identify its overall spatial clustering characteristics.

Finally, based on the comprehensive resilience index and its spatial distribution, the K-means clustering algorithm was employed to delineate differentiated resilience management zones. This approach provides a scientific basis for formulating targeted protection and enhancement strategies for folk cultural spaces with varying levels of resilience.

2.3. Data Sources

This study utilizes multi-source datasets to construct a resilience assessment framework for folk cultural spaces on Xiamen Island. The base road network data were obtained from OpenStreetMap, including information on multi-level road hierarchies; public transportation and subway station data were integrated from official Xiamen public transport sources and open mapping platforms; the core research objects—288 folk cultural spaces—were identified through a combination of Gaode Map POI extraction, the literature review, and field investigations; and data for other indicators were collected through field surveys and questionnaire distribution in August 2024. Among them, the respondents were local residents living within 300 m of the sites. The questionnaires were distributed on-site at and around the 288 identified folk cultural spaces between July and August 2024. A total of 1500 questionnaires were distributed in the study, with 1272 valid ones recovered, resulting in an effective recovery rate of 84.8%. For sites with multiple responses, we first verified the completeness of the questionnaires, then aggregated the data by calculating the average score of each item, and finally incorporated the aggregated results as questionnaire indicators into the resilience assessment framework. All spatial datasets were uniformly projected using the WGS 1984 UTM Zone 50N coordinate system to ensure accuracy in spatial measurement and analysis. This coordinate system is well suited to the study area, effectively minimizing projection distortion, and all data underwent quality-control procedures, including geometric correction and topological inspection.

2.4. Methods

2.4.1. Space Syntax Network Analysis

Spatial Design Network Analysis (sDNA) is a graph theory-based spatial analysis method. It can quantify the structural attributes of urban spatial networks and is often used to analyze the interaction between urban form and human activities. To identify the spatial accessibility characteristics of folk cultural spaces, with reference to urban life circle theory and existing research, the analysis employed three search radii for multi-scale measurement: 800 m (walking scale), 1200 m (community life circle scale), and 5000 m (island-wide scale). After correlating with the “ease of use” indicator from the questionnaire, the betweenness centrality at the 1200 m radius showed the strongest correlation. Therefore, this study adopts the betweenness centrality (radius: 1200 m) in the sDNA model to represent the network accessibility of each folk cultural space. Its calculation formula is as follows:

$$Betweenness(i)={\sum }_{s\ne i\ne t}{\displaystyle \frac{{\sigma }_{st}\left(i\right)}{{\sigma }_{st}}}$$

(1)

Among them, ${\sigma }_{st}$ represents the total number of shortest paths from the start node s to the end node t, and ${\sigma }_{st}\left(i\right)$ represents the number of shortest paths passing through node i.

To identify the spatial scale that best reflects the actual user experience of folk cultural spaces, this study calculated betweenness centrality values at three search radii: 800 m (walking scale), 1200 m (community life circle scale), and 5000 m (island-wide scale). These values were then correlated with the “ease of use” indicator obtained from the questionnaire surveys using Pearson correlation analysis. The results (

Table 3. Correlation Analysis Between Betweenness Centrality at Different Scales and Questionnaire Scores for “Ease of Use”.

Search Radius	Scale Concept	Correlation with “Ease of Use” (r)	p-Value
800 m	Walking Scale	0.41	p < 0.01
1200 m	Community Life Circle Scale	0.52	p < 0.01
5000 m	Island-wide Scale	0.28	p < 0.05

) demonstrate that the correlation between betweenness centrality and ease of use was strongest at the 1200 m radius (r = 0.52, p < 0.01). This finding aligns with the theoretical concept of the urban community life circle, indicating that the 1200 m radius more effectively captures the transportation convenience experienced by residents in their daily visits to folk cultural spaces. Consequently, the betweenness centrality at the 1200 m search radius was selected as the core metric for the “connectivity” dimension in this study.

