Green Transition Assessment, Spatial Correlation, and Obstacles Identification: Evidence from Urban Governance Data of 288 Cities in China

Abstract

The green transition of China’s cities is crucial for ecology civilization realization. Based on the driver–pressure–state–impact–response (DPSIR) framework, an integrated technique for order preference by similarity to ideal solution (TOPSIS) model with entropy weight, this study achieved the comprehensive assessment of the green transition of 288 province-level municipalities and prefecture-level cities in China over 18 years from 2002 to 2019, in addition to the spatial correlations and obstacles analysis. The results indicate that major cities in China have a more significant green transition value, and the eastern region is developing fast, while the northeast region is relatively slow. There was heterogeneous spatial distribution for green transition, because of the disequilibrium sustainable development of 288 cities. Green transition has a significantly positive spatial autocorrelation in the cities of China, the high–high significant clusters greatly increased, and the main locations changed from the northeast to southeast of China. Frequent obstacles were also found, including road infrastructure construction, water resources, and the green coverage of urban built-up areas. Based on these results, several policy implications were put forward, including the optimization of environmental laws and regulations, the development of green transportation infrastructure, resource conservation and the circular economy, the establishment of a green financial system, and increasing the linkage for the green transition of different cities.

1. Introduction

The world is experiencing fast urbanization, with cities and towns accounting for 52% of the world’s population in 2010, a figure that was set to reach 57% in 2022 [1]. Rapid urbanization and industrialization bring increasing demands to cities for energy, food, and water, as well as environmental problems such as habitat fragmentation [2], biodiversity decline [3], and overloaded waste disposal systems [4]. This has led to the transformation of cities into vast centers of energy consumption and carbon emissions, which are statistically responsible for nearly 75% of global carbon emissions and 75% of global energy consumption [5].

In the past 40 years, China’s reform and opening-up policy has enabled cities to make significant contributions to national economic growth [6]. However, the long-term crude development model has inevitably brought a series of problems to Chinese cities, such as the gradual depletion of resources, deformed economic structure, rising unemployment, severe damage to the ecological environment [7,8], and underdeveloped alternative industries. As a result, many cities in China have gradually lost their investment attractiveness, which may hinder further economic development [9,10].

Ecological civilization appeared for the first time at the 17th National Congress of the Communist Party of China (CPC) in 2007. At the 18th National Congress of the CPC in 2012, the CPC Central Committee further placed the construction of ecological civilization and ecological environmental protection at the forefront of national governance. In the 13th Five-Year Plan, green and eco-friendly development appeared as one of the five new guiding principles for development leading to the future [11]. The 19th National Congress of the CPC stresses modernization with a Chinese character, a feature in harmony between man and nature [12]. In 2021, carbon neutrality and a carbon peaking policy framework further discussed a shared future for all life and a community with a shared future for humankind. These policies reflect China’s emphasis on ecological civilization and the green transition of its cities.

This study aims to reveal further to what extent the green transition has happened in China and shed light on how China’s sustainable development policy affects it. Under the DPSIR (drivers–pressures–state–impact–responses) framework, an integrated GIS (Geographic Information System) with entropy-weight-based TOPSIS (technique for order preference by similarity to an ideal solution) model, this study extensively evaluates the green transition performance of 288 province-level municipalities and prefecture-level cities in China from 2002 to 2019, in addition to the spatial correlations. Although there are some previous studies focused on this topic [13], for example, on the industry scale [6] and district scale [14], or focused on some aspect of green development [15], our study has the most comprehensive coverage of all the current 288 province-level municipalities and prefecture-level cities in China during the 19-year period, which is the most crucial period for China’s development [16]. The characteristics of the spatiotemporal evolution of the 288 cities’ green transition are also clarified, which shed a light on China’s urban governance and sustainable development. This study is organized as follows: after the introduction, a literature review section presents the recent development of related studies; the method section talks about the methods and data sources used in this work; the results section then offers the detailed findings of this article; and finally, policy implementations and conclusions are made based on all the sections above.

2. Literature Review

2.1. DPSIR Framework

Since its first proposal from PSR (pressure–state–response) and DSR (driver–state–response) models in 1993 [17], DPSIR has been widely used as a tool for assessing the sustainability of social–ecological systems [18], covering a wide range of environmental issues including urban ecosystems [19,20], ecotourism [21], marine and coastal management [22], greenhouse agriculture policies [23], and desertification [24]. It has been applied by multiple international organizations such as the Organization for Economic Cooperation and Development (OECD), the European Environment Agency (EEA), and the United Nations Environment Program (UNEP) [25]. These scholars have used the characteristics of DPSIR to reveal the relationship between the environment and the economy.

