Spatial–Temporal Decoupling of Urban Carbon Emissions and Socioeconomic Development in the Yangtze River Economic Belt

Abstract

The spatial–temporal pattern, influencing factors and driving variables of carbon emissions are essential considerations for achieving China’s carbon peak and neutrality targets, which support high-quality development. This study was designed to explore and evaluate the spatial–temporal evolutionary characteristics, trends and main influencing factors of carbon emissions in the Yangtze River Economic Belt (YREB), focusing on the decoupling of carbon emissions and socioeconomic development in the YREB. In total, 11 provinces and key cities were focused on as the research objects of the YREB district Tapio decoupling model, which examined the decoupling relationship between carbon emissions and socioeconomic development. Combined with a geographic detector, the Tapio, Logarithmic Mean Divisia Index (LMDI) and gray prediction models were employed in a comprehensive evaluating pipeline, which was constructed to decouple the main influencing factors and corresponding impacts of carbon emissions. Particularly, the gray prediction model was employed to predict the carbon emission differences in the YREB sub-regions in 2030. The results indicated the following: (1) The total carbon emissions showed a periodic fluctuation and upward trend with obvious spatial differences, and energy consumption was mainly dominated by coal. (2) The center of carbon emissions was located in Hubei Province in the middle reaches of the Yangtze River, with a standard deviation ellipse showing a “Southwest–Northeast” trend, and most provinces were concentrated in the L-H (low-high) cluster. (3) The entire YREB had achieved carbon emissions decoupling, but it was mainly in a weak decoupling state. (4) Carbon emissions were significantly affected by the indicator E for economic growth, with the indicators EI for energy consumption and I for the added ratio of GDP also bringing greater impacts on carbon reduction contributions. The carbon emission prediction results indicated that the upper and middle reaches of the YREB were more likely to achieve carbon neutrality.

1. Introduction

The continuous increase in CO₂ concentration is primarily responsible for global warming, which presents significant risks to human health and the natural ecology of the earth and has attracted widespread attention. To this end, the international community has proposed achieving net zero CO₂ emissions by the mid-21st century and keeping the global temperature increase within 2 °C [1]. Under this constraint, countries have set emission reduction targets based on their emission reduction responsibilities and elevated “carbon neutrality” to a national strategy. Since the initiation of its reform and opening-up policy, China has experienced a steady rise in carbon emissions, contributing to approximately 27% of global CO₂ emissions and positioning itself as the top carbon emitter worldwide [2]. In addition, the 28th United Nations Climate Change Conference COP28 agreed that countries will carry out climate action under the Paris Agreement in a “Self-determined Contribution +” model, and China will continue to face pressure to update its NDC in climate multilateral and bilateral processes [3]. The task of emission reduction faces enormous difficulties [4], as well as significant challenges [5]. In order to gradually achieve net zero carbon dioxide emissions, the national government has established specific targets, including achieving peak carbon emissions by 2030 and realizing carbon neutrality by 2060 [6]. It aims to incorporate addressing climate change as a national strategy into the overall framework of ecological civilization development and socioeconomic progress [7].

As a significant hub for socioeconomic development and ecological conservation in China, the urban agglomeration within the Yangtze River Economic Belt (YREB) plays a crucial role in reducing carbon emissions. This region spans the eastern, central, and western parts of China, forming an economically gradient development pattern. The lower reaches areas are economically developed but face saturated environmental carrying capacity, the middle reaches are experiencing accelerated industrialization alongside increasing ecological pressures, and the upper reaches are ecologically sensitive yet rich in clean energy resources. This unique regional heterogeneity endows the YREB with significant strategic importance in China’s overall carbon reduction framework.

As a key pilot area for the country’s ecological civilization development, the YREB region assumes the crucial responsibility of leading the way in attaining “Carbon Peak” and “Carbon Neutrality.” However, the YREB currently confronts substantial environmental challenges. The conventional development paradigm led to progressively intensifying environmental burdens, while the tension between industrial restructuring and ecological conservation has become markedly pronounced. The central government has clearly pointed out that the strategic positioning of the development of the YREB must adhere to ecological priority and green development, jointly focus on major protection, and not engage in major development. Thus, the primary objective of this research is to explore how to develop practical low-carbon development strategies for the YREB under varying levels of carbon emissions and socioeconomic growth, facilitate the region’s transition to a low-carbon economy, and harmonize the interplay between carbon emission control and socioeconomic progress.

