Measurement and Spatial-Temporal Evolution Characteristics of Low-Carbon Cities with High-Quality Development: The Case Study of the Yangtze River Economic Belt, China

Abstract

Carrying out measurements of low-carbon city development levels and exploring their core driving factors are focuses of attention in the field of building sustainable low-carbon cities (LCC). Previous studies have mainly focused on the national or provincial level, ignoring the problem of heterogeneity among different cities, and the consideration of the influencing factors of low-carbon cities has not been comprehensive enough. Given this, the authors of this paper selected 107 cities in the Yangtze River Economic Belt from 2006 to 2019, constructed a general comprehensive index system for measuring the high-quality development level of low-carbon cities at the prefecture-level city level, and explored the spatial and temporal evolution trends and core drivers of the high-quality development level of low-carbon cities in the Yangtze River Economic Belt using the CRITIC–VIKOR method and an ensemble learning algorithm. The empirical results showed that most of the cities in the Yangtze River Economic Belt showed an overall upward trend in the level of high-quality development and a certain degree of “central collapse” in the spatial distribution. In addition, this paper further confirms that industrial structure is the most central driver of low-carbon urban development, the importance of urban carbon emissions and the level of science and technology innovation are gradually increasing, and a certain aggregation effect is formed in space that has led to a significant urban “siphon effect”. These results provide new evidence on the spatial and temporal evolution of the high-quality development of low-carbon cities in China and can help authorities formulate more targeted policies and strategic plans to enhance the high-quality development of low-carbon cities.

1. Introduction

According to statistical results, more than 70% of global CO₂ emissions come from cities [1]. As the core activity of promoting national economic and industrial development, building low-carbon cities (LCCs) has important strategic significance in the achievement of global sustainable development goals (SDGs) [2]. All along, China’s CO₂ emissions and related issues such as reaching their peak have been the focus of the international community [3]. For the first time ever in 2016, the Chinese government proposed reaching their peak in CO₂ emissions by 2030 and to cut CO₂ intensity per unit of emission by 60–65 percent [4]. China’s overall CO₂ emissions have climbed 318.87% since 2000 as of 2019 [5]. In light of the aforementioned context, how can one embrace adopting effective control measures to reduce domestic CO₂ emissions is now a top priority for the Chinese government. Furthermore, mega-city clusters with higher urban economic volumes and industrialization levels will play key roles in accomplishing the overall CO₂ reduction goal [6].

In 1997, the Kyoto Protocol included carbon dioxide emissions as a legally binding provision in the global greenhouse gas emission reduction plan for the first time, as well as quantifying the global greenhouse gas emission limits. In 2016, the Paris Agreement proposed a warming target of “within 2 °C”. By 2021, global surface temperatures rose by 1.1 degrees Celsius compared to the end of the 19th century [7], a difference of only 0.4 degrees Celsius compared to the Paris Agreement’s target of “preferably 1.5 degrees Celsius”. An increase of 2 °C would double the probability of extreme weather compared to 1.5 °C [8]. This is certainly more than most countries expected, and it means that we are left with less time to reduce emissions. Global carbon emissions fell by 5.7% year-on-year in 2020 due to the COVID-19 epidemic but rapidly rebounded in 2021 and exceeded the 2019 total [9]. This is not a positive response to the global emission reduction plan. In addition, at COP26 at the end of 2021, in addition to submitting more nationally owned contributions (NDCs), politicians proposed a new plan to gradually reduce the use of fossil energy through government means—the first time the conference publicly proposed a reduction in the proportion of fossil energy [10].

Currently, fossil energy combustion, such as that of natural gas, liquefied petroleum gas, and coal, is still the main source of CO₂ emissions. Each year, more than 10% of the world’s population prematurely dies due to atmospheric pollution caused by fossil energy combustion, with China accounting for 38% of the total [11]. Accordingly, it is imperative to increase the share of non-fossil energy sources, such as hydrogen energy [12]. The use of hydrogen to generate electricity seems to be an effective solution to achieve the storage of 100% renewable energy (RE), but it is also a great challenge to ensure a high quality of cleanliness throughout the process of producing hydrogen [13]. At present, hydrogen can be broadly classified into three categories according to the different raw materials and pathways used to prepare it. The first category is the use of fossil energy combustion to produce hydrogen, commonly known as gray hydrogen; this method is currently the mainstream hydrogen production method [14]. The second category also mainly uses fossil fuels to produce hydrogen, but carbon capture, carbon utilization, and storage (CCUS) technologies are used in the production process to capture and collect the produced carbon dioxide; the hydrogen produced by this process is also known as blue hydrogen [15]. The third category is the use of renewable energy to produce hydrogen, e.g., using electrolysis of water; this is known as green hydrogen [16]. Limited by the high cost and low electrolysis efficiency, the technology is not expected to completely replace commercially available gray hydrogen in the next decade [17]. In addition, electrical energy, a more mature alternative energy source for current technology development, and emerging technologies such as electric vehicles, decarbonized electricity, and electrochemical energy storage have been widely used worldwide [18]. Compared to 2010, the price of lithium battery packs decreased by about 58% by 2018, and the total global inventory of electric vehicles exceeds three million [19]. Some scholars are currently focusing on the recycling of batteries and the sustainability of rechargeable batteries, as well as other issues related to the impact of electricity on the urban environment, to improve the reuse of batteries and reduce the environmental pollution of used batteries through technological means. In addition, other scholars are focusing on the impact of battery recycling and rechargeable battery sustainability on the urban environment [20]. It is thus clear that among the appropriate paths to explore the quality development of low-carbon cities in terms of energy, technological advancement in energy is an effective way to reduce urban carbon emissions [21]. Technological innovation contributes to the reduction in urban CO₂ emissions and can help achieve low-carbon urban development while significantly improving production efficiency [22]. This is a necessary path to further improve the quality of economic growth in low-carbon cities worldwide in the future [23].

In addition, to enhance the level of urban science and technology innovation, the research and governance form of the urban living laboratory (ULL) has emerged in developed European countries to meet the requirements of the sustainable development of low-carbon cities [24]. At present, this means has become an important way to enhance the level of sustainable urban development in developed European countries [25]. A ULL is a kind of media platform developed based on realistic urban community living scenarios, and, compared to the closed innovation of traditional enterprises, its biggest advantage lies in open innovation, which can enhance the technological innovation capacity of cities, enable more diversified technological innovation results, and improve the technological innovation levels of cities by absorbing multiple parties to participate in innovation and coordinate cooperation. However, the method still faces many challenges in the future because the value of the future outputs and results of the method cannot be precisely measured, and there may be problems such as conflicts among different stakeholders [26].