Based on a socialist analytical perspective and examples of biodiversity, a critical review of the theoretical underpinnings is provided [26], through which it is argued that the driver–pressure–state–impact–response (DPSIR) framework sometimes leans on the side of conservationists compared to the side of other perspectives. The model must improve in establishing good communication between researchers, stakeholders, and policymakers. Another study argued that the DPSIR model can structure the communication between scientists and end-users in environmental information [27]. However, it needs to be more analytical and accurate in the diversity dimensions of uncertainty and causality inherent in complex environmental and socioeconomic aspects. The authors reconstruct the model with a complex institutional methodology to overcome the shortcomings of the original DPSIR framework. But in this work, we still use the traditional DPSIR framework.

2.2. TOPSIS Model with Entropy Weight

The concept of entropy was initially introduced into Information Theory by C.E. Shannon, which evaluates disorder and chaos and reveals the degree of fuzziness reflected by the varying extent of the indicators [28]. While TOPSIS is one of the fundamental methods in the multiple attribute decision-making domains and has been immensely popular in applications and as a foundation for numerous method developments [29], the combination of the TOPSIS method with entropy weight provides a unique non-parametric, mathematically based multi-criteria decision-making method to rank samples in accordance to their performance. It has been used to measure the maturity of the carbon market in China [30], the financial performance of a real estate company in Borsa Istanbul [31], crisis management systems in petrochemical industries [32], the ecological security of marine ranching in Yantai, China [33], etc.

2.3. Studies in China’s Industrial Transformation and Green Transition

The topic we are discussing is a rising one. A couple of studies are already discussing the industrial transformation [34], market structure [35], green credit policy [36], and green transition of cities in China [14], especially for resource-based ones whose lifecycle followed a term called Resource Curse [20,37,38,39,40,41,42]. Xu et al. (2021) assessed the sustainability level and screened obstacles of 26 cities in the Yangtze River Delta (YRD) with the combined method of GIS–entropy–TOPSIS [13]. However, our study covers the whole part of China, with the most comprehensive coverage of province-level municipalities and prefecture-level cities and the most extended period to date, which can draw the most complete conclusions.

3. Method

This paper studies green transition assessment, spatial correlation, and obstacles identification with the data of urban governance data of 288 cities in China. With the DPSIR scheme, an assessment index system was put forward, and some methods were combined and used, including entropy–TOPSIS analysis, obstacles analysis, and spatial correlation. Figure 1 shows the methodology in each part.

3.1. Data Sources

For the data collection of the comprehensive study, we prioritize adopting official statistics and academic reports, which have to be collected from different sources. In this paper, the data are obtained from China Statistical Yearbooks on Environment, China Statistical Yearbooks for Regional Economy, China City Statistical Yearbooks, Chinese Energy Statistical Yearbooks, China Population and Employment Statistical Yearbooks, official documents of the United Nations Environment Program (UNEP), and various governmental environmental notices. The panel data for China’s 288 province-level municipalities and prefecture-level cities in 31 Chinese provinces and municipalities are used. To eliminate the influence of price factors, 2002 is taken as the base period, and deflation for all the variables related to value is uniformly conducted. Also, the interpolation method is employed to complete the missing data, while outlier processing is minorized mean at the 0.5% level.

In addition, due to data availability, the data from 2002 to 2019 are chosen, and six periods can be classified for green transition development: 2002–2004, 2005–2007, 2008–2010, 2011–2013, 2014–2016, and 2017–2019. Accordingly, 2004, 2007, 2010, 2013, 2016, and 2019 are chosen as the sample years for the case study (Figure 2).

Furthermore, in order to scientifically assess the green transition performance, the socioeconomic development status of different regions should be considered. According to policies and implementations, including the “Several Opinions of the Central Committee of the Communist Party of China and the State Council on Promoting the Rise of the Central Region”, the “Implementation Opinions of the State Council on Several Policy Measures for Western Development”, etc., we divided China’s economic regions into four major regions: eastern, central, western, and northeastern. All the details can be found in Table S1.

3.2. TOPSIS Model with Entropy Weight

The specific process of building the assessment model for the green transition of cities is similar to other TOPSIS models with entropy weight such as Alao et al. (2020), Xu et al. (2021), and Wei et al. (2023) [13,43,44], and as follows:

(1)

Construction of a standardized evaluation matrix

The original evaluation index matrix for the carrying capacity of cities is set as follows:

$$V=\left[\begin{array}{ccc}{v}_{11}& \cdots & v_{1n}\\ ⋮& \ddots & ⋮\\ v_{m1}& \cdots & v_{mn}\end{array}\right]$$

(1)

We can use the normalized method to process the original data to obtain the standardized evaluation matrix. The processing method is shown in Formulas (2) and (3). Therefore, the standardized matrix is obtained in Formula (4).