The previous literature on the connection between carbon emissions and socioeconomic development mostly reported and discussed it from the perspectives of different countries, regions, and industries. For example, a multi-sectoral analysis of carbon emissions was conducted to reveal the driving factors of consumption-based carbon emissions from 2002 to 2017 in Guangdong province, China [8]. To identify effective strategies for reducing carbon emissions at the national scale, the interplay between carbon emissions and socioeconomic development was frequently analyzed and measured in China based on the global value chain embedding estimation [9]. Numerous studies have aimed to explore the connection between economic income and carbon emissions within the framework of the Environmental Kuznets Curve (EKC), which suggests that an inverted U-shaped relationship exists between environmental deterioration and economic growth [10,11]. To achieve this objective, the association between carbon emissions (or carbon emission intensity) and social economies (or people’s income) was often reported to be related to GDP per capita, and this correlation was frequently examined in the inverted U-shape implications for the EKC figuring [10,11,12,13,14]. A long-term equilibrium connection between carbon emissions and GDP per capita income was examined across a sample of 86 developing and developed countries spanning the period from 1990 to 2015 in the Americas [13]. The findings showed that the EKC at least held for three sectors, i.e., the electricity and heat production sector, the other energy industry own use sector, and the commercial and public service sector [13]. The connection between carbon emissions and socioeconomic progress has also been examined and quantified in the United States through the lens of the EKC [14]. Meanwhile, the relationship between energy-related carbon emissions and socioeconomic progress has been investigated and estimated in African countries too [15]. Recently, the relationship between carbon emissions and socioeconomic progress was analyzed and explored in Morocco based on co-integration tests [16]. To identify effective strategies for reducing carbon emissions at the regional level, the decoupling state between energy production and consumption emissions was examined in relation to economic factors in Guangdong Province, using an approach based on implicit carbon flow analysis [17]. The decoupling relationship between carbon emissions and economy was also investigated to assess the low-carbon development level in Sichuan Province [18]. Later, the decoupling status between transportation carbon emissions and economy was assessed to support the advancement of low-carbon transportation in Hainan Province [19]. To identify effective strategies for reducing carbon emissions at the industrial level, the connection between carbon emissions in key industrial sectors and socioeconomic progress was analyzed in China [20] in order to achieve green and sustainable development within the industrial sector. The relationship between carbon emissions from the logistics industry and the economy was also analyzed to reduce carbon emissions and enhancing the efficiency of the logistics industry [21]. Based on the existing research status, few scholars have studied the relationship between carbon emissions and socioeconomic development in urban agglomerations, and the conclusions and recommendations drawn are not highly targeted towards specific urban agglomerations.

At present, the academic community has adopted multiple research approaches and diverse perspectives regarding carbon emission calculation methodologies, primarily including Life Cycle Assessment (LCA), Input–Output Analysis (IO), and the IPCC emission factor method. The LCA approach requires tracking the entire construction cycle, with difficult data acquisition and high accounting costs. Therefore, it is more suitable to study the carbon emissions of individual structures in the construction projects [22]. The IO model input–output method can provide a more comprehensive perspective for carbon emission research, including identifying implicit carbon transfers [23]. As a valuable tool for conducting macro-level research on carbon emissions, it can provide insights and references for formulating effective carbon reduction policies and measures [24]. However, this approach demands high-quality data inputs, and since China’s input–output tables are only updated quinquennially, the data lacks sufficient timeliness for analyzing the Yangtze River economic zone. The IPCC (i.e., Intergovernmental Panel on Climate Change) method is an international recognized approach used to estimate and investigate the national greenhouse gas emissions [25]. Due to its adaptable data requirements and broad applicability, this method is frequently employed to estimate carbon emissions across diverse sectors and geographical areas [25,26,27]. Therefore, the emission coefficient method was selected for this study due to its comparative advantages in applicability and data accessibility.

