Spatiotemporal Evolution and Driving Factors of Green Transition Resilience in Four Types of China’s Resource-Based Cities Based on the Geographical Detector Model

Abstract

Promoting synergistic economic–resource–environmental development in resource-based cities (RBCs) is a fundamental requirement for ensuring national energy security and advancing regional sustainable and coordinated development. This study innovatively proposes the theoretical framework of “green transformation resilience (GTR)” based on evolutionary resilience theory, and then empirically explores the GTR of 114 RBCs in China from the perspective of urban development stages using multiple data models. The findings indicate that the GTR demonstrated an overall upward trend, though it remained at a consistently low level. Regenerative RBCs exhibited the highest GTR levels. GTR exhibits an uneven spatial distribution, primarily caused by super-variation density. The factor detection results indicate that factors such as government intervention, income level, and human capital have strong explanatory power for the spatial variation of GTR. Interaction analysis confirmed the significant nonlinear enhancement or bivariate enhancement of all pairs of factors. This study provides a basis for the differentiated development paths of GTR in China’s RBCs. Moreover, through factor interaction testing, it also offers guidance on policy combinations and prioritization for RBCs in different development stages.

1. Introduction

Resource-based cities (RBCs) are cities with an economy largely based on the extraction and processing of natural resources, such as minerals and forests. These cities play a major role in global industrialization and urbanization, and have a significant impact on the economy and ecological environment of the region and country. However, excessive dependence on natural resources often leads to the so-called “resource curse”, intensifying a range of challenges, including a mono-industrial structure, environmental degradation, and diminished social resilience [1,2]. Aligning with the global Sustainable Development Goals (SDGs) and China’s “Dual Carbon” strategy, transitioning resource-based cities (RBCs) from traditional practices to green and low-carbon development is crucial. This shift is vital for ensuring national energy and resource security and fostering balanced regional growth [3]. In this transformation, the concept of “resilience”—which emphasizes the core ability of a system to absorb shocks, adapt to changes and pursue sustainable development—provides a powerful framework for us to analyze the dynamic process, potential risks and long-term effects of RBCs transformation [4,5].

Figuring out how to scientifically evaluate and effectively promote this complex transformation has become an important research direction in recent years. Scholars’ research basically focuses on two related topics—urban green transition and urban resilience, which are also the theoretical basis of our research.

Regarding the urban green transformation, the current research mainly focuses on the evaluation framework and influencing factors, and uses two main methods to evaluate the urban green transformation. The first method is to establish a multi-faceted evaluation system, which includes indicators such as economic restructuring, social progress, environmental quality, innovation ability, living environment and resource utilization [4,5,6,7]. The second method is to use the input–output efficiency model to evaluate the green transformation. For example, Yin et al. [8] used a three-stage data envelopment analysis model to measure the green transformation efficiency of 110 mineral RBC, and found that the overall efficiency level was still low. Similarly, Liu et al. [9] applied the super-efficiency relaxation measurement model to estimate the green development efficiency of 12 major cities in Gansu, China Province. Studies have examined variations in variable selection, sample scope, and methodological design as influencing factors. Zhang et al. [10], using the generalized method of moments, demonstrated that innovation-driven strategies significantly enhance green economic growth in RBCs, with heterogeneous effects observed across city types. Through Panel Vector Autoregression models, Wang et al. [11] confirmed the positive impact of information and communication technology (ICT) on the green transition efficiency of resource-exhausted cities, noting marked regional differences. Jiang and Sun [12] used a method called double machine learning to study 282 prefecture-level cities in China, and found that the construction of smart city can greatly promote green production, green life and green economic growth. These studies establish a basis for identifying green transition pathways in resource-based cities. However, they exhibit several limitations; notably, most focus on single factors or broad regional analyses, and insufficient attention is given to the unique nature of resource-based cities as a distinct category.

Since Holling [13] first proposed it, resilience theory has evolved from “engineering resilience” (emphasizing restoration to the original state) to “ecological resilience” (emphasizing the system’s ability to absorb disturbances), and then to “evolutionary resilience” (emphasizing system adaptation, learning, and fundamental transformation) [14]. Evolutionary resilience particularly focuses on the adaptability and transformative potential of systems in long-term changes, providing core theoretical support for the construction of the “green transformation resilience” framework in this study. In recent decades, the concept of resilience related to cities has received increasing amounts of theoretical and empirical attention from researchers in various disciplines. Among the newly developed concepts, economic resilience and ecological resilience are closely related to the GTR discussed in this paper. Research on economic resilience typically focuses on three areas: level measurement, spatial pattern analysis, and influencing factors. Measurement approaches include single-indicator methods—such as changes in employment [15,16], unemployment [17,18], and GDP growth rates [19,20]—as well as composite index systems that provide more holistic assessments [21,22,23]. Studies on spatial patterns have examined units ranging from urban agglomerations and provinces to prefecture-level cities and districts [24,25,26]. Influencing factors identified include innovation capacity, sound economic policies, industrial diversity, human capital levels, trade openness, financial liberalization, and industrial specialization [25,27,28]. Research on ecological resilience focuses on its theoretical basis, measurement, and influencing factors. Scholars frequently use evolutionary resilience theory to define ecological resilience as a system’s ability to remain stable, adapt its structure and functions, and transform when faced with external pressures [29]. For measurement, multi-dimensional indicator systems are frequently employed [30,31,32]. Common methods include the entropy weight method (EWM) [33,34], analytic hierarchy process (AHP) [35], and hybrid approaches such as Criteria Importance Through Intercriteria Correlation (CRITIC)–EWM combinations [36]. Regarding influencing factors, widely used analytical tools include the Spatial Durbin Model (SDM) [37], multiple linear regression [38], geographically weighted regression (GWR) [39], and obstacle degree models [40]. Existing research is relatively abundant, but it often focuses on a single dimension of economy or ecology, and as such lacks an integrated analysis of the resilience of economic–environmental synergy.