Normalization of positive indicators:

$$r_{ij}=\frac{v_{ij}}{\mathit{max}\left(v_{ij}\right)}$$

(2)

Normalization of negative indicators:

$$r_{ij}=\frac{min\left(v_{ij}\right)}{v_{ij}}$$

(3)

$$R=\left[\begin{array}{ccc}{r}_{11}& \cdots & r_{1n}\\ ⋮& \ddots & ⋮\\ r_{m1}& \cdots & r_{mn}\end{array}\right]$$

(4)

In the above formulas, V is the initial evaluation matrix, $V_{ij}$ is the initial value of the j year of the i index, R is the standardized evaluation matrix, $r_{ij}$ is the standardized value of the j year of the i index, i = 1, 2, …, m, m is the number of evaluation indicators, j = 1, 2, …, n, and n is the evaluation year.

(2)

Establishment of evaluation matrix based on entropy weight

The entropy weight method is a technique for calculating indicator weights that objectively assesses indicator values. To prevent analytical problems brought on by too slight changes in indicator selection, it can reflect implicit information from data and improve the significance of and difference in indicators while comprehensively reflecting a variety of information. The concept is that the evaluation object is more significant and, therefore, has a more considerable weight if its value differs from that of the object on a particular index [45].

The calculation steps of the entropy weight method are as follows:

$$w_i=\frac{1-H_i}{m-{{\displaystyle \sum }}_{i=1}^m{H}_i}$$

(5)

Among them, $H_i=-k{\displaystyle \sum }_{j=1}^n{f}_{ij}ln{f}_{ij}$; $f_{ij}=\frac{r_{ij}}{{{\displaystyle \sum }}_{j=1}^n{r}_{ij}}$; $k=\frac{1}{ln\mathrm{n}}$.

In the above formulas, $r_{ij}$ represents the standardized value of the j year of the i index, i.e., the data of j year under the i evaluation index, i = 1, 2,…, m; j = 1, 2,…, n; m is the evaluation index; n is the number of evaluation years; $r_{ij}$ is the standardized data matrix of the original index data matrix (using extremum method); and $H_i$ is the entropy of the index i. $f_{ij}$ is the proportion of the standardized value of the index in the j year under the index of item i in the whole evaluation year series; k is the Boltzmannn constant; and W is the weight value of the index of item i and meets $0\le w_i\le 1$ and ${{\displaystyle \sum }}_{i=1}^m{w}_i=1$.

Using the concept of weighting and the entropy weight $w_i$, the weighted normalized evaluation matrix Y is built to increase further the objectivity of the assessment matrix of the carrying capacity of cities. The following is the concrete calculation formula:

$$Y=\left[\begin{array}{ccc}{y}_{11}& \cdots & y_{1n}\\ ⋮& \ddots & ⋮\\ y_{m1}& \cdots & y_{mn}\end{array}\right]=\left[\begin{array}{ccc}{r}_{11}·w_1& \cdots & r_{1n}·w_1\\ ⋮& \ddots & ⋮\\ r_{m1}·w_m& \cdots & r_{mn}·w_m\end{array}\right]$$

(6)

(3)

Determination of ideal solutions

Let $Y^{+}$ as the maximum value of index I in the evaluation data in j year, i.e., the most preferred scheme, be called the positive ideal solution; and let $Y^{-}$ as the minimum value of index i in the evaluation data in j year, i.e., the least preferred scheme, be called a negative ideal solution. The calculation methods are as follows:

$$Y^{+}=\left\{\underset{1\ll i\ll m}{\max }{y}_{ij}\right|i=1,2,\dots ,m\}=\left\{y_1^{+},y_2^{+},\dots ,y_m^{+}\right\}$$

(7)

$$Y^{-}=\left\{\underset{1\ll i\ll m}{\min }{y}_{ij}\right|i=1,2,\dots ,m\}=\left\{y_1^{-},y_2^{-},\dots ,y_m^{-}\right\}$$

(8)

(4)

Distance calculation

Let $D_j^{+}$ be the Euclidean distance between index i and $y_i^{+}$, and $D_j^{-}$ be the Euclidean distance between index i and $y_i^{-}$. The calculation methods are as follows:

$$D_j^{+}=\sqrt{{{\displaystyle \sum }}_{i=1}^m{\left(y_i^{+}-y_{ij}\right)}^2}$$

(9)

$$D_j^{-}=\sqrt{{{\displaystyle \sum }}_{i=1}^m{\left(y_i^{-}-y_{ij}\right)}^2}$$

(10)

In the formulas, $y_{ij}$ is the normalized value weighted in the j year of the index I, and $y_i^{+}$ and $y_i^{-}$ are the most preferred and least preferred values of index i in the n years, respectively.