The primary methods for decomposing carbon emission factors are indicator decomposition analysis [1] and the structural decomposition analysis (SDA). For example, the SDA approach was adopted to examine the driving factors of carbon emissions generated by energy consumption in Gansu Province and identified the critical final demand and industries for carbon emissions [28]. Compared to SDA, IDA offers greater operational simplicity and enhanced suitability for longitudinal studies and cross-period comparisons. In fact, IDA includes many types of models, and the LMDI model is the most widely used one. For example, the LMDI model has been applied to investigate the determinants of greenhouse gases in Spain [29]. The LMDI model was also adopted to decompose the carbon emissions and main driving factors of the transportation sector in Hungarian [30]. The carbon emissions of primary and recycled aluminum products and the main contributing forces of the aluminum industry’s carbon footprint were analyzed and estimated with the LMDI model [31]. The study explored and decomposed carbon emissions from China’s transportation and the key factors that led to the largest increase with the LMDI model [31]. It is evident that the decomposition process of the commonly applied LMDI model does not produce residual terms, and it also allows for the integration of both additive and multiplicative decomposition approaches [31]. Of course, the other decomposition methods have also been used by scholars. For example, the generalized Divisia Index approach was applied to break down carbon emissions in China’s construction industry and identify the contributors to carbon emissions [32]. In contrast, the Tapio decoupling framework is widely utilized for analyzing the separation of natural resource consumption or environmental deterioration from economic expansion and has been implemented across various regions and industrial sectors. For instance, the decoupling coefficient between socioeconomic progress and carbon emissions was analyzed in the Guanzhong Plain urban cluster using the Tapio decoupling model [33]. Using the Tapio model, the decoupling connection between land use-related carbon emissions and socioeconomic development was investigated in Hubei Province. The decoupling status of economic growth and carbon emissions was also analyzed in China’s transportation industry with the decoupling models and the EKCs [34]. Then, while decoupling models alone cannot assess driving factors, their integration with exponential decomposition models enables both the identification of emission drivers and the examination of factors influencing decoupling dynamics. Therefore, researchers often combine decoupling analysis with index decomposition methods to examine the determinants influencing carbon emission decoupling patterns.

For the regional research, several regional studies have examined the relationship between carbon emissions and economic growth using integrated analytical approaches. The Tapio decoupling model coupled with LMDI decomposition has been applied to analyze this dynamic in China’s provincial capital cities [35], while similar methodology has been employed to investigate both decoupling states and driving factors between economic development and carbon emissions in Xinjiang [36]. For the industry research, the decoupling effect of carbon emissions was analyzed and estimated in China’s manufacturing sector through a dual-dimensional framework encompassing both regional and industrial perspectives, with the Tapio decoupling model and the LMDI model [37]. Similarly, the carbon emissions and carbon sink functions of agriculture was calculated and measured in the Yellow River Basin with the Tapio decoupling model and the LMDI model [38]. The decoupling status of transportation carbon emissions and socioeconomic development was also explored and measured in Hainan Province with the Tapio decoupling model and the LMDI model, as well as the contribution values and the contribution rates of various influencing factors [19].

In general, space research helps identify regional spatial differences. Recent studies have utilized the MGTWIPR model to estimate annual carbon emissions across China’s 31 provincial-level administrative units during the 2015–2017 period. [39]. The study found that the inter provincial carbon emission transfer in China had an obvious asymmetric and unbalanced characteristics [39]. A comprehensive analysis of energy-related carbon emissions was conducted for 282 Chinese prefecture-level cities (2003–2019) using Theil–Kaya decomposition and an enhanced Atkinson index, with particular focus on examining emission inequality patterns and energy equity metrics. [40]. In order to make the research result analysis more accurate and the emission reduction recommendations more specific and thorough, it has been found that many scholars combine driving factors and decoupling structures with spatial research to analyze carbon emissions through a literature review. Moran’s I index and the slope index were employed to analyze and explore its spatiotemporal changes, combined with the Kaya equation and LMDI method to analyze driving factors [41]. Between 2010 and 2019, China’s construction sector exhibited a consistent rise in carbon emissions across provincial regions, demonstrating distinct spatial characteristics with eastern provinces emitting significantly higher levels than their western counterparts. [41]. Moreover, spatial autocorrelation analysis was also adopted to identify and clarify the spatiotemporal differentiation characteristics of agricultural carbon emissions in the three northeastern provinces [42]. The driving factors and their interactions of agricultural carbon emissions were also estimated and explored through LMDI decomposition models and the geographic detectors [42]. Furthermore, researchers employed Tapio decoupling analysis combined with LMDI decomposition to investigate the key drivers of emissions while simultaneously assessing the spatial distribution characteristics of carbon emissions from agricultural cultivation activities across Henan Province [43]. The analysis revealed a distinct north–south divergence in agricultural carbon emissions across Henan Province, with southern regions consistently demonstrating higher emission intensity compared to northern areas [43].