2.4.2. Kernel Density Estimation

Kernel Density Estimation (KDE) can effectively reveal the spatial agglomeration characteristics and distribution patterns of geographic elements. It is often used to analyze the spatial density distribution of point elements and identify hot spot areas. This study introduces Kernel Density Estimation (KDE) to quantify the spatial distribution pattern of folk cultural space points, thereby revealing the inherent resilience characteristics of their spatial organization. Specifically, a kernel density value is calculated for each cultural space point; this value reflects the agglomeration degree of folk cultural space points within a certain search radius centered on the point in question. High kernel density values identify core areas with highly concentrated cultural spaces, while low kernel density values correspond to peripheral areas with sparsely distributed cultural spaces. Its mathematical expression is as follows:

$$f_n\left(x\right)={\displaystyle \frac{1}{nh}}\sum _{t=1}^{n}k\left({\displaystyle \frac{X-X_1}{H}}\right)$$

(2)

Among them, $k\left({\displaystyle \frac{X-X_1}{H}}\right)$ represents the kernel density function, where H denotes the bandwidth (threshold), and n stands for the number of points within the bandwidth (threshold range).

2.4.3. Analytic Hierarchy Process

The Analytic Hierarchy Process (AHP) is a systematic multi-criteria decision-making method proposed by Thomas L. Saaty in the 1970s [39]. It constructs a hierarchical structure to decompose complex decision problems into multiple levels, including objectives, criteria, and alternatives. Through expert judgment and pairwise comparison, AHP calculates the relative weights of multiple factors, thereby providing a quantitative basis for structured decision-making. In this study, 15 domain experts were invited to evaluate the relationships between folk cultural spaces and their surrounding built environments using the AHP approach. This evaluation determined the relative importance of different spatial relationships in terms of environmental compatibility, thereby identifying the corresponding weight coefficients for each type (Table S3). In this study, AHP is applied only within the “environmental compatibility” indicator to weight different types of spatial relationships between folk cultural spaces and surrounding buildings. The resulting composite environmental compatibility score is then treated as a single indicator within the adaptability dimension.

2.4.4. Entropy Weight Method

The Entropy Weight Method (EWM) calculates the entropy value of each indicator based on the impact of changes in each indicator’s values on the whole, and then determines the weight. When dealing with multi-indicator weighting issues, this method can eliminate result deviations caused by subjective human assignment, avoid the influence of subjective factors, and improve the objectivity and accuracy of evaluation results.

In the second step, after obtaining the four dimension-level indices (connectivity, modularity, diversity, and adaptability), the Entropy Weight Method (EWM) is applied at the dimension level to determine the relative contribution of each dimension to the overall resilience. In this study on the resilience of folk cultural spaces, the Entropy Weight Method (EWM) was employed to determine the weights of four dimensions—connectivity, modularity, diversity, and adaptability. This approach ensures that the comprehensive resilience index is objectively driven by the intrinsic properties of the data, revealing which dimension contributes most to the overall resilience of folk cultural spaces within the specific context of Xiamen Island. Consequently, the evaluation results are more scientific, reliable, and convincing.

First, the original indicators of connectivity, modularity, diversity, and adaptability are processed using Min-Max normalization:

$$x^{\prime }={\displaystyle \frac{x-min}{max-min}}$$

(3)

Subsequently, the information entropy is calculated in accordance with the Entropy Weight Method (EWM):

$$H_j=-{\displaystyle \frac{\sum _{i=1}^n(p_{ij}·\ln p_{ij})}{\ln n}}(i=\mathrm{1,2}...,n;j=\mathrm{1,2}...,m)$$

(4)

Among them, ${\mathrm{H}}_{\mathrm{j}}$ represents the information entropy of the j-th indicator, $p_{ij}$ denotes the proportion of the standardized value of the i-th sample under the j-th indicator (i.e., $p_{ij}={\displaystyle \frac{r_{ij}}{{\sum }_{i-1}^n{r}_{ij}}}$), and n stands for the number of samples.