(5)

Computing the closeness between the evaluation object and the ideal solution

The closeness of the TOPSIS model is generally referred to as the closeness of the carrying capacity of cities in the j year. The larger the T_j is, the closer the carrying capacity of cities is to the optimal carrying capacity in that year. When T_j = 1, the carrying capacity of that city is the lowest. In this paper, the degree of closeness is used to express the carrying capacity of cities. According to the degree of closeness each year, the carrying capacity of cities can be judged, and the order of advantages and disadvantages can be determined. The calculation method is shown as follows:

$$T_j=\frac{D_j^{-}}{D_j^{+}+D_j^{-}}$$

(11)

3.3. Calculation of the Moran’s I Statistic

The Moran’s I statistic was used to identify spatial clusters of statistically significant high and low values [46]. The formula was as follows:

For global Moran’s I:

$$I=\frac{{{\displaystyle \sum }}_{i=1}^n{{\displaystyle \sum }}_{j=1}^n{w}_{ij}\left(x_i-\overline{x}\right)\left(x_j-\overline{x}\right)}{s^2{{\displaystyle \sum }}_{i=1}^n{{\displaystyle \sum }}_{j=1}^n{w}_{ij}}$$

(12)

For Anselin Local Moran’s I:

$$I_i=\frac{\left(x_i-\overline{x}\right){{\displaystyle \sum }}_{j=1}^n{w}_{ij}\left(x_j-\overline{x}\right)}{s^2}$$

(13)

x_i and x_j are the spatial unit observations, and w_ij is the spatial weight, $s={{\displaystyle \sum }}_{i=1}^n{{\displaystyle \sum }}_{j=1}^n{w}_{i,j}$. In addition, if global Moran’s I > 0, Anselin Local Moran’s I was divided into four types: high–high cluster, high–low outlier, low–high outlier, and low–low cluster, in which high and low represent the state of the I_i value of the local area and z_i value in which $z_i=x_i-\overline{x}$.

3.4. Degree of Obstacle

In order to further explore and identify the obstacles restricting the green transition of cities, improve the level of economic system construction, and provide a decision-making reference for formulating targeted development policies, we introduced the degree of obstacle model. The degree of the obstacle model determines the impact of each obstacle factor on the green transition, and its calculation formula is as follows [44]:

$$P_{ij}=\frac{\left(1-x_{ij}\right)\times w_j}{{{\displaystyle \sum }}_{j=1}^n\left(1-x_{ij}\right)\times w_j}$$

(14)

P_ij represents the degree of obstacle of the factor indicator j to the level of green growth; x_ij indicates the normalization of the element indicator; w_ij represents the weight of the feature indicators; and n is the number of feature indicators.

4. Results

4.1. Calculated Weights of Indicators

Based on former research [47], because of the accessibility of statistical data, only data from 2002 to 2019 were introduced for the situation in China’s cities. With the DPSIR model, 29 indicators were chosen, and their entropy weights were calculated. The details can be found in

Table 1. Indicators used in this study and their weights.