While the existing literature has established a strong foundation for research in this field, several important gaps persist. First, most studies on urban green transition focus on the isolated impacts of individual factors (e.g., economic or social factors) on transition efficiency, and they do not adequately address the complex interactions among multiple dimensions. This isolated perspective makes it difficult to comprehensively assess a city’s overall adaptability in the face of complex shocks, and it also weakens the systematicity and coordination of policy design. Second, research has primarily concentrated on national or regional urban agglomerations, and there are relatively few systematic investigations specifically for RBCs. As a result, existing conclusions and policy recommendations often fail to precisely match the actual situations and special challenges that resource-based cities face during their transformation. Third, the concept of resilience has not yet been deeply integrated into the framework of green transformation, with the lack of a comprehensive analysis framework combining economic resilience with ecological resilience being especially notable [41]. This limits current understandings of the mechanism and assessments of the path by which resource-based cities can achieve sustainable and coordinated economic, ecological, and social development during their transformation.

To bridge these gaps, this study focuses on RBCs while explicitly incorporating the synergistic relationship between economic and environmental systems. By integrating resilience theory into the analytical framework of green transition, we propose the novel concept of “green transition resilience” (GTR) and undertake a comprehensive empirical investigation. The GTR framework breaks through the limitation of previous studies, which usually only examined a single dimension of the economy or environment. By systematically integrating various views on economy, ecology and society, GTR framework constructs a capability system. This system can fully reflect the ability of resource-based cities to cope with complex shocks and achieve coordinated development in the process of transformation. Accordingly, this study reveals the temporal and spatial evolution, regional differences and driving mechanism of GTR, which enables us to analyze the green transformation of resource-based cities more comprehensively, dynamically and harmoniously. The marginal contribution of this research has three aspects. Firstly, it introduces resilience theory into the green transformation framework of RBCs for the first time, breaks through the limitation of traditional single-dimensional analysis, constructs a resilience analysis framework to understand the co-evolution of economy and environment, and provides an innovative perspective of system analysis. Secondly, it develops a set of indicator system to measure GTR in RBCs, which broadens the research of urban resilience and provides a practical methodology for quantitative evaluation. Thirdly, based on the typical samples of RBCs in China, the proposed system identifies the performance of GTR in different development stages of RBCs in differentiation characteristics, spatial evolution, key factors and interaction mechanisms by using a variety of econometric and spatial models, which provides a scientific basis for formulating differentiated policies.

This paper is organized as follows: Section 2 explains the methodology and modeling framework, as well as an explanation of the data. Section 3 discusses the research findings. Section 4 summarizes the main results, explains the policy implications, and discusses future research directions.

2. Methods and Data

2.1. Criteria Importance Through Intercriteria Correlation (CRITIC)–Linear Weighting Model

The CRITIC–linear weighting method determines indicator weights by jointly evaluating contrast intensity and inter-indicator conflict. Specifically, contrast intensity is quantified using standard deviation, which reflects data variability, while inter-indicator conflict is measured using correlation coefficients to reduce redundancy among highly correlated indicators [42]. Based on the weights derived from the CRITIC method, this study employs a linear weighting model to comprehensively assess the GTR of RBCs. We calculated a set of global weights using pooled data from all years and applied them to each year to compute the composite index. This approach facilitates both cross-sectional comparisons across regions and longitudinal comparisons over the entire period, as it avoids distortions caused by annual fluctuations in the weighting scheme. The computational procedure is as follows:

$$S_j = \sqrt{{\displaystyle \frac{\sum _{i=1}^m{x}_{\mathit{ij}}}{n-1}}}, R_j = {\sum }_{i=1}^n\left(1-r_{\mathit{ij}}\right),$$

(1)

$$C_j={ S}_j \times { R}_j,$$

(2)

$$W_j={\displaystyle \frac{C_j}{\sum _{j=1}^n{C}_j}},$$

(3)

$$U_i=\sum _{j=1}^n{x}_{\mathit{ij}}{W}_j,$$

(4)

where $m$ is the number of evaluated samples, $n$ is the number of evaluation indicators, $x_{\mathit{ij}}$ is the dimensionless value of the evaluation indicator $j$ for the sample $i, S_j$ is the standard deviation of the indicator $j, r_{\mathit{ij}}$ is the correlation coefficient between evaluation indicators $i$ and $j,$ and $C_j$ is the information content of the evaluation indicator $j$ within the entire indicator system. A greater $C_j$ value indicates an increased contribution of the indicator $j$ to the overall evaluation system, and thus it should be assigned a higher weight. $U_i$ is the comprehensive GTR index of the sample $i$.

To eliminate the effects of different measurement units, we standardized all indicators using the min-max normalization method, scaling all values to the range [0, 1]. This approach was selected because it preserves the original distributional relationships in the data while demonstrating good compatibility with the CRITIC weighting method. To ensure the robustness of the research conclusions, we conducted robustness tests by employing alternative standardization methods and alternative weighting methods, with detailed results available in File S1.

2.2. Kernel Density Estimation Method

Kernel density estimation (KDE) is widely recognized as a fundamental tool for analyzing the spatial distribution of quantitative data due to its low model dependence and high robustness [43]. The resulting density surfaces effectively reveal the spatial location, morphological structure, and extent of the observed phenomenon. This study utilizes KDE to analyze density surfaces, revealing dynamic changes, agglomeration effects, and polarization patterns in the GTR of Chinese RBCs. Assuming that the density function of the random variable X is f(x), the estimated probability density at point x is given by:

$$f_x={\displaystyle \frac{1}{\mathit{Nh}}}\sum _{i=1}^{N}K\left({\displaystyle \frac{X_i-x}{h}}\right),$$

(5)

where $N$ is the total number of observations, $K\left(·\right)$ is the kernel function, $X_i$ represents independent and identically distributed observations, $x$ is the mean, and $\mathrm{h}$ denotes the optimal bandwidth.

2.3. Dagum–Gini Coefficient

To address the limitations of conventional regional inequality decomposition methods, this study adopts the Dagum–Gini decomposition approach proposed by Dagum [44]. This method is particularly effective in identifying the specific sources of disparity. By implementing the Dagum decomposition in MATLAB 2022b, we analyze the underlying causes of variation in GTR across different types of RBCs. The defining equation is:

$$G=\sum _{j=1}^k\sum _{h=1}^k\sum _{i=1}^{n_j}\sum _{r=1}^{n_h}|j_{\mathit{ji}}-y_{\mathit{hr}}|/2n^2\overline{y},$$

(6)

where $k$ denotes the number of categories; $n$ denotes the number of RBCs; $y_{ji}\left(y_{hr}\right)$ denotes the GTR of city $i\left(r\right)$ within category $\left(h\right);$ and $y$ denotes the mean value of GTR in each RBC.