Calculate the information entropy, redundancy and determine the weights:

$$d_j=1-H_j$$

(5)

$$w_j={\displaystyle \frac{d_j}{{\sum }_{i=1}^m{d}_j}}$$

(6)

Among them, $d_j$ represents the calculated information entropy redundancy, and $w_j$ represents the weight; finally, the comprehensive score is calculated:

$$S_i=\sum _{j=1}^m(w_j·r_{ij})$$

(7)

2.4.5. K-Means Clustering Analysis

K-means clustering analysis is an unsupervised machine learning algorithm widely applied in data mining and pattern recognition. It aims to automatically partition an unlabeled dataset into a predefined number of mutually exclusive clusters based on the similarity among data objects. Through an iterative optimization process, the algorithm calculates the Euclidean distance between each data point and the cluster centroids, assigns each point to the nearest cluster, and subsequently updates the centroids. This process continues until the cluster assignments converge, thereby achieving the dual objectives of minimizing intra-cluster variance and maximizing inter-cluster separation.

To identify differentiated management patterns of resilience among folk cultural spaces on Xiamen Island, this study employed the K-means clustering algorithm. Based on the comprehensive resilience index and the scores of each resilience dimension for each cultural space, the study area was divided into distinct resilience management zones exhibiting significant characteristic differences. The within-cluster sum of squared errors (denoted as E) is commonly used as a standardized evaluation function [40], and its calculation is expressed as follows:

$$\mathrm{E}={\sum }_{\mathrm{i}=1}^{\mathrm{k}}{\sum }_{\mathrm{x}\in {\mathrm{C}}_{\mathrm{i}}}{\left|\mathrm{x}-{\mathsf{\mu }}_{\mathrm{i}}\right|}^2$$

(8)

Among them, x represents the value of the sample object, and μ_i represents the mean value of category C_i. Its calculation method is as follows:

$${\mu }_i={\displaystyle \frac{1}{\left|C_i\right|}}{\sum }_{x\in C_i}x$$

(9)

2.4.6. Spatial Autocorrelation

The spatial autocorrelation analysis method can effectively reveal the spatial dependence and heterogeneity of geographic attribute values. It is commonly used to determine whether geographical phenomena exhibit significant spatial clustering, dispersion, or random distribution patterns. In this study, spatial autocorrelation analysis was employed to quantify the spatial correlation structure of the resilience index of folk cultural spaces, thereby revealing the underlying patterns and driving mechanisms of its spatial differentiation. Specifically, spatial autocorrelation statistics were calculated at two scales—global and local—to systematically identify and diagnose the spatial correlation patterns of the resilience index at both overall and local levels. The core statistic, Global Moran’s I, is expressed as follows:

$$\mathrm{I}={\displaystyle \frac{\mathrm{n}}{{\mathrm{S}}_0}}\times {\displaystyle \frac{\sum _{\mathrm{i}=1}^{\mathrm{n}}\sum _{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{w}}_{\mathrm{i}\mathrm{j}}({\mathrm{x}}_{\mathrm{i}}-\overline{\mathrm{x}})({\mathrm{x}}_{\mathrm{j}}-\overline{\mathrm{x}})}{\sum _{\mathrm{i}=1}^{\mathrm{n}}{({\mathrm{x}}_{\mathrm{i}}-\overline{\mathrm{x}})}^2}}$$

(10)

Among them, n is the total number of sample points of folk cultural spaces in the study area; ${\mathrm{x}}_{\mathrm{i}}$ and ${\mathrm{x}}_{\mathrm{j}}$ are the resilience indices of sample points i and j, respectively; $\overline{\mathrm{x}}$ is the average value of the resilience indices of all sample points; ${\mathrm{w}}_{\mathrm{i}\mathrm{j}}$ is the spatial weight matrix, which is used to measure the spatial adjacency relationship between sample points i and j; ${\mathrm{S}}_0$ is the aggregated value of all spatial weights, i.e., ${\mathrm{S}}_0$ = $\sum _{\mathrm{i}=1}^{\mathrm{n}}\sum _{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{w}}_{\mathrm{i}\mathrm{j}}$.

This study employed a distance threshold spatial weights matrix to define the spatial adjacency relationships among sample points. Based on the spatial scale of the study area and the distribution characteristics of the sample points, the distance threshold was set to 1000 m after several tests. This means that two folk cultural space sample points were considered spatial neighbors (weight = 1) if the Euclidean distance between them was less than or equal to 1000 m; otherwise, they were considered non-neighbors (weight = 0).