	No.	Indicator Name	Code	Weight	Direction
Driver	1	GDP per capita (RMB)	DX1	0.022	Positive
	2	Per capita water resources (cubic meters per person)	DX2	0.043	Positive
	3	Urban land area per capita	DX3	0.038	Positive
	4	Population density (people/sq km)	DX4	0.039	Negative
Pressure	5	Industrial electricity consumption (10,000 kWh)	PX1	0.082	Negative
	6	Total water supply in cities (10,000 tons)	PX2	0.054	Positive
	7	Electricity consumption in urban households (10,000 kWh)	PX3	0.035	Negative
	8	Number of registered unemployed persons in urban areas at year-end (person)	PX4	0.022	Negative
State	9	Gross industrial output value above designated size (current price, 10,000 yuan)	SX1	0.051	Positive
	10	Actual use of foreign capital in the current year (US $10,000)	SX2	0.076	Positive
	11	Number of regular secondary schools	SX3	0.001	Positive
	12	The proportion of added value of tertiary industry to GDP (%)	SX4	0.005	Positive
	13	The built-up area of the city	SX5	0.039	Positive
	14	Urbanization rate (%)	SX6	0.010	Positive
Impact	15	Industrial soot emissions (tons)	IX1	0.053	Negative
	16	The discharge of industrial wastewater (10,000 tons) includes untreated and treated	IX2	0.050	Negative
	17	Industrial sulfur dioxide emissions (tons)	IX3	0.064	Negative
	18	CO2 (million tons)	IX4	0.031	Negative
Response	19	Proportion of employees in the tertiary sector (percentage)	RX1	0.005	Positive
	20	Total investment in fixed assets (10,000 yuan)	RX2	0.047	Positive
	21	Science expenditure (10,000 yuan)	RX3	0.090	Positive
	22	Education expenditure (10,000 yuan)	RX4	0.036	Positive
	23	Comprehensive utilization rate of general industrial solid waste (percentage)	RX5	0.0066178	Positive
	24	Centralized treatment rate of sewage treatment plants (percentage)	RX6	0.0079192	Positive
	25	Harmless treatment rate of domestic waste (percentage)	RX7	0.007	Positive
	26	Green coverage of urban built-up areas	RX8	0.005	Positive
	27	Urban green space per capita (sq m)	RX9	0.033	Positive
	28	Buses per capita (10,000 people)	RX10	0.019	Positive
	29	Road area per capita (sq m)	RX11	0.018	Positive

and Figure 3. The 29 indicators cover drivers–pressures–state–impact–responses, a total of five aspects within the DPSIR framework. Most of the calculated weights were in the range of 0.005–0.090, suggesting that this framework is primarily comprehensive, and no dominating weights were found. Also, 22 of the indicators were positive, and 7 of them were negative.

4.2. Green Transition Score of 288 Cities

The green transition indices of cities from 2002 to 2019 were calculated based on the entropy–TOPSIS method section, as shown in Figure 4. There was heterogeneous spatial distribution for green transition, because of the disequilibrium sustainable development of 288 cities. We can see from Figure 5 that in the northeast region, Harbin, Daqin, Dalian, Shenyang, Changchun, and Heihe experienced the most significant extent of green transition during the study period; in the eastern region, Sanya, Shanghai, Beijing, Tianjin, Guangzhou, and Shenzhen had the most extent of green transition; in the central region, Hefei, Zhangjiajie, Zhenzhou, and Changsha had the most significant transition; and in the western region, Lijiang, Dingxi, Chengdu, Lhasa, and Chongqin had the most significant green transition index value.

Similar to the former research, large and very large cities have a high degree of sustainable development [48]. Our results indicated that significant cities usually had more green transitions. Moreover, this could be explained in several ways:

(1)

Resource advantage: Major cities usually have more resources, including human, financial, and technical support. They can invest more money and workforce to drive the green transition and attract more professionals and innovators to participate in green development efforts. For example, as the economic capital of China, Shanghai ranks top in green transition during the 18 years.

(2)

Awareness and demand: Major cities usually have higher environmental awareness and demand for sustainable development. These cities face more environmental problems, such as air pollution, waste disposal, and energy consumption. Due to high population density and economic activity, these cities are more concerned with improving residents’ quality of life and environmental sustainability.

(3)

Policy support: Government policy support for implementing green transitions in major cities is often more active. These policies include encouraging renewable energy development, promoting energy efficiency and emission reduction, and providing financial support for green infrastructure development. Active government involvement and policy support are essential in promoting green transition. Beijing was once notorious for haze, but after the Regulations on the Prevention and Control of Air Pollution were issued, the air quality was increased obviously, which is an important aspect of green transition.

(4)

Opportunities for cooperation: There are more opportunities for cooperation between major cities. They can share experiences, exchange best practices, and work together to solve common problems. Such cooperation facilitates the rapid development of green technologies and innovations, accelerating the transition. In summary, significant cities can better address environmental challenges, promote sustainable development, and provide role models and lessons for other cities.

In terms of regional average value, their performance can be seen in Figure 6. In general, all four regions had an increasing trend. The eastern region has the most significant increase, while the northeast region has fallen and stagnated. It is noticeable that except for the northeast region, the other three regions have had a sudden increase since 2014, which could be attributed to the effect of China’s Sustainable Development Plan of National Resource-based Cities Act at the end of 2013, which involved 262 resources-based cities. In addition, several reasons exist for the relatively slow green transition of cities in the northeast region, including economic structure, few resource advantages, population mobility, employment problems, etc. [49]. However, cities in the northeast are responding positively and stepping up their green transformation efforts as the country pays great attention to and promotes policies for sustainable development. For example, they gradually realize green transformation by increasing environmental governance, encouraging technological innovation, and guiding industrial upgrading [50]. Our results were similar to the research of Yu (2008), who found the development performance ranked from eastern to central to western cities in geographical location [48].