These studies indicated that the spatiotemporal evolution path of carbon emissions became relatively stable and clear. In addition, many scholars have also combined driving factors and decoupling structures with predictive models to analyze carbon emissions. For example, researchers have applied the IGA-BP hybrid model to assess and forecast the decoupling dynamics between carbon emissions and socioeconomic development in China’s construction sector, while simultaneously identifying key influencing factors [44]. The study further employed LMDI decomposition to analyze emission patterns, coupled with scenario analysis and Monte Carlo simulation to project future carbon emission trajectories for China’s energy-intensive industries across multiple development scenarios [45]. Subsequent research integrated the Tapio decoupling index with panel quantile regression analysis to empirically determine key determinants while additionally employing a gray prediction model to forecast China’s low-carbon eco-city development levels by 2030 [46]. In summary, a comprehensive understanding of emission reduction potential and the formulation of scientifically sound policy recommendations require not only the analysis of driving factors and decoupling relationships but also the projection of future carbon emissions.

The preceding analysis reveals that employing decoupling models to examine the link between carbon emissions and socioeconomic progress has gained widespread acceptance. Meanwhile, the application of LMDI techniques to explore underlying micro-level driving factors has also become a commonly adopted approach within academic circles. However, there are still issues in the existing literature. First, there is lack of notably targeted research on the special economic zones (e.g., YREB), as existing findings predominantly focus on the national or provincial levels, failing to reflect the heterogeneous characteristics within urban agglomerations. Second, most studies examine only single dimensions, either driving factors of carbon emissions or decoupling status, without establishing a comprehensive framework that can integrate decomposition analysis, decoupling assessment, and spatial characteristics. Third, the current predictions of future carbon emission trends primarily rely on scenario assumption methods, with a limited application of historical data-based objective modeling. To address these research gaps, this study develops an integrated “decomposition–decoupling–prediction” analytical framework, including (1) a systematic combination of the LMDI decomposition model with the Tapio decoupling indices to quantitatively assess each driving factor’s contribution to decoupling elasticity, (2) a specific examination of the YREB as a national strategic research object, and (3) the employment of the standard deviation ellipse and gray prediction models to reveal spatiotemporal differentiation patterns of carbon emissions and predict peak pathways for 2030. This approach may provide some novel perspectives and methodological support for formulating regionally differentiated emission reduction policies.

This study potentially offers the following contributions (Figure 1): On the one hand, it analyzes and predicts carbon emissions and their decoupling, enabling the government and enterprises to establish targeted carbon reduction objectives and strategic measures for energy structure industry transformation. On the other hand, this study combines decomposition and decoupling analysis to construct a decoupling-driven model, analyzing the impact of various effects on the decoupling dynamics between carbon emissions and socioeconomic progress. It can also serve as an additional extension to carbon emission analysis, applicable to relevant research in other regions or different industries. Finally, this study made relevant predictions of carbon emissions in 2030, which can provide suggestions for the development of the dual carbon strategy in the YREB. Figure 1 presents the framework diagram of the article’s analytical pipeline.