2. Literature Review

In this study, we review the status of previous research in three respects: carbon-dioxide-accounting methods, low-carbon city evaluation indicator system, and core driver exploration methodology.

2.1. Carbon-Dioxide-Accounting Methods

For the measurement of total urban CO₂ emissions, there are two types of methods based on technical means and data sources: the traditional method, based on the IPCC Guidelines for National Greenhouse Gas Inventories, relies on a linear combination of the total consumption of various energy sources from statistical yearbooks and the CO₂ emission factors from experimental measurements to obtain annual total CO₂ emission data [27]. For example, the annual CO₂ emission values of China from 2003 to 2014 were obtained by multiplying the annual total energy consumption data of natural gas, liquefied petroleum gas, and electricity of the whole society with the corresponding emission factors [28]. Additionally, with the development of science and the increasing maturity of satellite remote sensing technology, more and more scholars have started to use the remote sensing data of meteorological satellites, combined with neural networks and other frontier disciplines, to inversely estimate the CO₂ emissions of Chinese prefecture-level cities [29]. This method also allows for the real-time accounting and supervision of the net carbon flux (NCF) of each country [30]. The multi-source approach has the advantage of being timely and comprehensive compared to traditional CO₂ measurement methods. Real-time CO₂ monitoring data can help the academic community and society in general to better understand the CO₂ emissions in a specific region, as well as to observe the seasonal trends and times of day when the emission peaks are concentrated [31]. Unfortunately, most of the publicly available annual CO₂ emission data of Chinese prefecture-level cities only reach 2017 and the multi-source data estimation method has not yet fully covered the real-time CO₂ emissions of all cities in China from 2018 to 2019, which have led to the lack of timeliness of the current research on CO₂ emissions related to Chinese prefecture-level cities. Therefore, the authors of this paper accounted for the carbon emission data of 107 cities in the Yangtze River Economic Belt from 2006 to 2019 with the help of national public energy data, which enhanced the timeliness of the research data related to urban carbon emissions.

2.2. Low-Carbon City Evaluation Indicator System

For the evaluation of the quality development status of low-carbon cities, it is obvious that individual indicators cannot meet the needs for comprehensive and integrated evaluation. Therefore, a comprehensive evaluation index system needs to be constructed for measurement and analysis. Ecological environment and economic development are the dimensions that should be taken into consideration by default in the establishment of a comprehensive evaluation system for cities [32]. Considering a wider range of factors, some scholars have established a low-carbon city evaluation index system from socio-economic, energy consumption and structure, living standards, carbon emissions, ecological environment, and urban mobility, while other researchers have conducted subjective surveys on the satisfaction of residents through questionnaires to improve the existing comprehensive evaluation index system [33]. Considering that this paper is mainly focused on the macro panel data of 107 cities in the Yangtze River Economic Belt for relevant analysis and research, objective evaluation indexes were chosen to construct the comprehensive evaluation indicator system of low-carbon cities from the seven dimensions of carbon emissions [34], industrial structure [35], economic development [36], ecological environment [37], scientific and technological innovation [38], transportation and population [39], and energy consumption [40].

Furthermore, in studies related to objective evaluation, comprehensive evaluation methods have mostly been used to solve such multicriteria decision-making (MCDM) problems [41]. Common methods include the entropy value method [42], the principal component method [43], the AHP hierarchical analysis method [44], the fuzzy comprehensive evaluation method [45], and the TOPSIS superior and inferior solution distance method [46]; of these, the TOPSIS method is the most widely used. The TOPSIS method only determines the distances between each solution, the positive and negative ideal solutions, and their corresponding relative postings without considering the relative importance of these distances, while the VIKOR method not only considers the distances between each solution, the superior and inferior solutions, and their relative postings but also includes the relative importance of these distances, which makes it more objective than TOPSIS [47]. Therefore, the VIKOR method was chosen for the subsequent analysis of the level of quality development of low-carbon cities in the Yangtze River Economic Belt.

2.3. Core Driver Exploration Methodology

In terms of exploring the core drivers of carbon emissions, most scholars generally consider the factors of population, economy, technology, and energy. The methods can be broadly divided into two categories: the first category uses panel data in statistical econometric models, for which scholars usually set the carbon emission factor as the dependent variable and the core drivers to be studied as the independent variables [48,49], analyze the degree and direction of influence of each variable on the dependent variable through regression [50], and use the environmental Kuznets curve hypothesis (EKC hypothesis) to explain the intrinsic influence mechanism of low-carbon cities [51] or use spatial econometric models to analyze the dynamic distribution and intrinsic relationship of core drivers of carbon emissions in space [52]. The other category comprises factor decomposition methods, such as Laspeyres exponential decomposition [53], Divisia exponential decomposition [54], and Kaya constant decomposition [55]. The most basic one is the Kaya constant decomposition, which only considers carbon emissions, energy consumption, and economic and demographic factors while ignoring the influence of structural factors; the Laspeyres index decomposition method quantifies the impact of each influence on the total by fixing the base period level of the influence in order to determine the degree of influence of each influence factor on the total. Compared to the Kaya constant decomposition method, the Laspeyres index decomposition method considers the influence of economic and energy structure factors. The Divisia index decomposition method is used to differentiate each influence factor over time before decomposing the degree of influence of each influence factor on the total. The two improved Divisia index decomposition methods commonly used today are arithmetic mean Divisia index decomposition (AMDI) and logarithmic mean Divisia index decomposition (LMDI), of which LMDI is the most widely used for decomposing carbon emission factors because of the absence of residual values. In addition, Liu et al. (2019) pioneered the use of random forest in machine learning for the identification of key influencing factors [56], which has provided new ideas and methods for the subsequent investigation of the key influencing factors of urban carbon emissions.

However, we believe the random forest method may still have certain shortcomings. First, the construction of the indicator system is not reasonable. This may lead to a serious deviation in the probability of selecting the core drivers for different categories of primary indicators, i.e., primary indicators with a higher number of secondary indicators are more likely to be selected as core drivers and vice versa. Second, the method uses information entropy and Gini impurity as the basis for the classification of each decision tree, which is questionable because the indicator of carbon intensity is a continuous numerical form that reflects the magnitude rather than the type, so it may not be reasonable to treat this type of problem as a classification problem. Third, the choice of algorithm is relatively singular, and there has been a lack of discussion regarding the rationality of the method. Liu et al. calculated the importance of the indicator for the random forest algorithm, but they did not evaluate the effect of the algorithm or perform comparisons with other similar algorithms.