Regarding spatial relationships, the distribution of green transition in 288 cities was divided into four levels, as shown in

Table 2. Division of the level of green transition.

Value Range	Status	Level	Feature Description
0~0.2	Bad	I	Low level of green transition
0.2~0.4	Normal	II	The level of green transition is normal and generally meets the needs of people
0.4~0.6	Good	III	Good green transition level
0.6~1.0	Excellent	IV	Green transition level is very good

. In addition, the study period was divided into six periods and the green transition divisions at the end of each period are shown in Figure 4.

Generally, it can be seen in Figure 6 that most cities fell into level I, indicating that the green transition in China is still at a low extent during the study period. Also, some cities fell into level II, while only a few are in the division of level III, and barely any are in the division of level IV.

In terms of regional characteristics, cities in the northeast region were mostly in level I, except Heihe in 2004, and Dalian and Shenyang in 2010 and 2013; in the eastern region, many cities in the Yangtze River area, including Suzhou, Hangzhou, and Ningbo, had a normal or good status of green transition, i.e., level II and III, while Shanghai has fallen into level IV since 2007; in the Central region, some cities, including Wuhan, Zhengzhou, and Changsha, had a normal or good status; and in the western region, a few cities, including Lhasa, Chengdu, Chongqin, Jiuquan, and Lijiang, had a normal, good, or even excellent status of green transition, i.e., level II, level III, and level IV. In terms of the city scale, the green transitions of cities were mostly aggregated in big cities, with a clear trend since 2007.

While on a national point of view, the relatively advanced cities in terms of green transition are more dispersed in the four regions rather than aggregated. No other significant geography-related factors were found.

4.3. Spatial Characters Analysis

Based on the green transition index calculated from the above, AcGIS 10.8 was selected for the spatial analysis. The results of the global Moran’s I from 2002 to 2019 show that there is spatial correlation between green transition in different cities in China, and the details can be found in

Table 3. Global Moran’s I indicator in this study.

Year	Moran’s I	Z	p
2004	0.134	3.533	<0.001
2007	0.140	3.983	<0.001
2010	0.169	4.472	<0.001
2013	0.164	4.340	<0.001
2016	0.163	4.297	<0.001
2019	0.209	5.486	<0.001

. All the p-values were less than 0.01, with the Z value in the range of [−2.58, 2.58], showing that the urban green transition has a significantly positive spatial autocorrelation, at the confidence level of 99% [14]. In addition, the value of the global Moran’s I varied from 0.13 to 0.21, showing that the spatial correlation of the green transition of 288 cities was weak (see Table 3), and increased steadily during the 18-year period. And all the spatial correlations are irregular and unbalanced, but between different cities.

Similarly, the study period was divided into six periods and the green transition divisions at the end of each period are shown in Figure 7. In 2004, 2007, 2010, 2013, and 2016, the Local Moran’s I show four types of agglomeration, including high–high, low–high, low–low, and high–low. High–high agglomeration refers to a city with a high green transition index value that is surrounded by other corresponding high-value cities. Low–high agglomeration refers to a city with a low green transition index value that is surrounded by four corresponding high-value cities. Low–low agglomeration refers to a city with a low green transition index value that is surrounded by four cities that are corresponding low-value cities. High–low refers to a city with a high green transition index value that is isolated by four cities with a corresponding low value.

More specifically, in 2007 and 2013, there was no spatial agglomeration for the green transition of different cities. Because most cities have a different green transition paradigm, they seldom have interaction, showing that there was no aggregation effect on the green transition of these cities, and that the radiation effects of the good green transition performance cities have less benefit for adjacent cities. As shown in

Table 4. Clusters and outliers in Anselin Local Moran’s I results.