2. Study Area and Data

2.1. Study Area

The YREB region, as a crucial element in realizing China’s two centennial objectives and the revitalization of the Chinese nation, as well as a pioneering model area for the development of ecological civilization, holds a pivotal position (Figure 2). This economic region spans across the three major regions of China, covering a total area of roughly 2.0523 million square kilometers, which constitutes approximately 21.4% of the country’s total area. This study selects the YREB as the research subject based on three key considerations: First, the study area comprehensively covers three major sections (upper, middle, and lower reaches), including three national-level urban agglomerations, the Yangtze River Delta, Middle Yangtze River, and Chengdu-Chongqing economic zones, exhibiting typical watershed characteristics. Second, according to the “YREB Development Plan Outline” issued by the State Council, these 11 provinces and municipalities are explicitly designated as core implementation zones, bearing the strategic mission of coordinating ecological protection with economic development. Third, the region has established a relatively complete energy consumption and socioeconomic statistical system, providing a solid data foundation for multi-model integrated analyses, including LMDI decomposition and Tapio decoupling. However, in the process of rapid development, the YREB also faces some severe challenges. Among them, the severe situation of the ecological environment and the problem of uneven regional development are particularly prominent. In the future, with the continuous implementation of relevant policies and measures, the YREB is expected to overcome these challenges and become an emerging growth driver for China and even the global economy.

2.2. Data Collection

This study selected panel data from 11 provincial and municipal units in the YREB from 2006 to 2022 as research samples. The various energy consumption data and energy conversion coefficients of each province and city were sourced from the China Energy Statistical Yearbook, and the carbon dioxide emission coefficients were sourced from the Guidelines for Provincial Greenhouse Gas Inventory Compilation. This study utilizes regional GDP, population statistics, and industrial value-added data for all provinces and cities, which were obtained from the China Statistical Yearbook and respective provincial statistical yearbooks, with complete data coverage ensuring no missing values.

3. Research Methods

3.1. The Calculation of Carbon Emission

According to the IPCC’s research, the combustion of fossil fuels is the main source of global carbon emissions. Therefore, in the process of calculating the carbon emissions of the YREB, this study selected 8 main energy sources and further converted the energy consumption units into standard coal units. Due to the different socioeconomic development status of cities in the YREB, there are significant differences in carbon emissions. Therefore, this study adopted the IPCC rules and approaches to calculate and measure carbon emissions [25]. The specific calculation formula is listed as follows:

$$C_t=\sum _{i=1}^n{E}_{it}\cdot {\delta }_i\cdot {\theta }_i$$

(1)

In the formula, C_t represents the total carbon emissions in the t-th year; E_it is the total energy consumption of type i in year t; δ_i means the conversion standard coal coefficient for the i-th type of energy; and θ_i indicates the carbon emission coefficient of the i-th type of energy. The conversion standard coal coefficient and carbon emission coefficient for various types of energy are shown in

Table 1. Standard coal conversion factors and carbon emission factors.

Type	Conversion Factor for Standard Coal/(104 tce/104 t)	Carbon Emission Conversion Factor/(104 t/104 tce)
Coal	0.7143	0.7559
Crude Oil	1.4286	0.5857
Coke	0.9714	0.8550
Gasoline	1.4714	0.5714
Diesel	1.4571	0.5921
Petrol	1.4714	0.5530
Fuel Oil	1.4286	0.6185
Natural gas	1.3300	0.4483

.

3.2. The Tapio Decoupling Model

The Tapio decoupling model is a quantitative analytical method that evaluates the decoupling status between carbon emissions and economic growth by calculating their elasticity coefficient. It includes the dynamic concept of elasticity based on the decoupling theory [47], effectively improving the accuracy of decoupling analysis.

This study introduces the Tapio decoupling model for analysis and research. The specific calculation formula is listed as follows:

$${\mathsf{\epsilon }}_t={\displaystyle \frac{\Delta C/C_s}{\Delta Y/Y_s}}={\displaystyle \frac{\frac{C_t-C_s}{C_s}}{\frac{Y_t-Y_s}{Y_s}}}$$

(2)

In the formula, ε_t is the Tapio decoupling elasticity index for year t, ΔC means the difference in carbon emissions, and ΔY represents the difference in industrial added value. C_t and C_s are the carbon emissions for year t and year s, respectively. Y_t and Y_s indicate the total industrial production value of each region for year t and year s, respectively. The integrated decoupling state and its corresponding numerical characteristics are shown in Figure 3.