Therefore, in this study, seven primary indicators of carbon emissions, industrial structure, economic development, ecological environment, science and technology innovation, transportation and population, and energy consumption, as well as 22 secondary indicators, were selected to establish a high-quality evaluation index system for low-carbon cities and to ensure the relative balance in the number of the secondary indicators corresponding to the primary indicators of different categories in order to overcome the unfairness of the results caused by probability bias and to effectively solve the limitations of the traditional measurement models in the selection of the number of variables. Second, in terms of method selection, we combined frontier disciplines and proposed an improved method for factor-importance identification based on the establishment of an ensemble learning algorithm. Using the high-quality low-carbon city development index as the dependent variable and each influencing factor as the independent variables, a regression-like ensemble learning algorithm model was established to perform index importance calculations. In addition, four different ensemble learning algorithms were selected for training in this paper, and the model with the best training effect was chosen for the calculation of indicator importance. The method was shown to be able to further improve the robustness and rationality of the results, as well as to provide new ideas and methods for further explorations of indicator importance in the future.

2.4. Research Pathways

To address some possible limitations of the previous studies discussed above, the authors of this paper used 107 prefecture-level cities in China’s Yangtze River Economic Belt from 2006 to 2019 as the research objects, combined comprehensive evaluation models and GIS technology, and constructed a multivariate comprehensive evaluation of low-carbon cities from the seven dimensions of industrial structure, ecological environment, carbon emissions, science and technology innovation, transportation and population, economic development, and energy consumption The spatial and temporal evolution of the coordinated development of low-carbon cities and the core driving factors of 107 cities in the Yangtze River Economic Belt were explored, and the spatial differences of the core driving factors were selected and analyzed. Based on the social reality, we expand the ideas of low-carbon city construction and development in order to expect a breakthrough in the measurement of low-carbon construction status, the analysis of low-carbon level changes in cities, the exploration of the core drivers of low-carbon cities, and spatial correlation analysis (Figure 1).

3. Materials and Methods

3.1. Study Area

The cities along the Yangtze River Economic Belt (Figure 2) generated 46.6% of the country’s annual gross domestic product (GDP) [57] in 2020 as China has worked to meet its emission reduction targets. The Yangtze River Economic Belt is the largest integrated industrial base in China [58] and the lifeblood of China’s economic development, playing a decisive role in the implementation of China’s “double carbon” goal. In 2014, the Chinese government incorporated the Yangtze River Economic Belt into its national development strategy for the first time [59], and in 2016, the Chinese government proposed the establishment of the Chengdu–Chongqing City Cluster [60] in the upstream region, the Yangtze River Midstream City Cluster [61] in the midstream region, and the Yangtze River Delta City Cluster [62] in the downstream region based on the Yangtze River Golden Waterway, thus encompassing the entire Yangtze River Economic Belt. In the past five years, China has strengthened the awareness and concept of environmental protection along the Yangtze River through a series of policies and top-level strategies. As of 2021, the percentage of good water quality in the Yangtze River Economic Belt exceeded 90% [63]. In March 2021, the Yangtze River Protection Law (China’s first watershed law) became fully and officially enforced within China [64]. In addition, the Yangtze River Economic Belt is currently being built along the Yangtze River as the world’s largest green and clean energy corridor. By the end of 2021, the total annual cumulative power generation and emission reduction of the six step hydropower plants along the Yangtze River was about 216 million tons of CO₂ [63]. Thus, it can be seen that the cities along the Yangtze River Economic Belt are the core of China’s development, so the sustainable development of low-carbon cities along the Yangtze River Economic Belt has strong research significance and an important role in promoting China’s overall green quality.

3.2. Data Resources

The data used in this paper on the carbon dioxide emissions of Chinese prefecture-level cities from 2006 to 2019 were measured by referring to the method of Han et al. (2017) [28]. The calculations were carried out through the specifics of the total energy consumption of three categories of natural gas, liquefied petroleum gas, and electricity of the whole society in each city (Equation (1)), and the corresponding energy consumption data were derived from the China Energy Statistical Yearbook. Carbon dioxide emissions per capita and carbon dioxide emissions per unit were calculated, and the original data of the remaining environmental and economic subsystem-related indicators were obtained from the China Urban Statistical Yearbook and the China Urban and Rural Construction Statistical Yearbook.

$${\mathrm{CO}}_2=C_1+C_2+C_3=k{E}_1+v{E}_2+\phi (\eta \times E_3)$$

(1)

where $C_1$, $C_2$, and $C_3$ denote the CO₂ emissions of urban natural gas, LPG, and electricity consumption of the whole society, respectively; $E_1$ denotes total natural gas supply; $E_2$ denotes total LPG supply; $E_3$ denotes the total electricity consumption; $k$, $v$, and $\phi$ are CO₂ conversion coefficients; and $\eta$ is the share of coal power generation in total power generation.

3.3. Methods

3.3.1. CRITIC–VIKOR Method

(1)

Determination of index weights

The CRITIC method is an objective weighting method that, in contrast to the general entropy weighting method, uses the multiplication form for variability between elements and the conflict between elements to reflect the information size of the data. The CRITIC method was used in this paper as the basis for determining the objective weights of the indicators [65] as follows.

First, the raw indicator data were normalized.

$$x_{ij}=\frac{b_{ij}-\min (b_{ij})}{\max (b_{ij})-\min (b_{ij})}\left(If b_{ij} is a positive indicator\right)$$

(2)

$$x_{ij}=\frac{\max (b_{ij})-b_{ij}}{\max (b_{ij})-\min (b_{ij})}\left(If b_{ij} is a negative indicator\right)$$

(3)

where $x_{ij}$ denotes the data value corresponding to the $j$-th indicator in the year $i$ after standardization, and $b_{ij}$ denotes the data value corresponding to the $j$-th indicator in the year $i$ of the original data.