3. Results

3.1. Overall Distribution Characteristics of Folk Cultural Spaces

The 288 identified folk cultural spaces on Xiamen Island exhibit a markedly uneven spatial distribution (Figure 3). Kernel Density Estimation (KDE) analysis reveals a “multi-core clustered” spatial pattern, with cultural spaces primarily concentrated in the historical urban areas of Siming District—such as the Zhongshan Road area, the waterfront opposite Gulangyu Island—and several traditional villages in Huli District. The global spatial autocorrelation analysis of the comprehensive resilience index of folk cultural spaces yielded a Global Moran’s I value of 0.799. This result was assessed for significance using a permutation test with 999 permutations, which produced a p-value of less than 0.001. This strongly rejects the null hypothesis of spatial randomness, indicating that the resilience of folk cultural spaces on Xiamen Island exhibits a highly significant spatial clustering pattern at the global level. The hot–cold spot analysis map (Figure 4), derived from local spatial autocorrelation, identifies two prominent high-resilience cores located in the old urban area of Siming District and the Wuyuanwan area, respectively. In contrast, low-resilience cold spots are widely dispersed across the eastern coastal new town. Between the peripheries of the hot-spot cores and cold-spot zones lie extensive “insignificant” areas that form spatial transition zones between high- and low-resilience regions.

3.2. Spatial Differentiation Characteristics of Resilience Across Various Dimensions

The connectivity of folk cultural spaces on Xiamen Island exhibits a clustered spatial differentiation pattern characterized by “high connectivity in two cores and low connectivity at the periphery” (Figure 5a). Areas with high connectivity are mainly concentrated in the old urban core of Siming District and the Wuyuanwan area. Questionnaire feedback from respondents in the Siming Old Town indicates that the average score for ease of use exceeds 4 points. In contrast, the sDNA model results for the Wuyuanwan area show relatively high betweenness centrality values, whereas the corresponding questionnaire scores for ease of use are comparatively lower.

The spatial differentiation of modularity is characterized by high-value areas densely concentrated around Wuyuanwan in the northeast, where kernel density values exceed those of the Siming Old Town in the southwest (Figure 5b). The cultural spaces in the Wuyuanwan area display an agglomerated distribution pattern, forming a distinctive high-density cultural enclave within the modern built environment.

The diversity dimension also exhibits a spatial pattern of “high diversity in two cores and low diversity at the periphery,” with high-diversity zones primarily concentrated in the Siming Old Town core and the Wuyuanwan area (Figure 5c). The Siming Old Town demonstrates both a high diversity of cultural space functions and a rich mix of surrounding business types, whereas the diversity in Wuyuanwan is mainly reflected in the variety of adjacent commercial formats.

High-value zones in the adaptability dimension are primarily concentrated in the Siming Old Town core (Figure 5d). This area records high questionnaire scores for both environmental compatibility and resistance to spatial disturbance. The adaptability level in Wuyuanwan is moderate, whereas other areas generally exhibit low adaptability.

3.3. Spatial Pattern and Classification of Comprehensive Resilience

The spatial distribution of the comprehensive resilience of folk cultural spaces on Xiamen Island is not random. Instead, it reflects the city’s spatiotemporal development trajectory—from the southwest to the northeast, and from old urban cores to newly developed areas—exhibiting a distinct spatial configuration characterized by “dual-core leadership, corridor transition, and marginal vulnerability.” This configuration results from the combined effects of the four resilience dimensions—connectivity, modularity, diversity, and adaptability—shaped by the island’s specific historical context and urban planning background (Figure 6).

To evaluate the robustness of the comprehensive resilience assessment model developed in this study, a sensitivity analysis was conducted (Table S4). By sequentially adjusting the weights of each dimensional indicator and observing variations in the comprehensive resilience index and its classification, the following results were obtained: across all eight weight-adjustment scenarios, the average rate of change in the comprehensive resilience index remained below 1.3%, which is far lower than the commonly accepted 5% threshold. This finding indicates that the model output is largely insensitive to variations in input weights, demonstrating a high level of stability. The variations in weighting exert minimal influence on the classification of resilience levels among sample points. Therefore, the study categorizes the study area into high-, medium-, and low-resilience zones using the natural breaks method.