The Gini coefficient for category $j$ is expressed as:

$$G_{\mathit{jj}} = {\displaystyle \frac{\sum _{i=1}^{n_j}\sum _{r=1}^{n_j}\left|y_{\mathit{ji}}-y_{\mathit{jr}}\right|}{2n_j^2\overline{Y_j}}}.$$

(7)

The Gini coefficient between category $j$ and category $h$ is expressed as:

$$G_{jh}={\displaystyle \frac{\sum _{i=1}^{n_j}\sum _{r=1}^{n_h}\left|y_{\mathit{ji}}-y_{\mathit{hr}}\right|}{n_j{n}_h\left(\overline{Y_j}+\overline{Y_h}\right)}}.$$

(8)

$G_w$, $G_{nb}$, and $G_t$ are denoted as:

$$G_w=\sum _{j=1}^k{G}_{\mathit{jj}}{p}_j{s}_j,$$

(9)

$$G_{nb}=\sum _{j=2}^k\sum _{h=1}^{j-1}{G}_{\mathit{jh}}\left(p_j{s}_h+p_h{s}_j\right)D_{\mathit{jh}},$$

(10)

$$G_t=\sum _{j=2}^k\sum _{h=1}^{j-1}{G}_{\mathit{jh}}\left(p_j{s}_h+p_h{s}_j\right)\left(1-D_{jh}\right).$$

(11)

The relationship between the three satisfies $G=G_w+G_{\mathit{nb}}+G_t$, where $p_j=n_j/n$, $s_j=n_j\overline{Y_j}/n\overline{Y}\left(j=1,\dots ,k\right)$. $n_j\left(n_h\right)$ denotes the number of cities in category $j_h$ and $Y_j\left(Y_h\right)$ denotes the mean value of GTR in category $j_h$.

2.4. Geographical Detector Model

The Geographical Detector is a tool used to detect spatial heterogeneity and its associated driving factors [45]. The magnitude of the q-value obtained through the Geographical Detector reflects the degree of spatial association between variables. While the q-value cannot directly validate causal relationships, it can identify factors that exhibit significant spatial associations with the studied phenomenon. Therefore, in this study, the term “driving factors” specifically refers to those factors identified by the Geographical Detector as being significantly associated with the spatial heterogeneity of GTR in RBCs. Furthermore, the Geographical Detector’s capacity to detect spatial differentiation is not limited to geographic stratification, but it also extends to attribute-based stratification, such as the classification of resource-based cities into different types based on their developmental stage [45]. In this study, the factor detector module was used to identify core associated factors, while the interaction detector module was applied to examine the interdependencies among these factors. The mathematical formulation of the factor detector is as follows:

$$q=1-{\displaystyle \frac{\sum _{h=1}^L{N}_h{\sigma }_h^2}{N{\sigma }^2}},$$

(12)

where $q$ denotes the explanatory power of an associated factor on the spatial heterogeneity of the dependent variable, $h$ denotes the stratification of the variable, ${\sigma }_h^2$ and ${\sigma }^2$ represent the variance of units in stratum $h$ and the global variance, respectively, while $N_h$ and $N$ correspond to the number of units in stratum $h$ and the total number of units in the entire study area.

The interaction detector is used to quantify the interactive effects of two independent variables on the response variable. The interaction outcomes are primarily categorized into five types: nonlinear weakening, single-factor nonlinear weakening, bi-factor enhancement, independence, and nonlinear enhancement.

2.5. Indicator System and Data

According to the evolutionary resilience theory, we put forward the concept of “green transition resilience” (GTR), and apply it to the specific scene of green transformation of resource-based cities. Unlike previous studies that always focus on economic fluctuations and natural disasters, the GTR framework pays more attention to various driving forces of transformation, such as environmental regulation, resource depletion and green innovation. It is a conceptual expansion and situational adjustment of resilience theory under the background of urban green transition [46]. This study divides GTR into three dimensions: resistance, adaptability, and innovation. The core definition of GTR is as follows: under the restriction of resources and environment, cities are not only able to withstand disturbances caused by environmental degradation, regulatory tightening, and other factors (Resistance), actively adjust its economic structure and change its development mode (Adaptability), and at the same time make good use of scientific and technological innovation and institutional reform to successfully guide and achieve green and low-carbon development goals (Innovation) [13,14]. These three dimensions are interconnected and evolve progressively, together constituting a complete resilience assessment framework. To more intuitively present the theoretical structure of GTR, Figure 1 illustrates the theoretical relationships among the three dimensions of “Resistance—Adaptability—Innovation”. Resistance provides a stable foundation for the system, Adaptability enables the system to adjust its functions on this basis, and Innovation is the key driving force to promote the system to upgrade to a higher toughness state. This study defines GTR as a comprehensive capacity system, which not only emphasizes the stability of the system under impact, but also emphasizes the dynamic ability of the system to actively adjust, continuously evolve and finally achieve the goal of green transition [14].

Building upon this conceptualization—and integrating insights from existing literature [41,47,48] along with requirements for urban green transition and sustainable development [49,50,51,52,53,54]—an indicator system was constructed to evaluate GTR. The system comprises indicators selected across three key dimensions: Resistance, Adaptability, and Innovation (detailed in

Table 1. Indicator system for GTR.

Level 1 Indicators	Level 2 Indicators	Level 3 Indicators	Specific Meaning	Nature	Weight
Green transition resilience	Resistance	GDP per capita	GDP/Urban population (CNY/person)	+	0.0488
		Fiscal self-sufficiency ratio	Local Government Revenue/Local Government Expenditure (%)	+	0.0961
		Energy consumption	Total energy consumption estimated from nighttime light data (10,000 tons of standard coal equivalent)	−	0.0485
		Water consumption per capita	Total urban water consumption/Urban population (m3/person)	−	0.0399
		Industrial wastewater emissions per unit GDP	Total industrial wastewater emissions/GDP (10,000 tons/CNY)	−	0.0404
		Industrial fume and dust emissions per unit GDP	Total industrial fume and dust emissions/GDP (tons/CNY)	−	0.0268
		Industrial SO2 emissions per unit GDP	Total industrial SO2 emissions/GDP (tons/CNY)	−	0.0163
	Adaptability	Economic diversification	Employment in the service sector/Total Employment (%)	+	0.0919
		Industrial upgrading	Value-added of the service sector/Value-added of the industrial sector (%)	+	0.0729
		Self-purification capacity	Greening coverage of built-up areas (%)	+	0.0462
		Waste treatment capacity	Innocuous treatment rate of municipal solid waste (%)	+	0.0357
		Wastewater treatment capacity	Percentage of wastewater treated at centralized plants (%)	+	0.0663
		Sustainable transportation	Annual passenger volume of public bus services/Urban population (passenger trips per capita per year)	+	0.0677
	Innovation	Development of emerging industries	Number of AI enterprises (units)	+	0.0358
		R&D expenditure	Government expenditure on R&D/Local government expenditure (%)	+	0.0669
		Education expenditure	Government expenditure on education/Local government expenditure (%)	+	0.0908
		Green innovation output	Number of granted green patents per 10,000 population	+	0.0512
		Innovative human capital	Number of college students per 10,000 population	+	0.0580

).