Then, the standard deviation of each element and the correlation coefficient was calculated, thus reflecting the variability and conflict of each element in the comprehensive measure of low carbon levels in cities [66]. Finally, the weights of the elements were determined according to the calculation results (where n represents the number of research objects and m represents the number of indicators), and the results of the weight assignment are shown in

Table 1. Indicator system for high-quality development of low-carbon cities in the Yangtze River Economic Belt.

First-Grade Indicator	Second-Grade Indicator	Unit	Weight	Type
Industrial Structure	Primary industry share of GDP	%	0.031	−
	Secondary industry share of GDP	%	0.075	−
	Tertiary sector share of GDP	%	0.072	+
Ecological Environment	Industrial wastewater discharge	million tons	0.036	−
	Industrial sulfur dioxide emissions	ton	0.053	−
	Smoke and dust emissions	ton	0.015	−
Carbon Emissions	CO2 emissions	million tons	0.053	−
	Per capita CO2 emissions	ton	0.050	−
	CO2 emission intensity per unit of GDP	Tons/million yuan	0.032	−
Technology Innovation	Number of Invention Patents	Pieces	0.038	+
	Number of utility model patents	Pieces	0.034	+
	Number of design patents	Pieces	0.027	+
Traffic and Population	Per capita road area	m2/person	0.054	+
	Total number of passengers transported by city public buses and trams	10,000 people	0.057	+
	Population density	People/square kilometer	0.083	-
Economic Development	Per capita GDP	Yuan/person	0.068	+
	Government Expenditures	million yuan	0.026	+
	Government Revenues	million yuan	0.028	+
	VAT payable by industrial enterprises above the scale	million yuan	0.040	+
Energy consumption	Total artificial gas and natural gas supply	million m3	0.028	−
	Total LPG supply	ton	0.046	−
	Total annual electricity consumption	million kWh	0.056	−

.

$$S_j=\sqrt{\frac{{\displaystyle \sum _{i=1}^n{(x_{ij}-{\overline{x}}_j)}^2}}{n-1}}$$

(4)

$$R_j={\displaystyle \sum _{i=1}^n(1-r_{ij})}$$

(5)

$$w_j=\frac{S_j{R}_j}{{\displaystyle \sum _{j=1}^m{S}_j{R}_j}}$$

(6)

where $S_j$ denotes the standard deviation of the $j$-th indicator, reflecting the variability of each element; ${\overline{x}}_j$ denotes the mean of the $j$-th indicator; $r_{ij}$ denotes the correlation coefficient between the $i$-th indicator and the $j$-th indicator; $R_j$ denotes the conflict of the $j$-th indicator; and $w_j$ denotes the weight of the $j$-th indicator.

(2)

Calculation of Low-Carbon City Composite Development Index

Firstly, after standardizing the indicators with Equations (2) and (3), the positive and negative ideal solutions were determined to calculate the maximum value of group benefit $T$ and the minimum value of individual regret $Q$. Among them, $p_j^{+}$ represents the maximum value of the $j$th indicator, $p_j^{-}$ represents the minimum value of the $j$th indicator, and $w_j$ represents the weight value of the j-th indicator after the combination of weighting [67].

$$T_i={\displaystyle \sum _{j=1}^m{w}_j(\frac{p_j^{+}-x_{ij}}{p_j^{+}-p_j^{-}}})$$

(7)

$$Q_i=\max \{w_j(\frac{p_j^{+}-x_{ij}}{p_j^{+}-p_j^{-}})\}$$

(8)

Then, the low-carbon city composite development index lcc_i of the $i$th city was calculated, and the default $\lambda$ value was set to 0.5 here. Meanwhile, considering that $lc{c}_i$ is a negative indicator, the original index was forwarded for subsequent analysis, and the low-carbon city composite development index $LC{C}_i$ of the city after forwarding was obtained.

$$lc{c}_i=\lambda \frac{T_i-\min T_i}{\max T_i-\min T_i}+(1-\lambda )\frac{Q_i-\min Q_i}{\max Q_i-\min Q_i}$$

(9)

$$LC{C}_i=1-lc{c}_i$$

(10)

3.3.2. Ensemble Learning

By ensemble learning, we refer to the process of obtaining an integrated learner by building and using multiple learners [68]. Most current scholars believe that the generalization ability of an integrated learner is superior to that of a single base learner [69]. Integration learning algorithms are now broadly classified into two categories in the mainstream: parallelization methods and serialization methods [70]. In this paper, we used the data of 22 indicators of 107 cities each year as a dataset, with the urban low-carbon development index as the dependent variable and each comprehensive evaluation indicator as the independent variable. Additionally, in the process of establishing the index system, we ensured equal numbers when assigning the secondary indicators to each primary indicator in order to eliminate the errors of the results caused by an unequal number of indicators. During the selection of the model type, the classification model was replaced by the regression model, four types of ensemble learning algorithms were selected, and the model with the smallest MSE value was selected to calculate the index importance level by calculating the MSE value corresponding to each algorithm. In addition, when comparing the importance of each level of indicators, the mean value of the importance of the second-level indicators included in each level of indicators was calculated as the importance level of the first-level indicators, which further eliminated the influence of the imbalance of the number of indicators on the true fairness of the results.

3.3.3. Moran’s I Index

The global Moran’s I index [71] was calculated to test and analyze the spatial relevance of the core drivers of low-carbon city development in the Yangtze River Economic Belt. The spatial distribution characteristics among cities along the Yangtze River Economic Belt were clustered or dispersed (i.e., not correlated) as follows.

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

(11)

where $I$ denotes the global Moran’ s I index; $I\in \left[-1, 1\right]$; $x_i$ and $x_j$ denote the observed values of the core drivers of the $i$th city and the $j$th city, respectively, along the Yangtze River economic belt; $\overline{x}$ denotes the mean value of the core drivers of each city; and w denotes the spatial weight coefficient matrix. We used the square of the inverse of the geographical distance to construct the corresponding coefficient matrix (see Equation (12)), and then we standardized the corresponding coefficients.

$$w_{ij}=\frac{1}{d_{ij}{}^2}$$

(12)

The global Moran’s I index was mainly used to study the spatial correlation degree of the Yangtze River Economic Belt as a whole, avoiding the heterogeneity problem of the study area. Accordingly, we used the local Moran’s I index to study the local spatial autocorrelation of each region of the Yangtze River Economic Belt and drew a corresponding local Moran’s I index LISA map.