Year	Agglomeration Type	Number of Cities	Cities
2004 (53)	LL	31	Shanghai; Shangrao; Dongwan; Lincang; Ulanqab; Wuhai; Xinyang; Nantong; Shamen; Jilin; Tianjin; Xiaogan; Ningde; Anshun; Laibin; Liuzhou; Zhuzhou; Guilin; Wuzhou; Yulin; Wuhan; Yongzhou; Hanzhong; Shantou; Shanwei; Jiangmen; Zhanjiang; Laiwu; Xining; Ganzhou; Handan
	HH	7	Panjin; Shaoxing; Zhongqing; Suizhou; Qingdao; Anshan; Huangshi
	HL	6	Neijiang; Wenzhou; Maoming; Tongliao; Shaoyang; Jinhua
	LH	9	Xianning; Loudi; Yichang; Yichun; Xian; Hezhou; Zhangzhi; Fangchenggang; Longnan
2010 (51)	LL	28	Ulanqab; Wuhai; Shamen; Shuangyashan; Changde; Guangyuan; Yanan; Kaifeng; Zhangjiajie; Wuzhou; Shantou; Heyuan; Taian; Taizhou; Baise; Yiyang; Suzhou; Maoming; Laiwu; Xining; Chifeng; Dazhou; Tongliao; Zunyi; Xingtai; Shaoyang; Zhengzhou; Jinhua
	HH	10	Sanya; Sanmenxia; Daqing; Weihai; Loudi; Yichun; Xuancheng; Panjin; Shaoxing; Suihua
	HL	7	Shangrao; Tianjin; Luzhou; Wenzhou; Zigong; Quzhou; Xiangyang
	LH	6	Zhongshan; Linyi; Anyang; Baoji; Xuzhou; Xian
2017 (8)	LL	3	Bozhou; Nantong; Xining
	HH	3	Loudi; Yichun; Shaoxing
	HL	2	Neijiang; Baotou
2019 (55)	LL	24	Shanghai; Lincang; Lishui; Urumqi; Bozhou; Baoshan; Xinyang; Lanzhou; Shiyan; Nanjing; Nanchong; Nantong; Nanyang; Shamen; Shuangyashan; Jian; Jilin; Lüliang; Hulun buir; Zhoushan; Xining; Dazhou; Zhengzhou; Zhangchun
	HH	18	Daqing; Weihai; Loudi; Yichun; Baoji; Xuancheng; Xiuzhou; Chongzuo; Pingdingshan; Xuzhou; Mudanjiang; Yulin; Yuxi; Baishan; Panjin; Qinhuangdao; Shaoxing; Suihua
	HL	12	Ulanqab; Neijiang; Baotou; Wuzhong; Tianjin; Jincheng; Hanzhong; Jiangmen; Luzhou; Wenzhou; Zigong; Quzhou
	LH	1	Linyi

, in 2004, the results showed 53 cities achieved significant spatial agglomeration, comprising 31 low–low significant clusters, 7 high–high significant clusters, 6 high–low significant clusters, and 9 low–high significant clusters. Most of the high–high and low–high significant clustering areas were located in the western regions, while the low–low significant clustering areas were located in the middle regions. In 2010, the results showed 51 cities achieved significant spatial agglomeration, comprising 28 low–low significant clusters, 10 high–high significant clusters, 7 high–low significant clusters, and 6 low–high significant clusters. The aggregation spatial pattern is similar to 2004. In 2017, the results showed only eight cities achieved significant spatial agglomeration, comprising three low–low significant clusters, three high–high significant clusters, and two high–low significant clusters. The aggregation was mainly in the east of China. In 2019, the results showed 55 cities achieved significant spatial agglomeration, comprising 24 low–low significant clusters, 18 high–high significant clusters, 12 high–low significant clusters, and 1 low–high significant cluster. The aggregation spatial pattern is similar to 2004. The high–high significant clusters greatly increased, and mainly in the east and south of China; the number of high–high significant clusters also increased, indicating more and more cities have positive effects on adjacent cities with spatial correlation.

4.4. Degree of Obstacles

In terms of the degree of obstacles, the top 10 degrees of obstacles and their values can be seen in Figure 8. From the figure, we can see no dominating obstacle in the process of green transition. All the top degrees of the obstacle were less than 0.1. In addition, we can also see that none of the obstacles can remain constant in the first three places; they are almost changing each year. Many of the obstacles change rapidly up and down every year, indicating the complicated situation in the process of green transition. To illustrate the result more clearly, the frequencies of indicators in the top 10 degrees of obstacles are shown in Figure 9. RX11, DX2, and RX8 are the most frequent top three obstacles (Figure 9), indicating the transport infrastructure development, water resource, and green coverage in urban areas should be focused on in China. Similar to our research, as a model example of sustainability, Stockholm in Sweden became the first European Green Capital, and one of the most effective instruments is enlarging the area of public green space [15].

In addition, all obstacles appeared at least four times, indicating that all obstacles mentioned as indices can affect the improvement in the green transition of cities in China.