3.4. Analysis of K-Means Clustering Results

To further examine and validate the characteristics of resilience patterns at different levels, this study performed K-means clustering analysis on 288 sample points of folk cultural spaces using their standardized indicator scores across multiple dimensions. The purpose of this analysis was to obtain a more comprehensive understanding of the spatial differentiation of folk cultural spaces on Xiamen Island. After multiple clustering iterations and validation using the silhouette coefficient and the elbow method, the optimal number of clusters was determined to be three. The silhouette coefficient is highest at K = 2 (0.3936), but at K = 3 it reaches 0.3517 with a more balanced distribution of cluster sizes. Meanwhile, the elbow method shows that the decreasing trend of within-cluster sum of squares slows down markedly after K = 3. Therefore, selecting K = 3 provides a good balance between clustering performance and practical interpretability (Figure 7), and the three clusters represent distinct combinations of resilience characteristics (Figure 8). To further clarify the internal characteristics of each resilience type, we calculated the mean standardized scores of the four dimensions and the composite resilience index for each cluster (

Table 4. Mean standardized scores of resilience dimensions for the three K-means clusters of folk cultural spaces on Xiamen Island. (Note: Scores are standardized (Z-scores). Cluster C1 has the highest mean composite resilience index, Cluster C2 an intermediate level, and Cluster C3 the lowest level of resilience among the three types).

Cluster	Resilience Type	Connectivity (Z)	Modularity (Z)	Diversity (Z)	Adaptability (Z)	Comprehensive Resilience (Z)
C1	Historic-core robust–adaptive type	0.85	0.82	0.88	0.90	0.86
C2	Transitional mixed type	0.65	0.70	0.45	0.4	0.55
C3	Marginal fragile fringe type	0.30	0.25	0.20	0.15	0.23

) and assigned interpretable labels to the three types. As shown in Table 4, Cluster C1 records the highest mean composite resilience index (0.86), Cluster C2 an intermediate value (0.55), and Cluster C3 the lowest value (0.23), corresponding to the historic-core robust–adaptive type, the transitional mixed type and the marginal fragile type, respectively.

The results reveal that areas with high comprehensive resilience are mainly distributed in the old urban core of Siming District and the Wuyuanwan area. These spaces perform strongly across all four dimensions—connectivity, modularity, diversity, and adaptability—demonstrating high levels of system coordination and structural integrity. Medium-resilience areas are primarily found along the periphery of the old urban zone and in several well-preserved traditional villages. Although these spaces exhibit sound modularity and connectivity, they remain weaker in functional diversity and environmental adaptability, indicating a solid structural foundation but limited functional and environmental integration. In contrast, low-resilience areas are predominantly located in the eastern coastal new town, the northern industrial belt, and the central transitional zone. Characterized by dispersed spatial distribution, weak structure, and low performance across all dimensions, these areas face an elevated risk of cultural function degradation and loss of vitality.

4. Discussion

This study reveals a spatial pattern of folk cultural resilience on Xiamen Island characterized by a structure of “dual-core leadership, corridor transition and marginal vulnerability”, which closely aligns with the island’s broader urban development trajectory. As the core enclave of traditional Minnan culture, Siming Old Town exhibits a resilience profile that combines structural robustness with process-oriented adaptability. On the one hand, its historically evolved “small-scale, high-density” street–alley morphology provides a robust spatial framework that supports frequent everyday interactions and fine-grained spatial continuity, illustrating a synergistic stability among spatial morphology, functional structure and social networks [41,42]. On the other hand, the clustering results indicate that Siming’s folk cultural spaces also score relatively high on adaptability and transformability, reflected in the coexistence of long-standing rituals and festivals with emerging cultural, tourism and service-oriented functions. In this sense, they are not only robust in the narrow sense of withstanding disturbance, but also resilient in the broader evolutionary sense adopted in this study, namely being able to reorganize, learn and innovate while maintaining core cultural functions. In contrast, the relatively high resilience of the Wuyuanwan area depends primarily on a modern, planning-led pathway of “design intervention and functional implantation”. During its renewal process, morphological identification, functional reconfiguration and environmental improvements have been applied to preserve key cultural signifiers and enhance environmental resilience [43], while integrating folk cultural spaces into new waterfront, leisure and tourism systems. Nevertheless, excessive commercialisation may erode cultural authenticity and undermine long-term resilience if visitor-oriented consumption crowds out everyday community uses. Taken together, these two pathways—a historic-core robust–adaptive pathway and a planning-led functional innovation pathway—form the “dual engines” sustaining Xiamen Island’s cultural resilience and offer a comparative model for understanding the adaptive trajectories of cultural spaces in both traditional and modern urban contexts.