3.3.4. Advanced Industrial Structure Index

To better quantify the quality of the industrial structure of each city in the Yangtze River Economic Belt, we referred to the spatial vector entrainment method proposed by Fu (2010) [72] to calculate the advanced industrial structure index (AIS) (see Equations (13) and (14)), where the value added of each industry as a share of the current year’s regional GDP $x_{i,0}$ is regarded as a component of the spatial vector $X_0=\left(x_{1,0},x_{2,0},x_{3,0}\right)$. We also used Equation (14) calculate the angle ${\theta }_j$ between the spatial vector $X_0$ and the vectors $X_1=\left(1,0,0\right)$ and $X_2=\left(0,1,0\right)$, $X_3=\left(0,0,1\right)$ of industries arranged from the low level to the high level, respectively, thus generating the advanced industrial structure index.

$${\theta }_j=\arccos \left(\frac{{\displaystyle \sum _{i=1}^3(x_{i,j}·x_{i,0})}}{({\displaystyle \sum _{i=1}^3{(x_{i,j}^2)}^{1/2}·{\displaystyle \sum _{i=1}^3{(x_{i,0}^2)}^{1/2}}}}\right),j=1,2,3$$

(13)

$$AIS={\displaystyle \sum _{k=1}^3{\displaystyle \sum _{j=1}^k{\theta }_j}}$$

(14)

4. Results

4.1. Analysis of the Results of Measuring the Level of High-Quality Development of Low-Carbon Cities

According to the results of the cross-sectional comparison of the low-carbon city quality development index of large city clusters in the Yangtze River Economic Belt (Figure 3), the best low-carbon development level is in the Yangtze River Delta city cluster. The Yangtze River Delta region is located in the eastern coastal region of China, where Shanghai, as the central economic city of China, holds the lifeline of China’s urban development and has the Shanghai–Nanjing–Hangzhou integrated industrial base, which is the largest integrated industrial base in China and relies on high-quality shipping resources; accordingly, the region’s annual GDP accounts for about one-fifth of the country’s annual GDP [73]. The next best low-carbon development level was found to be in the Chengdu–Chongqing region city cluster. In 2012 and before, the high-quality development level of low-carbon cities in the Chengdu–Chongqing region was significantly lower than the overall level of the Yangtze River Economic Belt, which was caused by the unreasonable long-term industrial structure in the upstream area of the Yangtze River Economic Belt. The upstream area comprises several heavy industrial-type enterprises with high energy consumptions and low output values, and it is also the source of raw material supply for production enterprises in the downstream area of the Yangtze River, which was found to have overburdened resource demand and caused the city to produce a large amount of CO₂ emissions. However, with the inward shift of the national economic development strategy, accelerating the construction of the twin-city economic circle in the Chengdu–Chongqing region has led to qualitative leaps in the region’s strategic orientation and scientific/technological innovation, the establishment of an integrated, synergistic talent-development mechanism that has attracted high-tech talent to the western region, and the establishment of the Western Science City and the Chengdu–Chongqing Science and Technology Innovation Corridor; accordingly, the Chengdu–Chongqing cluster leads the entire western region in science and technology innovation development. The development level of the central region of the Yangtze River Economic Belt was maintained near the average level from 2006 to 2012, but with the rise of Chengdu–Chongqing twin cities in the upstream area, the development of cities in this region has gradually but obviously lagged behind the development level of the overall Yangtze River Economic Belt. At present, Hubei Province, Hunan Province, and Jiangxi Province are still dominated by heavy industry, and the ratios of the added value of the secondary industry to regional GDP in Hubei Province, Hunan Province, and Jiangxi Province as of 2020 were 39.19%, 38.15%, and 43.15%, respectively, all of which are high levels. Furthermore, this industrial structure has seriously restricted the high-quality development of urban clusters in the middle reaches of Yangtze River and therefore requires considering for future development.

The low-carbon city quality development indexes of 11 provinces in the Yangtze River Economic Belt for 2008, 2012, 2016, and 2019 were selected at the provincial level to conduct an in-depth study and analysis of the low-carbon development level at the provincial level in the Yangtze River Economic Belt (Figure 4). The results showed that Chongqing City had the most significant improvement of the 11 studied provinces in the Yangtze River Economic Belt over the period of 2006–2016, with an average annual growth rate of about 9.62%. This result also reflects that Chongqing improved its low-carbon urban quality development status to a significant extent in the past nearly fifteen years. For Shanghai, although the pace of low-carbon city development was slower than that of Chongqing during 2008–2012 due to the financial crisis, it was also significantly higher than that of other provinces. Additionally, after 2012, the national 12th and 13th Five-Year Plans enabled Shanghai to accelerate its development of international finance and trade such that is has become an international financial center.

The level of the high-quality development of low-carbon cities in other provinces was found to be more balanced, but it is noteworthy that the level of high-quality development of low-carbon cities in Jiangxi Province was lower than that of other provinces, and its improvement was not significant. In addition to industrial structure, energy consumption was found to be an important factor affecting the high-quality development of low-carbon cities in Jiangxi Province. By the end of 2019, raw coal production in Jiangxi Province was 504 tons, accounting for only about 1.34% of the country, but thermal power generation in Jiangxi Province accounted for 80.01% of the total annual developed amount, indicating that Jiangxi Province currently has a high degree of dependence on coal but is limited by a lack of natural resources in its region and has a strong reliance on external markets, thus increasing the province’s total CO₂ emissions, constraining the province’s local emissions, and constraining local economic development.

We divided cities into four levels according to their comprehensive evaluation scores of low-carbon cities’ high-quality development levels, where Level I indicates a low level with a range of [0, 0.25], Level II indicates a medium level with a range of [0.25, 0.50], Level III indicates a medium-high level with a range of [0.50, 0.50, 0.75], and Level IV indicates a high level with a range of [0.75, 1.00]. By looking at the specific low-carbon development of each city in the Yangtze River Economic Belt in 2008, 2012, 2016, and 2019, we found the following (see Figure 5). First, by 2019, the core cities in each region of the Yangtze River Economic Belt were at a high level of low-carbon development, the most prominent of which were Chengdu and Chongqing in the upstream region; Changsha and Wuhan in the midstream region; and Shanghai, Hangzhou, and Nanjing in the downstream region. Second, due to the influence of China’s economic development plan, most of the cities in the Yangtze River Economic Belt had a low level of low-carbon development in the first decade of the 21st century due to their single mode of economic growth, sloppy development, lack of attention to the urban environment, and negative impact of overcapacity. Thirdly, the spatial distribution trend was high for upstream and downstream regions and low for the midstream region. A common trend for each region was that the core cities had high spatial distributions and the peripheral cities had low spatial distributions, and the closer the core cities were, the lower their low-carbon development levels were, reflecting a certain extent the “siphon effect” of large cities [74]. In addition, at the city level, the low-carbon development level of all cities was found to have increased overall. The Mann–Kendall trend test revealed that among the 107 cities in the Yangtze River Economic Belt studied in this paper, 88 cities had a significant upward trend in their low-carbon development levels, 18 cities were still in a dynamic state of change with no significant trend, and the remaining city (Yichun City) showed a significant downward trend. By observing the spatial distribution of their trend changes (Figure 6), it was found that the cities with no trend and decreasing trends were mainly located in the middle and lower reaches of the Yangtze River Economic Belt, and these regions shared a common characteristic, namely, the relative lack of production factors, an important reason for the lagging development of this type of city.