5. Policy Implementations

Based on the current results, we found that most cities in China had a certain extent of green transition, but the current transition needs to be more thorough. Also, they mainly focused on several major cities. In addition, the spatial correlation is weak, and the temporal change is huge, with no obviously fixed degree of obstacles. Thus, we have the following policy recommendations.

Because of the accessibility of statistical data, only data from 2002 to 2019 are introduced in our research; the latest data cannot be found, which could affect the timeliness of the policy implementation we put forward. Additionally, the statistical calibration is different in countries; some calculating methods for indicators in this study did not dovetail with international practices. So, because of that, it is hard to give comparative research in an international context.

5.1. Optimize the Environmental Protection Laws and Regulations

According to our results, most cities we studied were in a rapidly changing state, meaning that their regulation needs to be improved. We should strengthen the formulation and implementation of laws and regulations. The regulations should include measures to establish a sound regulatory and assessment mechanism for urban green transformation to ensure the implementation and effectiveness of policies and make timely adjustments and improvements to policy measures in light of the actual situation. China should strictly enforce the law and impose severe penalties for environmental violations in order to ensure that the urban environment is effectively protected.

5.2. Promote the Development of Public Transportation and Green Infrastructure

According to our results, transport infrastructure development is the second important obstacle for green transition. Thus, we should increase investment in the public transportation system, improve public transportation facilities and service levels, increase convenience and comfort, reduce the use of personal automobiles, and reduce traffic congestion and tailpipe emissions. For example, nature-based green infrastructure (GI) should be considered to promote the development of urban green transition and mitigate the urban issues resulting from rapid urbanization [51], including the constructing of a GI network. Also, we should promote the urban planning and construction of green cities in line with ecological principles, focusing on protecting and restoring the ecological environment, promoting green building technologies and materials, improving the efficiency of building energy utilization, and reducing the generation of construction waste.

5.3. Develop Resource Conservation and Circular Economy

China should further promote urban residents and enterprises to adopt measures to conserve resources, such as water and electricity conservation, waste recycling, etc., and encourage a circular economy model to minimize the consumption and waste of resources. In addition, cities should be encouraged to develop industrial parks to realize a circular economy, reduce carbon emissions, reserve resources, and improve energy utilization efficiency [52].

5.4. Establish a Green Financial System Radiating from Big Cities to Small Cities

According to our results, major cities had a relatively advanced state of green transition. Effective and efficient economic incentives are crucial for green development [52]. China should take advantage of this, encourage banks and financial institutions to develop green financial products, support environmentally friendly projects and enterprises, and provide financial support and policy favoritism for green transformation.

5.5. Increase Linkage for Green Transition of Different Cities

Our study revealed that the spatial correlation of the green transition of 288 cities was weak but increased steadily during the 18 years. And all the spatial correlation is irregular and unbalanced, but between different cities, and more and more cities have positive effects on adjacent cities with spatial correlation. Therefore, it is necessary to encourage different municipal government participation so that a broader range of cities can be involved in urban green transition. Therefore, the central government should play a key role in the creation of linkage on urban governance experiences for green transition. For instance, an information platform should be convened regularly so that different governments can share their transition information and experiences with others and create much more cooperation opportunities. Based on these, collaborative governance will be realized, and a better urban green transition performance will be presented.

6. Conclusions

China has to balance urbanization and the environmental challenges of green transition on a city level. It is therefore important to carry out a comprehensive assessment of the green transition level. With the DPSIR framework, the integrated GIS with entropy-weight-based TOPSIS model, this study evaluates the score of the green transition of 288 prefecture-level. Spatial correlation and obstacles identification have also been specifically detailed. The conclusions made are as follows:

Firstly, we discovered that most Chinese cities had undergone some degree of green transition; nevertheless, more work has to be undertaken on the current transition. There was heterogeneous spatial distribution for green transition, because of the disequilibrium sustainable development of 288 cities.

Secondly, many major cities in China have a more significant achievement, and the eastern region is developing fast in terms of green transition, while the northeast region is relatively slow.

Thirdly, the urban green transition has significantly positive spatial autocorrelation in different cities in China, and the high–high significant clusters greatly increased, and the main locations changed from northeast to southeast of China.

Finally, the most frequent obstacles are mainly in the area of road infrastructure construction, water resources, and the green coverage of urban built-up areas.

Based upon these results, several policy recommendations were raised, including the formulation of laws and regulations, promotion of green transportation infrastructure, development of resource conservation and a circular economy, establishing a green financial system, and increasing the linkage for the green transition of different cities. Hopefully, the results of this study could contribute to promoting sustainable development, combating climate change, and shed light for policymakers to lead cities, businesses, and individuals toward a sustainable development path.