Resistance is concerned about the extent to which a city can rely on the existing economic stock, financial health, and adjust the intensity of resource and environmental consumption when resource dependence is weakened or environmental regulations are tightened. GDP per capita measures the overall economic development level of a city and serves as the economic foundation for maintaining social operations and providing public services. The fiscal self-sufficiency rate reflects the fiscal independence of local governments under the potential fluctuation of resource revenues. Total energy consumption is a negative indicator that primarily measures the overall pressure on resources and the environmental impact imposed by the urban economic system. Water consumption per capita, industrial wastewater emissions per unit GDP, industrial fume and dust emissions per unit GDP, and industrial SO₂ emissions per unit GDP collectively reflect a city’s efficiency and intensity in its resource utilization and pollution emissions.

Adaptability focuses on the ability of urban systems to optimize internal processes, improve governance efficiency, and transform behavioral patterns to reestablish stability after being impacted. Economic diversification and industrial upgrading measure the effectiveness of adjustments in the diversification and sophistication of economic structure, reflecting the results of a city’s efforts to drive industrial transformation. Self-purification capacity represents the interventions by which a city actively enhances environmental capacity and quality of life through ecological construction. Waste treatment capacity and wastewater treatment capacity measure a city’s investment and operational level in environmental management infrastructure. Meanwhile, sustainable transportation measures the effectiveness of a city in guiding residents’ travel habits towards greener and low-carbon options.

Innovation primarily assesses the potential and achievements in cultivating new knowledge, technologies, industries, and human capital, thereby fundamentally transforming the path of green transition. The development of emerging industries, measured by the existing stock of urban artificial intelligence enterprises, is mainly reflected in the potential to foster future competitiveness and new growth engines within the green transition. AI is a horizontal general-purpose technology capable of comprehensively enabling green transformation. The degree of AI industrial agglomeration reflects a city’s overall ability to embed intelligent and green drivers within its economic system, rather than being limited to any specific sector. In addition, within China’s development strategy, AI is explicitly designated as a core component of strategic emerging industries. Using the AI industry as a representative of emerging industries allows for the nuanced capture of cities’ efforts and achievements in competing for and positioning themselves within the most cutting-edge future industries. R&D expenditure and education expenditure measure the government’s investment in the foundation of innovation. Green innovation output measures the direct output and knowledge accumulation of innovation activities in the field of green technology, while innovative human capital measures the reserve and supply potential of innovative human capital.

There are currently 126 prefecture-level cities in China that are classified as resource-based [55]. Based on differences in resource security capacity and sustainable development potential, these cities are categorized into four types: Mature RBCs, Growing RBCs, Declining RBCs, and Regenerative RBCs. Due to severe data gaps for certain cities and years—and to ensure data continuity and availability—this study employs a panel dataset covering 114 prefecture-level cities in mainland China from 2010 to 2022. To eliminate the effects of inflation, all GDP-related variables were deflated to constant 2010 prices. Primary data sources include the National Bureau of Statistics, provincial statistical bureaus, the China City Statistical Yearbook, the China Urban Construction Statistical Yearbook, provincial and municipal statistical yearbooks, and municipal statistical bulletins. A limited number of missing values were imputed using interpolation techniques. Specifically, urban energy consumption was estimated using nighttime light data [56], obtained from the National Earth System Science Data Center (https://www.geodata.cn) [57].

3. Results and Discussion

3.1. Analysis of Basic Characteristics of Indices

3.1.1. Temporal Evolution Characteristics

Figure 2 illustrates the overall GTR level and dimensional trends from 2010 to 2022. The GTR of China’s RBCs showed a general increase, rising from 0.255 in 2010 to 0.300 in 2022. This indicates a notable enhancement in resilience, although the overall level remained relatively low.

A relatively rapid increase in GTR was observed between 2010 and 2012. However, a brief decline occurred in 2013, primarily due to reductions in both Innovation and Resistance, with the decline in Innovation being more pronounced. Since 2013, GTR has been growing, reaching the highest value of 0.304 in 2021. This continuous improvement is mainly influenced by the continuous improvement of Adaptability, which coincides with the implementation time of The National Sustainable Development Plan for Resource-based Cities (2013–2020) [55]. After the implementation of this Plan, RBCs found ways to deal with resource constraints and environmental problems under the new development background, and they adopted some strategies to enhance their adaptability to internal and external shocks. These actions directly led to a continuous improvement in Adaptability, which significantly improved GTR in the implementation period of the Plan. Although a slight decrease was observed in 2022, GTR remained at a relatively high level compared with previous years in the study period. Across the entire study period, Adaptability always had the highest value, which made it the dimension that contributed the most to GTR. In contrast, Innovation and Resistance remained at lower levels and contributed to a lesser extent.

To further analyze the evolution of GTR levels across the four RBC types, Figure 3 presents GTR trends for each city category from 2010 to 2022.