Furthermore, the influencing mechanisms of each resilience dimension display pronounced spatial heterogeneity. In Siming Old Town, connectivity manifests as pedestrian-friendly accessibility derived from organically evolved street networks, whereas in Wuyuanwan it primarily depends on a planned and predesigned road system. In the southwestern historic quarter, modularity arises from social cohesion grounded in kinship and territorial ties, while in the northeastern sector it manifests as the “island-like persistence” of cultural spaces amid rapid urbanization. Diversity in the historical urban core results from organically overlapping and redundant functions, whereas in new urban zones it appears as a hybrid pattern driven by planning guidance and market dynamics. Adaptability in Siming Old Town is reflected in the seamless integration of spatial form and community life, whereas newly developed areas face integration challenges within the modern built environment. This multidimensional and heterogeneous pattern demonstrates that the resilience of folk cultural spaces is not a mere summation of individual attributes but a coupled outcome of structural characteristics, social processes, and governance logics within specific spatial contexts [44,45].

The three resilience types identified by the K-means clustering provide a spatial basis for designing differentiated protection and renewal strategies. For the historic-core robust–adaptive type (Cluster 1, with the highest comprehensive resilience scores), planning should focus on maintaining key spatial and social structures while enhancing functional performance, and at the same time preventing the erosion of intrinsic cultural value through over-commercialisation or community displacement. In practical terms, this may include setting upper limits on the proportion of purely tourist-oriented businesses, introducing rent-stabilization or subsidy schemes for long-term residents, and establishing community co-governance mechanisms for the management of key cultural spaces [45]. For the transitional mixed type (Cluster 2, with intermediate resilience levels), the priority is to address shortcomings in functional integration and spatial cohesion by strengthening structural support and network connectivity, thereby mitigating risks of spatial fragmentation and social disintegration during urban renewal. Possible interventions include enhancing functional diversity through small-scale cultural programming (e.g., weekend markets and folk performances) and repurposing underused buildings into multifunctional community–cultural spaces [44]. For the marginal fragile type (Cluster 3, with the lowest resilience scores), policy should prioritize the revitalisation of core cultural nodes, the reconstruction of community networks and micro-scale spatial interventions in order to gradually restore basic cultural functions and sense of place. This can be operationalised through low-cost micro-regeneration strategies, such as improving pedestrian connections to isolated cultural sites, creating small public plazas around temples or ancestral halls, and organizing targeted cultural events that reconnect residents with these spaces [46]. This typology-based, zoned governance approach aligns with the planning principle of “adapting measures to local conditions” and provides a reference framework for cultural-space conservation in other urban contexts.

Although this study establishes a relatively systematic framework for resilience assessment and conducts empirical verification on Xiamen Island, several limitations persist. First, the data are primarily derived from cross-sectional spatial datasets and questionnaire surveys, limiting the ability to capture the dynamic evolution of cultural-space resilience throughout the urbanization process. Second, although the assessment system encompasses four key dimensions, it remains constrained in representing the “soft” aspects of resilience, such as cultural practices, symbolic meanings, and intergenerational transmission mechanisms. Furthermore, as this study focuses on Xiamen Island—a representative but geographically limited case—the generalization of its findings to other urban contexts should be made with caution.