4.2. Analysis of Ensemble Learning Identification Results for Core Drivers

It is generally believed that the higher the importance of an indicator, the stronger the degree of influence of that indicator on the level of low-carbon development of a city. Therefore, we propose that indicators can be used to explore the core drivers that influence the level of low-carbon development in cities. Although ensemble learning has been used to calculate the importance of indicators in existing studies, only one algorithm is often used without an evaluation of its strengths and weaknesses, so such an algorithm cannot be proven to be suitable for experimental data samples. Additionally, there are limitations in algorithm selection. Therefore, based on previous research and a comparison of the mean square error (MSE) values of the four abovementioned algorithms (Figure 7), the Extra Trees algorithm (which had the smallest MSE value) was finally selected in this paper for the subsequent calculation of the index importance level, and the reasonableness of the model selection was verified to some extent.

In the process of selecting the core drivers for the quality development of low-carbon cities, the number of key influencing factors should not be excessive. In this paper, the “majority referendum principle” was used as a reference, i.e., all indicators with a cumulative importance level of more than two-thirds were considered key influencing factors. As can be seen in Figure 8, after the first five indicators were selected, the cumulative importance level of key impact factors was already close to 66.7%, so it could be considered that less than half of the indicators could cover nearly 70% of the overall importance level, indicating that these key impact factors had good explanatory power for the level of high-quality development of low-carbon cities. Of these five key influencing factors, industrial structure indicators accounted for 40% of the total, demonstrating that industrial structure is currently the core driving force to empower the high-quality development of low-carbon cities.

By calculating the mean value and overall proportion of the importance of each subsystem, we tried to explore the trend changes of the importance level of each subsystem in relation to the level of quality development of low-carbon cities over time. According to Figure 9, the importance level of each subsystem had significant time-trend change characteristics during 2006–2019. First, since 2015, the government has gradually strengthened the importance and support for the science and technology innovation capacity of cities in the Yangtze River Economic Belt; by 2019, the technology market turnover in the Yangtze River Economic Belt had reached 797.8 billion yuan, and the number of patents per 10,000 people was 14. At present, the upper reaches of the Yangtze River mainly develop electronic information hardware equipment such as power batteries, the middle reaches of the Yangtze River focus on investment in the optoelectronic display industry, and the lower reaches of the Yangtze River are dedicated to the development of intelligent manufacturing and high-end biological and chemical material technology industries. In short, the current upstream and midstream areas of the Yangtze River Economic Belt are mainly focused on providing production components while the downstream areas are responsible for assembling and industrializing them, thus forming a complete high-tech industrial chain and promoting innovation-driven development.

Second, the results shown in Figure 9 demonstrate that the industrial structure has been the most critical core driver while urban transport and population are becoming less influential in low-carbon urban development. We observed an interactive relationship between industrial structure and economic development. Regional economic development determines the future trends of industrial structure, while industrial structure can counteract economic development. The reasonable allocation of the ratio of industrial structure at all levels and the synergistic allocation of resource factors can promote the quality of regional economic development, but the pursuit of economic growth rate may lead to the deterioration of the quality of industrial structure, in turn inhibiting economic development. The current development of China’s industrial structure is gradually transitioning from the primary industry to the secondary and tertiary industries, with the eventual aim of realizing a new pattern of tertiary-industry-led development. As of 2021, among the 11 provinces in the Yangtze River Economic Belt, the proportion of tertiary industry in the remaining 10 provinces exceeded 50%, except for Jiangxi Province (which reached 47.6%), and Shanghai had the highest proportion of the tertiary industry, reaching 73.27% (an impressive figure). This shows that the quality of industrial structure determines the future trend of regional economic development, so industrial structure should be a key concern for cities along the Yangtze River Economic Belt in the future.

In addition, science and technology innovation can greatly promote changes in industrial structure. On the one hand, science and technology innovation can optimize the allocation structure of natural resources, maximize output with fewer factors of production, improve factor production efficiency, and reduce unnecessary transaction costs in the production process. On the other hand, the improvement of science and technology innovation level is bound to lead to the emergence of new technologies and products, which may give rise to new industries. While increasing the diversity of the market, it can also gradually eliminate some of the industries with high energy consumption and low production capacity under the role of the market mechanism and improve the quality of the industry. Therefore, we argue that science and technology innovation is the core internal driving force to promote the adjustment of the industrial structure of China’s Yangtze River Economic Belt, and industrial structure is also an important pillar to promote the high-quality development of low-carbon cities.

4.3. Analysis of Spatial Autocorrelation Results

According to the conclusion in Section 4.2, it is known that industrial structure is the core driving factor affecting the high-quality development of low-carbon cities. To further investigate whether there was a certain correlation effect between the two factors of industrial structure and science and technology innovation in space, we calculated the global Moran’s I index and the local Moran’s I index, tested and analyzed the correlation that existed in space, and drew Moran’s scatter plot. Among studied factors, the industrial structure was expressed as the advanced industrial structure index, the level of scientific and technological innovation was expressed as the number of invention patents obtained by the city in that year, and the level of carbon emission of the city was expressed as per capita carbon emissions.

After calculating the global Moran’ s I index for the advanced industrial structure of 107 cities in the Yangtze River Economic Belt from 2006 to 2019, we found that there was a significant spatial aggregation of the degree of advanced industrial structure and carbon emissions. Additionally, for science and technology innovation, there was no obvious spatial autocorrelation in the period of 2006–2008, which showed a random distribution in space. However, from 2009 onward, there was a spatial aggregation phenomenon, and the significance increased. This indicates that the positive correlation between geospatial agglomeration and technological innovation is becoming more and more significant (

Table 2. Moran’s I index of industrial structure, science and technology innovation, and carbon emissions across the Yangtze River Economic Belt, 2006–2019.