According to the urban development stage, the GTR of all city types exhibited fluctuations, but it generally had an increasing trend across the entire research period, and it was almost aligned with that of the national aggregate. Regenerative RBCs exhibited relatively high resilience levels, with a mean value of about 0.304, which was higher than that of the national aggregate (0.279). Mature RBCs, Declining RBCs, and Growing RBCs had mean values of about 0.279, 0.270, and 0.267, respectively. In terms of growth rates, Declining RBCs ranked first, with a 25.25% increase in GTR. This was followed by Mature RBCs (16.47%), Regenerative RBCs (15.16%), and Growing RBCs (14.34%). Among the four types, Growing RBCs performed the worst in both overall resilience level and growth rate. Regenerative RBCs, such as Zibo and Xuzhou, have a relatively strong industrial base and initiated their transformation early. By systematically promoting the green transformation of traditional industries and fostering emerging business formats, these cities accumulated substantial capabilities in technological innovation and environmental management, with their GTR maintaining a leading position. Mature RBCs, such as Karamay, possess a complete resource-based industrial system and strong economic stability. In recent years, Karamay has gradually advanced its green transformation by extending its industrial chain and deploying clean energy, resulting in a steady improvement in GTR. Declining RBCs, such as Tongling, under conditions characterized by resource depletion and growth-related pressure, have actively cultivated successor industries, promoted ecological restoration, and carried out urban renewal with policy support, thereby achieving rapid GTR advancement. Growing RBCs, such as Ordos and Yulin, are still in a phase of resource-dependent expansion, with slower progress in green technological innovation and industrial diversification, leading to a low overall GTR and inadequate growth.

Kernel density curves of four representative years (2010, 2014, 2018, and 2022) in China’s RBCs are shown in Figure 4. The data in Figure 4 exhibit a rightward shift in the central position of the kernel density curves over time, signifying a consistent annual enhancement in GTR across China’s RBCs. The peak height of the curves initially increased and then decreased, suggesting that inter-city disparities in GTR initially narrowed before widening later. In terms of distributional shape, a secondary peak emerged only in 2022, indicating the onset of a polarization trend in GTR among RBCs during that year. This may be related to the COVID-19 pandemic. Specifically, in 2022, cities at the secondary peak included Shizuishan, Yan’an, and other cities located primarily in central, western, or remote regions, many of which were Growing or Declining RBCs. In contrast, cities at the primary peak included Zhangjiakou, Chuzhou, Panjin, Nanchong, and other cities with advantageous geographic locations, most of which were Mature or Regenerative RBCs. Under the impact of the COVID-19 pandemic, the vulnerability of Growing and Declining RBCs with single industrial structures and lagging transformation processes became especially pronounced, hindering or slowing their progress toward green transformation and thus placing them in the secondary peak range. Mature and Regenerative RBCs were able to leverage their stronger adaptability and recovery capabilities to effectively withstand shocks, and as such could maintain or even enhance their position at the primary peak.

3.1.2. Spatial Evolution Characteristics

To analyze the spatial evolution of GTR, this study selected four time points, namely, 2010, 2014, 2018, and 2022. Based on ArcGIS 10.5 software, the spatial distribution of GTR in China’s 114 RBCs is displayed in Figure 5. In addition, the Jenks Natural Breaks classification method in GIS was employed to categorize cities into five distinct GTR levels: High-value Zones, Moderately High-value Zones, Medium-value Zones, Moderately Low-value Zones, and Low-value Zones. This classification enabled the clear examination of spatial evolution patterns.

In general, China’s RBCs presented a spatially unbalanced distribution of GTR, which was concentrated in low-value cities and scattered in high-value cities. Although the national average GTR level improved over time, most cities remained concentrated in the Medium-value and Moderately Low-value Zones. Notably, the number of cities classified as High-value or Moderately High-value declined from 35 to 25 during the study period. Cities with higher GTR levels were primarily Regenerative and Mature RBCs, although the number of Mature RBCs in these upper categories progressively declined over time. Both Mature and Declining RBCs increasingly clustered in the Low- and Medium-value Zones, while Growing RBCs became more concentrated within the Low-value Zones. In contrast, Regenerative RBCs consistently gravitated toward the High-value Zones.

3.2. Regional Differences and Source Decomposition

The Dagum–Gini coefficient decomposition method was employed to calculate the overall Gini coefficient (G), within-group Gini coefficient (G_w), between-group Gini coefficient (G_nb), and transvariation density Gini coefficient (G_t) for GTR among China’s RBCs from 2010 to 2022. The temporal evolution of these coefficients is illustrated in Figure 6.

The overall Gini coefficient of GTR from 2010 to 2022 ranged from 0.076 to 0.091, and is average value was 0.081. This meant that the overall regional disparity of GTR was not large and was stable from 2010 to 2022.

The overall G_w fluctuated downward, from 0.030 to 0.028. All four RBC types experienced a decrease in internal disparities. Among them, Declining RBCs showed the most pronounced reduction, with G_w falling from 0.104 to 0.086, followed closely by Regenerative RBCs, which also demonstrated a substantial decline. The overall G_nb declined from 0.029 to 0.019, reflecting a reduction in disparities between different city types. The greatest inter-group disparity was observed between Declining and Regenerative RBCs (mean G_nb = 0.094), followed in order by Growing–Regenerative, Mature–Regenerative, Mature–Declining, Growing–Declining, and Mature–Growing. This pattern highlights that Regenerative RBCs differ more significantly from the other three types than the other types differ among themselves.

Figure 7 illustrated the contribution rates of disparity sources to the Dagum–Gini coefficient for GTR in China’s RBCs. Among the three components, transvariation density emerged as the dominant contributor, with an average contribution rate of 41.78%, followed by within-group differences (35.35%) and between-group differences (22.88%). This indicates that transvariation density is the primary source of relative disparities in GTR across different RBC types. This finding has significant policy implications, as it suggests that the GTR of China’s RBCs is influenced more by the divergence among cities of the same type than by the inherent gaps between different types. This is primarily due to the fact that cities at the same stage of development have taken markedly different paths of transformation, owing to differences in their location, resource endowments, and local governance capabilities.

3.3. Analysis of Associated Factors

3.3.1. Selection of Associated Factors

Disparities in GTR between different types of RBCs are caused by the joint effects of several economic, policy, and social factors. Based on a review of relevant literature and the practical context of RBC development [58,59,60,61,62], this study selected ten factors spanning three dimensions: economy, society, and policy. The specific indicators used are defined in

Table 2. GTR-associated factors.