Future research can be expanded in several directions: (1) incorporating time-series data with historical maps and longitudinal surveys to reveal long-term evolutionary trajectories and key transitional mechanisms of cultural-space resilience; (2) broadening the assessment framework by integrating process-oriented indicators such as cultural transmission capacity and governance participation to establish a more comprehensive diagnostic model of resilience; (3) conducting comparative studies across cities and regions to identify resilience pathways and regulatory models under different development stages and cultural contexts, thereby improving the theoretical framework’s applicability and generalizability.

5. Conclusions

This study developed a resilience evaluation framework based on four key dimensions—connectivity, modularity, diversity, and adaptability—and systematically addressed the research questions posed in the introduction. The main conclusions and direct responses to the RQs are summarized as follows:

(1) The resilience of folk cultural spaces on Xiamen Island presents a spatial configuration of “dual-core leadership, corridor transition and marginal vulnerability”, broadly mirroring the island’s development trajectory from the southwest historic core to the newly developed northeastern areas. As the core enclave of traditional Minnan culture, Siming Old Town demonstrates a resilience profile that combines structural robustness with process-oriented adaptability: its historically evolved “small-scale, high-density” street–alley fabric provides a stable spatial framework for intensive everyday use, while multiplex lineage- and neighborhood-based social networks support self-organization, collective problem-solving and the gradual adjustment of cultural practices and functions in response to urban change. By contrast, the relatively high resilience of the Wuyuanwan area is mainly achieved through a modern, planning-led “squeeze–preserve–reshape” pathway, in which morphological identification, functional reconfiguration and environmental improvements preserve key cultural signifiers and integrate folk cultural spaces into new waterfront, leisure and tourism systems, albeit with potential risks of over-commercialisation. Together, these two areas constitute the “dual engines” of Xiamen Island’s cultural resilience and illustrate how historic-core robust–adaptive dynamics and planning-led functional innovation jointly shape the spatial evolution of folk cultural spaces.

(2) This study successfully constructed a multidimensional and operable quantitative assessment framework by integrating GIS-based spatial analysis, questionnaire surveys, the entropy weight method, and K-means clustering. This framework balances structural and process-oriented indicators, and its robustness was verified through sensitivity analysis, providing a reliable methodological tool for diagnosing the resilience levels and identifying the spatial differentiation of folk cultural spaces. The comprehensive resilience index and clustering results derived from this framework laid a scientific foundation for developing differentiated protection strategies.

(3) The influencing mechanisms of the four resilience dimensions show significant spatial heterogeneity and functional differentiation. In Siming Old Town, connectivity manifests as pedestrian-friendly accessibility generated spontaneously by the organically evolved street network, whereas in Wuyuanwan it depends primarily on a planned and predesigned road system. In the southwestern historic quarter, modularity derives from social cohesion rooted in kinship and territorial ties, whereas in the northeastern sector it manifests as the “island-like persistence” of cultural spaces amid rapid urbanization. This multidimensional and heterogeneous pattern underscores that the resilience of folk cultural spaces is not a simple aggregation of independent attributes but a coupled outcome of structural features, social processes, and governance logics within specific spatial contexts.

The three resilience zones identified through K-means clustering provide a spatial basis for implementing categorized and targeted protection and renewal strategies. High-resilience zones should prioritize system maintenance and functional enhancement; medium-resilience zones should address deficiencies in functional integration and spatial connectivity; and low-resilience zones should begin with the revitalization of cultural nodes, the reconstruction of community networks, and micro-scale spatial interventions to gradually restore their cultural functions and sense of place. This zoned governance approach aligns with the planning principle of “adapting measures to local conditions.”

Given the cross-sectional design, our findings should be interpreted as associations rather than causal adaptive dynamics. Future work with panel data and repeated field surveys will be required to trace temporal adaptation and path shifts.

In summary, by constructing a multidimensional and operable evaluation framework, this study reveals the structural patterns, mechanisms, and typologies of resilience within folk cultural spaces on Xiamen Island. It provides both a theoretical foundation and a practical pathway for promoting the “living” preservation and adaptive governance of cultural spaces amid urbanization. Future research should further advance in three aspects—data, dimensions, and scales—to establish a more systematic, dynamic, and refined regulatory framework for cultural space resilience.