Year	Moran’s I of Industrial Structure	Z Value	p Value	Moran’s I of Technology Innovation	Z Value	p Value	Moran’s I of Carbon Emissions	Z Value	p Value
2006	0.190	5.950	0.000	−0.018	−0.028	0.777	0.042	2.000	0.046
2007	0.239	7.417	0.000	0.011	0.672	0.501	0.153	4.942	0.501
2008	0.242	7.508	0.000	0.028	1.204	0.229	0.161	5.206	0.000
2009	0.197	6.159	0.000	0.052	1.977	0.048	0.155	5.005	0.000
2010	0.212	6.602	0.000	0.078	2.805	0.005	0.152	4.898	0.000
2011	0.243	7.534	0.000	0.090	3.146	0.002	0.116	3.827	0.000
2012	0.260	8.054	0.000	0.115	3.902	0.000	0.151	4.890	0.000
2013	0.258	7.997	0.000	0.130	4.379	0.000	0.161	5.127	0.000
2014	0.278	8.584	0.000	0.136	4.551	0.000	0.193	6.130	0.000
2015	0.287	8.840	0.000	0.185	6.054	0.000	0.206	6.496	0.000
2016	0.297	9.143	0.000	0.203	6.658	0.000	0.234	7.335	0.000
2017	0.263	8.145	0.000	0.153	5.057	0.000	0.224	7.111	0.000
2018	0.264	8.158	0.000	0.156	5.110	0.000	0.218	6.915	0.000
2019	0.204	6.405	0.000	0.104	3.578	0.000	0.209	6.612	0.000

).

According to the results of the global Moran’s I index calculation of the degree of advanced industrial structure and the level of science and technology innovation in the Yangtze River Economic Belt, these two factors had significant spatial autocorrelation in general. To further explore the spatial autocorrelations among the regional cities and the corresponding significance levels in-depth, we calculated the local Moran’s I index and drew a LISA diagram with the data for 2019 (Figure 10). In the diagram, HH-type (high–high type) regions refer to regions with an advanced level of industrial structure or a relatively high level of science and technology innovation, with neighboring regions also possessing high levels; LH-type (low–high type) regions refer to regions with a low advanced level of industrial structure or a high level of science and technology innovation, with neighboring regions showing high levels; LL-type (low–low type) regions refer to regions with low advanced levels of industrial structure and science and technology innovation, with similar levels for neighboring regions; and HL-type (high–low type) regions refer to regions with high advanced levels of industrial structure or science and technology innovation, with neighboring regions showing low levels.

For the advanced level of industrial structure, there were 15 cities of the HH type, mainly concentrated in the Yangtze River Delta region. The cities in this region showed a high-quality spatial gathering, good radiation, and driving effects on neighboring cities. Reflecting that the industrial structure in the lower reaches of the Yangtze River Economic Belt is developing in a benign direction, there were three HL-type cities, namely, Chengdu, Wuhan, and Qujing. Chengdu and Wuhan were the core development cities in the upper and middle reaches of the Yangtze River, respectively. In particular, Chengdu, with the promotion of the development strategy of Western China, has absorbed a large number of advantageous talent resources in the West, enabling the city’s economy to fully develop and its industrial structure to be increasingly rationalized. However, its radiation capacity for neighboring cities is poor, and, coupled with the lack of talent and resources in neighboring cities, the phenomenon of polarization is easily formed. There were 12 L-type cities, including Baoshan City, and there was only one LH city, Maanshan, which had a low level of industrial structure and a low absorption capacity for the advantageous industries of the neighboring cities.

For the level of science and technology innovation, there were eight cities of the HH type, all of which are located in the Yangtze River Economic Belt downstream of the Yangtze River Delta region in terms of spatial distribution. The cities in this region showed a high-quality spatial aggregation phenomenon, reflecting the close connection between these cities and their neighboring cities in science and technology innovation, as well as the trend of synergistic development in science and technology innovation, strengthening regional cooperation and promoting the high-quality development of science and technology innovation level in the Yangtze River Delta region through the mutual recognition of talents and the contribution of resources. There was only one HL-type city in Chongqing. There were nine LL-type cities, most of them located in the upstream area of the Yangtze River Economic Belt, indicating that the science and technology innovation level in the upstream area still needs to be strengthened. The four LH-type cities were Zhoushan City, Huzhou City, Xuan City, and Ma’anshan City, whose science and technology innovation levels are still in need of strengthening. Xuancheng and Maanshan, which are geographically located in the lower reaches of the Yangtze River, should strengthen cooperation with neighboring cities with strong scientific and technological innovation capabilities, enhance the exchange of scientific and technological talents, and improve their currently relatively backward level of scientific and technological innovation in the future.

In addition, for the level of urban carbon emissions, there were 11 HH-type cities, mainly in the lower reaches of the Yangtze River, indicating that industrial production is strongly connected in the region, the level of urban carbon emissions is high, and the level of low-carbon urban governance needs to be improved. The four LH-type cities were Shanghai, Hangzhou, Nanjing, and Zhoushan, of which Shanghai, Hangzhou, and Nanjing were the core cities in the lower reaches. As reflected by the results, although these cities have achieved relatively good results in low-carbon city development with low carbon-emission levels, they lack the positive influence of the core cities. On the contrary, in the Chengdu–Chongqing urban agglomeration in the upper reaches of the Yangtze River, most of the cities in the region showed a “low–low” spatial aggregation of carbon emissions by 2019, reflecting that the region has made significant breakthroughs in the process of low-carbon city development and greenhouse gas emission management and that the region has formed radiation and driving effects in space. The spatial effect of low-carbon cities in the region has been radiated, thus driving the whole region to achieve the high-quality development of low-carbon cities.

5. Discussion

In this paper, we found several interesting phenomena by analyzing the spatial and temporal evolution characteristics of the high-quality development level of low-carbon cities in the upper reaches of the Yangtze River Economic Belt.