Dimension	Factor	Specific Meaning
Economy	Income level (IL)	Per capita disposable income of urban residents (CNY/person)
	Market size (MS)	Share of total retail sales of consumer goods in GDP (%)
	Financial development (FD)	Share of the year-end outstanding deposits and loans of financial institutions in GDP (%)
	Openness level (OL)	Share of total trade in GDP (%)
	Economic vitality (EV)	GDP growth speed (%)
Society	Population agglomeration (PA)	Population density (people per square kilometer)
	ICT infrastructure (ICTI)	Output of postal and telecommunications services (million yuan)
	Human capital cultivation (HCC)	Share of teachers in higher education in total population (%)
Policy	Government intervention (GI)	Share of local government expenditure in GDP (%)
	Environmental regulation (ER)	Term frequency density of green-related terminology in local policy addresses (%)

. The specific steps for measuring environmental regulation are as follows. First, word segmentation was performed on the text of each included municipal government work report. Next, the frequency of words related to “green” was determined by counting, and the proportion of these words relative to the total word frequency in the government report was calculated. The core keywords include categories such as environmental protection, green production, and green ecology [63].

3.3.2. Factor Detection Results

This study utilized the geographical detector model for factor detection of 10 independent variables and conducted comparative analysis on the temporal evolution of the main driving factors. Prior to analysis, all factors were discretized and categorized into four levels using the quantile method. The q-value representing the explanatory power of each factor on GTR is shown in

Table 3. Factor detection results for GTR in China’s RBCs.

Factor	2010		2022
	q	Rank	q	Rank
IL	0.262 ***	2	0.300 ***	2
MS	0.046	8	0.059	8
FD	0.039	9	0.052	9
OL	0.047	7	0.139 ***	4
EV	0.054	5	0.060 *	7
PA	0.053	6	0.020	10
ICTI	0.118 ***	4	0.066 *	6
HCC	0.145 ***	3	0.230 ***	3
GI	0.340 ***	1	0.402 ***	1
ER	0.004	10	0.082 **	5

Note: *, **, and *** indicate significance at the levels of 10%, 5% and 1%, respectively.

.

At the national level, government intervention (GI), income level (IL), and human capital cultivation (HCC) were the top three factors most strongly associated with the spatial differentiation of GTR. In addition, the explanatory power of these factors was consistently enhanced. Specifically, the q-value for GI increased from 0.340 to 0.402, IL increased from 0.262 to 0.300, and HCC increased the most, from 0.145 to 0.230. The explanatory power of GI continued to strengthen and consistently remained at the forefront. This trend aligns with the institutionalization of environmental governance and regulation in China after 2015. Faced with stringent higher-level assessments, local governments generally expanded their fiscal expenditures to implement policies related to environmental governance and industrial adjustment. Therefore, the GI reflects the overall effort and resource mobilization capacity of local governments in executing green transformation policies. It is noteworthy that the explanatory power of HCC exhibits the greatest increase. This is directly related to China’s economic shift from being “factor-driven” to “innovation-driven”. Since the implementation of the 13th Five-Year Plan, the complexity of green technology and the knowledge intensity of various industries have increased significantly. A highly skilled workforce has become an essential factor for cities to adopt and adapt green technologies and nurture emerging industries. In addition, the q-value of OL and ER has progressively grown since 2010, now ranking as the fourth and fifth most significant factors, respectively.

Table 4. Single-factor detection results for GTR in different types of China’s RBCs.

Factor	2010				2022
	Mature RBCs	Growing RBCs	Declining RBCs	Regenerative RBCs	Mature RBCs	Growing RBCs	Declining RBCs	Regenerative RBCs
IL	0.321 ***	0.245	0.294	0.599	0.227 ***	0.453	0.590 ***	0.385
MS	0.095	0.659 **	0.049	0.101	0.030	0.166	0.384	0.134
FD	0.114 *	0.069	0.253	0.325	0.020	0.278	0.301	0.625 **
OL	0.052	0.433	0.117	0.215	0.077	0.287	0.311	0.452
EV	0.041	0.056	0.100	0.272	0.072	0.290	0.203	0.073
PA	0.007	0.381	0.354	0.285	0.055	0.346	0.245	0.079
ICTI	0.066	0.178	0.131	0.122	0.047	0.127	0.389	0.652 *
HCC	0.147 *	0.126	0.438 *	0.646 *	0.147 *	0.244	0.605 ***	0.326
GI	0.299 ***	0.423	0.565 ***	0.582 *	0.274 ***	0.728 **	0.599 ***	0.728 ***
ER	0.001	0.275	0.214	0.301	0.016	0.324	0.068	0.573 **

Note: *, **, and *** indicate significance at the levels of 10%, 5% and 1%, respectively.

illustrates the single factor detection results for GTR in different types of China’s RBCs. The results indicate that GI was always positively correlated with GTR. Meanwhile, the dominant factors for each city category were different, indicating that the factors most strongly associated with the spatial differentiation of GTR demonstrate significant heterogeneity among different types of RBCs.

In Mature RBCs, the top three factors remained unchanged, consistently comprising GI, IL, and HCC. Interestingly, the q-values of GI and IL gradually decreased, while that of HCC was relatively stable. These results indicate a relatively fixed trajectory for GTR development in Mature RBCs, which continues to rely heavily on policy-driven expenditures. In Growing RBCs, the primary factor shifted from MS to GI. This is because these cities are still in the resource dividend stage, in which the economic returns from being resource dependent are still highly attractive. Therefore, key urban stakeholders have displayed a relatively weak proactive willingness toward green transition, and GTR in these cities is largely driven by fiscal policy instruments. In Declining RBCs, HCC replaced GI as the primary factor. GI declined to second place, while IL rose to third. The increasing importance of HCC reflects its role in enhancing technological capacity and enabling cities to pursue leapfrog development through a “latecomer advantage.” In Regenerative RBCs, GI progressively became the dominant factor, followed by ICTI, FD, and ER. These cities had a relatively strong economic foundation and technological innovation capability, and they also had relatively mature environmental regulation systems. Fiscal expenditures were substantially directed toward promoting eco-friendly and technology innovation industries to achieve dual goals.

3.3.3. Interaction Detection Results

Figure 8 illustrates the interaction detection results of GTR. There were significant interactions among different factors. Most interactions improved explanatory power via nonlinear and bivariate enhancement. The detailed version is shown in Figure S1.

At the national level, GI showed the strongest explanatory power in interactions with other factors, and the overall trend of its explanatory power was upward. Among the various interactions, GI × HCC had the highest explanatory power, and its continuously increasing explanatory power may reveal a process of mutual empowerment. On one hand, this reflects the mechanism of financial resources supporting human capital development, in which government spending on education and technology directly enhances the city’s human capital and foundation for innovation. On the other hand, it may also reflect the mechanism of “high-quality talent improving policy effectiveness”, whereby a more skilled and innovative workforce can more effectively utilize government resources to promote the research, development, and application of green technologies, thereby increasing the marginal returns of transformation policies.

Across city types, the results of interaction detector showed that the interactive explanatory power of two factors was always higher than that of a single factor. For Mature RBCs, both IL and GI exhibited strong interaction effects, suggesting that their synergistic effects have significant explanatory power for the spatial differentiation of GTR. Notably, the interaction of IL × FD consistently demonstrates high explanatory power, highlighting the catalytic role of market dynamics and financial instruments in promoting urban green transformation. For Growing RBCs, the interaction effect of MS presented a decreasing trend, while the interaction effect of GI and PA always increased. The interactive explanatory power of PA × OL surpassed that of PA × MS, reaching 99.4%. For Declining RBCs, the explanatory power of GI combined with other factors exceeds 55%. It can be seen that the policy instrument plays a central role in promoting sustainability transition. In addition, the interaction effect of HCC also improved, which was consistent with the result of single factor detection analysis. For Regenerative RBCs, the interaction effect of HCC gradually weakened, while those of ER, GI, and FD progressively strengthened. The interaction of ER × OL displayed exceptional explanatory power, reaching 97.9%. This finding suggests that stringent environmental regulations create space for industrial transformation by mandating the elimination of obsolete industries. Simultaneously, greater openness promotes the introduction of international green technologies and capital. Through this synergistic combination of phasing out outdated capacity and fostering advanced systems, these dual forces jointly enhance urban resilience in the green transition process.

In summary, the interaction detection results reveal both shared and differentiated driving mechanisms across city types. GI consistently exhibited strong interaction effects in all RBC categories. In addition to this commonality, IL was particularly important in Mature and Growing RBCs, and ER and FD were particularly important in Regenerative RBCs.

3.4. Discussion

Under the global trend of sustainable development, promoting the green transformation of RBCs is crucial for achieving regional coordinated development. This study analyzed the spatiotemporal evolution characteristics, developmental stage heterogeneity, and underlying mechanisms of GTR.

This study found that the level of Adaptability was consistently the highest of all the factors and that it drove improvements following the Plan [55]. This finding strongly supports the core concept of resilience theory regarding system reorganization. The implementation of the Plan constituted a significant external disturbance, compelling China’s RBCs to abandon their previous state and necessitate internal adjustments and reorganization. As emphasized by Meerow et al. [46], urban resilience requires transformative capabilities that go beyond mere resistance. The enhancement of Adaptability reflects these cities’ ability to reorganize their systems through adjusting their industrial structures and improving resource efficiency, thereby achieving sustainable development amid change, which surpasses the passive defense model that relies solely on Resistance.

Our findings reveal notable differences in GTR across cities at varying development levels. This observation not only aligns with the findings of Dou et al. [64] regarding green development efficiency in western China’s RBCs, but it also closely resonates with the differentiated transformation paths of Rust Belt cities in the United States [65]. The heterogeneity of these transformation outcomes largely stems from differences in the initial conditions and development strategies of these cities. This suggests that heterogeneity in development stages is not unique to China, but a common challenge for the transformation of resource-based regions globally.

Factor detection results indicate that GI has strong explanatory power for the spatial heterogeneity of GTR. This finding contrasts with the research of Qian [66] on all prefecture-level cities nationwide, which found that the market mechanism played a stronger role. GI consistently demonstrates strong explanatory power across different types of RBCs, stemming from the unique institutional background of RBCs. In many RBCs, particularly Growing and Declining RBCs, market mechanisms are often underdeveloped due to the “resource curse” and path dependence, creating both the space and necessity for the intervention of the government’s “visible hand”. However, this may also be related to China’s political and economic system. Compared with market entities, local governments often control key resources such as land, finances, and project approvals, and they are deeply involved in the process of economic development and transformation through the performance evaluation mechanism. This also makes the government a central actor in the process of transformation.

Furthermore, research has found that HCC is an important factor associated with spatial heterogeneity in GTR, particularly in declining cities. This conclusion is highly consistent with international studies on deindustrialization. By studying cities in six countries, including France and Germany, Gagliardi et al. [67] found that after the decline of the manufacturing sector, cities with a higher share of university-educated populations experienced faster employment recovery, and this advantage accelerated over time. This suggests that, whether facing deindustrialization or resource depletion, human capital is the core endogenous capability allowing cities to withstand structural shocks. Building on this foundation, the interaction detector analysis in this study further reveals that government policy intervention can effectively activate the green transformation pathway of human capital, thereby extending and enriching existing research.

The GTR framework proposed in this study moves beyond the one-dimensional focus of traditional concepts such as “green efficiency” or “ecological resilience” by integrating environmental performance with economic adaptability, institutional responsiveness, and social carrying capacity. More importantly, by identifying interaction effects among key factors, it reveals how policy combinations play a decisive role in enhancing transitional effectiveness, offering a more actionable theoretical perspective for understanding how resource-based cities can achieve proactive transformation.

4. Conclusions, Policy Implications, and Limitations

4.1. Conclusions

Based on evolutionary resilience theory, this study proposes an innovative theoretical framework of GTR, offering a novel research perspective for examining the green transition of China’s RBCs. By employing multiple models to analyze the spatiotemporal patterns and developmental disparities of GTR, the main findings are as follows:

(1) From 2010 to 2022, GTR increased from 0.255 to 0.300. Although the general trend is an increase, the level is still relatively low, indicating there is still great space for improvement. Among the three dimensions, adaptive capacity consistently showed the highest value (mean: 0.155). Among city types, Regenerative RBCs demonstrated the highest resilience (mean: 0.304), followed by Mature, Declining, and Growing RBCs.

(2) GTR exhibited distinct temporal and spatial variation. Temporally, GTR increased overall, but the inter-city gap first narrowed, then widened, with late-stage polarization. Spatially, a pattern of disequilibrium emerged, characterized by the clustering of low-value cities and the dispersion of high-value cities, while Mature and Declining RBCs were increasingly concentrated in the Low-and-Medium Zones. Growing RBCs shifted toward the Low-value Zones, while Regenerative RBCs clustered in the High-value Zones.

(3) Overall inequality in GTR showed a decreasing trend. Both within-group and between-group disparities declined. However, counterpart density (transvariation) increased and emerged as the primary source of inequality, contributing an average of 41.78%. This indicates a high degree of distributional overlap in GTR among cities at different developmental stages.