First, in terms of temporal trends: the level of low-carbon city high-quality development was on an upward trend from 2006 to 2019. Development in 2008 was influenced by the global subprime crisis, which led to the full development of urban infrastructure construction and the increase in industrialization level, which eventually brought negative results such as overcapacity. In particular, Panzhihua city, as a resource-based city in the western region [75], was at a low level of low-carbon urban quality development during 2006–2016. This is because Panzhihua, as the largest steel smelting base in Western China, has abundant natural mineral resources and carries the whole West’s demand for steel production factors [76], which makes the industrial structure of the city more monotonous and the proportion of secondary industry higher compared to other cities [77], resulting in more serious local heavy metal pollution [78]. Specifically, the average proportion of secondary industry to the regional GDP in Panzhihua from 2006 to 2019 was 70.67%, and the average proportion of employees in the secondary industry to the total number of employees in the city was 62.79%; through these percentages, we can see that the industrial structure of the low-carbon city is the key influencing factor that restricts its level of quality development. Given the special characteristics of the city, in addition to the adjustment and optimization of the industrial structure, it is also necessary to strengthen the reconstruction of the ecological environment of the mining area. Local mining enterprises should build their own clean and environmentally friendly mineral-resource transportation chains and improve their dust-removal and environmental-protection technologies through technological innovation; in addition, the management of the acoustic environment and sewage discharge should also be taken into account in the reconstruction of the ecological environment of the mining area, the recycling rate of the treated tailing water should be increased, the recycling rate of wastewater after treatment should be increased, and wastewater that meets certain standards should be recycled [79].

Second, in terms of spatial characteristics, the mega-city clusters along the Yangtze River Economic Belt had an obvious “siphon effect” on large cities in terms of industrial structure and technological innovation level. The most significant phenomenon was seen in the twin cities of Chengdu and Chongqing. These two cities can absorb high-quality production resources and talents from the neighboring cities through their superior conditions of policy dividends and economic development, creating stronger dynamics for the development of their cities. This result is also in line with the national policy of establishing a twin-city economic circle in the Chengdu–Chongqing region with the Chengdu–Chongqing twin cities as the core. In addition, cities that are too close to core regions are also subject to the “siphon effect”, which creates a shadow of agglomeration [80]. In particular, Chongqing, according to the previous study, experienced the rapid development of low-carbon cities with high quality from 2006 to 2019, but the cost and low-carbon development of Deyang City, Neijiang City, Yibin City, and Zigong City were significantly constrained due to resource drain. In this paper, we argue that the “siphoning effect” of such large cities is not a bad thing, as the results of the empirical study showed that a moderate diversion of talents and resources to core cities can lead to the rapid development of strategically important cities and thus the rise of the whole region. This is an effective way to quickly enhance the core competitiveness of large city clusters, and it is also a necessary path for the development of mega-city clusters. Local governments and all sectors of society should correctly view this phenomenon and respect the objective facts and development laws of population migration, population growth, and resource allocation transfer. However, in the development process of such cities, it is necessary to strengthen the effect of radiation on the neighboring cities, enhance the development of “co-location”, accelerate the construction of regional transportation network, strengthen the accessibility of transportation between neighboring cities [81], and strengthen the construction of “regional economic integration” [82] to avoid the phenomenon of over-polarization and coordinate the high-quality development of the regional economy.

6. Conclusions and Suggestions

Taking 107 cities in the Yangtze River Economic Belt from 2006 to 2019 as examples, the authors of this paper constructed a generic set of comprehensive urban low-carbon development level measurement index systems from seven aspects: carbon emissions, industrial structure, economic development, ecological environment, science and technology innovation, transportation and population, and energy consumption. The authors also explored the temporal and spatial evolution trends of the core drivers of the low-carbon city development level in the Yangtze River Economic Belt. According to the empirical results, the low-carbon development level of more than 80% of the cities showed a significant upward trend and generally showed a spatial distribution characteristic of “central collapse”. At the provincial level, the gap between provinces was found to be narrowing and tended to develop synergistically, but there was still a gap with Shanghai. Through the exploration of the core drivers, it was found that the industrial structure was the core driver of low-carbon urban development over time, transportation and population play key roles in the development of low-carbon cities, the economic development of low-carbon city construction has maintained a steady output, and the importance of science and technology innovation has been significantly increased since 2016, indicating that the driving force of science and technology innovation will continue to increase in the process of low-carbon city construction in the future, thus helping cities to achieve the optimization of industrial structure and socio-economic development. In addition, in terms of spatial distribution trends, the authors of this paper found that the core cities of the Yangtze River Economic Belt generally had high degrees of advanced industrial structure and innovation, but the radiation drive to the surrounding cities was not strong, which led to the inhibition of the industrial structure and the science and technology innovation ability of the surrounding areas, as well as increases in the carbon-emission levels of the surrounding cities to a certain extent. The formation of an obvious big city with adjacent neighboring cities leads to a “siphon effect”.

Given the current development situation of the Yangtze River Economic Belt, the authors of this study provide reasonable suggestions from the following three perspectives. First, local governments and relevant authorities should accelerate the transformation and upgrading of energy-intensive cities in the middle reaches of the Yangtze River Economic Belt and in the West, reduce cities’ dependence on energy and heavy industry, open up new channels for economic development, combine local development levels and characteristics, form green industries with local characteristics, and pay attention to the environmental protection of local mining and other natural resource-extraction plants; second, we should increase the government’s support for urban science and technology innovation in terms of policy and finance so that more capital can be diverted to the green and high-quality development of science and technology innovation enterprises. At the same time, the effect of only relying on enterprises’ closed innovation to promote the city’s overall scientific and technological innovation capacity is also limited. Third, the construction of transportation networks in cities along the Yangtze River Economic Belt should be strengthened to improve the accessibility of transportation between neighboring cities, which is especially crucial for the neighboring cities that are close to the core cities in order to break the “siphon effect”. Additionally, the experience of the Yangtze River Economic Belt in low-carbon development can be used as a reference for other cities in China, as well as a positive initiative for cities along the “Belt and Road” in China and a reference for the UN SDGs to establish a scientific comprehensive index system for low-carbon cities.

The limitations of this paper include that fiscal taxation was not included as a core driver of high-quality development of low-carbon cities and that the influence of human factors such as tourism development on the development of low-carbon cities, which are also very important, was also not considered. Therefore, in the process of improving the comprehensive evaluation index system of low-carbon cities in the future, the abovementioned dimensions could be prioritized. In addition, the spatial influence of each core driver on the level of low-carbon city development is an interesting element that can also be studied more in the future.