(4) The factors associated with the spatial differentiation of GTR exhibit distinct temporal dynamics and heterogeneity across city types. The main factors were ranked as follows: Government intervention (0.402) > Income level (0.300) > Human capital cultivation (0.230) > Openness level (0.139) > Environmental regulation (0.082). The marginal effects of these factors varied significantly across city types. Interaction analysis revealed substantial enhancement effects between factor pairs, indicating that GTR is shaped by a complex interplay of multiple factors, rather than being determined by any single variable.

4.2. Policy Implications

Considering the rapid green transition and sustainable development in RBCs, a deeper understanding of GTR is necessary to achieve regional economic and environmental goals. In accordance with the results of this study, the following policy suggestions are made to promote GTR in China’s RBCs.

(1) The empirical results show that the factors associated with the spatial differentiation of GTR in different types of resource-based cities exhibit distinct characteristics. Therefore, a differentiated policy system should be established. In Mature RBCs, policy should maintain a strong governmental role, focusing on income growth and human capital development. This includes institutional design, green investment guidance, and industrial coordination through fiscal incentives and green procurement. Promoting green consumption and investing in vocational and green skills training can further harness human capital for green innovation. Growing RBCs are transitioning from market-led to government-guided development, revealing the limitations of market mechanisms alone in enabling a full green transition. These cities should establish environmental access thresholds, provide targeted support for emerging industries, and avoid the “pollute first, remediate later” trap. Declining RBCs, despite resource exhaustion and economic slowdowns, show growing potential in human capital and increasing influence of government intervention and income levels. Priorities should include mobilizing local talent, attracting external green technology teams, and strengthening public engagement in the transition. Regenerative RBCs, now in a phase of diversified development, should focus on aligning digital, green, and financial pathways—for example, through intelligent environmental monitoring, green credit and carbon finance products, and improved emissions trading—to create synergy between regulation and market incentives. Furthermore, transvariation density emerges as the key contributor to regional disparities. This suggests that, beyond category-based guidance, policies should also be tailored to each city’s geographic context, resource endowment, and governance capacity, adopting a “one city, one strategy” approach.

(2) Resource-based cities, especially Declining RBCs, typically operate under significant fiscal constraints that limit their ability to fund green transformation locally. To overcome this, a coordinated “central–provincial–municipal” resource mobilization framework should be introduced. Centrally, Declining RBCs could be incorporated into a national special transfer program for green transition, with allocations directed toward human capital development, environmental infrastructure, and green finance risk compensation. Provincially, intra-regional fiscal transfers could be encouraged, enabling more developed areas to support Declining RBCs within the same province. Provincial green industry funds should also be consolidated and prioritized for green projects in these cities. Municipally, the use of market-oriented mechanisms should be promoted to repurpose underutilized assets (e.g., abandoned mines, idle factories) and to attract private investment into green renewal initiatives.

(3) The interactions among factors in different types of RBCs shape their optimal policy mix and sequence. In Mature RBCs, the interaction of IL × FD remains central, suggesting that financial instruments should be used to amplify market forces. Policy should therefore prioritize innovating green financial products, coupled with consumer incentives, rather than relying on standalone subsidies. In Growing RBCs, the strong interaction of PA × OL indicates that labor advantages must be converted into green capacity through institutional openness. Policies should first improve the international business environment—for instance, by streamlining foreign investment approvals—before planning industrial parks, so as to avoid misallocating labor. In Declining RBCs, the most significant interaction of GI × HCC highlights that capacity-building is a prerequisite for transformation. Policy should link skills training with local green job creation; for example, by requiring government-led ecological restoration projects to preferentially hire locally certified workers. In Regenerative RBCs, prominent interactions among ER, OL, and FD call for policies that simultaneously introduce environmental access standards and green finance support, while opening cross-border financing or green power trading opportunities for compliant enterprises.

(4) Considering that the characteristics exhibited by a certain type of city may, to some extent, stem from the shared institutional environment of its region rather than being purely a result of its development stage, it is recommended to establish a collaborative governance system characterized by “type-driven, regionally-adjusted” strategies. Under this system, the central government designs and allocates differentiated fiscal, industrial, and environmental policies according to city type. Provincial governments tailor these policies to align with their development plans, functional zoning, and resource-environmental carrying capacities, formulating detailed implementation rules and supporting measures. Municipal governments then adapt policies innovatively based on local conditions. This tiered coordination balances the precision of macro-level policies with the flexibility of local practice.

The implementation of the above policy recommendations must fully consider the fiscal capacity and governance foundations of different regions. The effectiveness of these policies may be constrained by practical conditions such as local debt pressures, the degree of marketization, and the efficiency of policy coordination. Therefore, gradual and differentiated adjustments should be emphasized during the implementation process.

4.3. Limitations

This study offers a new perspective on green transition and urban resilience in RBCs. Despite this, several limitations remain. First, the city-level environmental indicators used may be influenced by local government reporting incentives, posing a risk of data distortion. Key indicators (such as total energy consumption, text analysis-based environmental regulation intensity, and the development of emerging industries, as measured by the number of AI enterprises) are all proxy variables, and their associated measurement errors may affect the accuracy of estimation. Second, the construction of the indicator system does not consider institutional quality, social capital, political networks, or natural factors, which may affect the estimation results. This study mainly relies on cross-sectional data, making it difficult to capture potential path dependence, threshold effects, and dynamic accumulation processes in the green transition. Furthermore, although the Geographic Detector model effectively identifies factors significantly associated with spatial differentiation of GTR, its q-value measures correlation rather than causation. This study found that the strong associated factors, especially IL and GI, may have endogeneity issues. Higher incomes may support green investments, while successful green transformation could also promote economic growth and residents’ incomes. Similarly, the intensity of government intervention may be endogenous to a city’s stage of development and transformation pressure. Future research could use instrumental variable methods or quasi-experimental designs to identify the causal effects of these factors on GTR. In addition, the “type effect” observed in this study may encompass the combined influence of geographic location and regional agglomeration characteristics. Future research should develop a more spatiotemporally adaptive and refined governance framework by more precisely disentangling the independent contributions and interaction mechanisms of “type” and “regional” factors. Finally, given China’s unique governance model, the generalizability of the research conclusions is limited. Future studies should consider expanding the scope to other countries to observe national commonalities or differences in GTR.