3.3. The Standard Deviation Ellipse

The standard deviational ellipse is a classical spatial statistical method initially proposed by Lefever (1926), primarily employed to characterize the directional patterns, dispersion degree, and spatiotemporal evolution of spatial distributions for geographic elements [48]. It is mainly based on the spatial configuration and locational attributes of the research object and quantitatively reveals the development trend and directional features of the variables through various factors such as the center of gravity, major and minor axes, area, azimuth, and flatness of the standard deviation ellipse. The specific calculation formulas are listed as follows:

The standard deviation formula:

$${\sigma }_{\mathrm{X}}=\sqrt{{\displaystyle \frac{\sum _{i=1}^n({\overline{X}}_i\mathrm{c}\mathrm{o}\mathrm{s}\theta -{\overline{Y}}_i\mathrm{s}\mathrm{i}\mathrm{n}\theta {)}^2}{\mathrm{n}}}}$$

(3)

$${\sigma }_{\mathrm{Y}}=\sqrt{{\displaystyle \frac{\sum _{i=1}^n({\overline{X}}_i\mathrm{s}\mathrm{i}\mathrm{n}\theta -{\overline{Y}}_i\mathrm{c}\mathrm{o}\mathrm{s}\theta {)}^2}{n}}}$$

(4)

The azimuth angle formula:

$$\mathrm{t}\mathrm{a}\mathrm{n}\theta ={\displaystyle \frac{\left(\sum _{i=1}^n{\overline{X}}_i^2-\sum _{i=1}^n{\overline{Y}}_i^2\right)+\sqrt{{\left(\sum _{i=1}^n{\overline{X}}_i^2-\sum _{i=1}^n{\overline{Y}}_i^2\right)}^2+4{\left(\sum _{i=1}^n{\overline{X}}_i{\overline{Y}}_i\right)}^2}}{2\sum _{i=1}^n{\overline{X}}_i{\overline{Y}}_i}}$$

(5)

The oblateness formula:

$$e=({\sigma }_x-{\sigma }_y)/{\sigma }_x$$

(6)

In the formulas, n means the total count of urban units; σ_x is the standard deviation of the X-axis; σ_y shows the standard deviation of the Y-axis; θ indicates the azimuth angle of the standard deviation ellipse; and X_i and Y_i are the coordinate deviations from the geographic coordinates of each city to the center of the ellipse, respectively.

3.4. The Gray Prediction Model

The gray prediction model, originally developed by Julong (1989) within the framework of gray system theory [49], represents a mathematical modeling approach that constructs differential equations using limited and incomplete datasets to forecast future trends.

This method extracts underlying patterns from small samples through accumulated generating operations, requiring merely four or more data points to generate reliable predictions. Thereby, it may overcome the stringent requirements of conventional time series analysis regarding large datasets and stationarity assumptions. Notably, this approach exhibits strong complementarity with the LMDI decomposition method. While LMDI decomposition effectively identifies the driving factors behind carbon emissions, the gray prediction model can provide a robust forecasting of future emission trends. This dual analytical framework offers valuable scientific support for the formulation of carbon peaking policies.

In this study, the gray prediction model (GM (1,1)) was used to predict the carbon emission level of the YREB in 2030. The calculation details are listed as follows.

Perform first-order accumulation on data to generate a data sequence:

$$X^{\left(1\right)}\left(k\right)=\sum _{i-1}^k{X}^{\left(0\right)}\left(i\right)$$

(7)

Calculate the nearest neighbor mean of variable X⁽¹⁾(k) [1] to generate variable Z⁽¹⁾(k) [50]:

$$Z^{\left(1\right)}(k)={\displaystyle \frac{1}{2}}\left(X^{\left(1\right)}\left(k\right)+X^{\left(1\right)}(k-1)\right)$$

(8)

Establish a differential equation model:

$${\displaystyle \frac{d{X}^{\left(1\right)}\left(k\right)}{dt}}+a{X}^{\left(1\right)}\left(k\right)=u$$

(9)

Estimate parameters a and u using the least squares method:

$$\begin{gathered} \left.B=\left[\begin{array}{c}-Z^{\left(1\right)}\left(2\right)1\\ -Z^{\left(1\right)}\left(3\right)1\\ \dots \\ -Z^{\left(1\right)}(.)1\end{array}\right.\right] \\ Y={\left[X^{\left(0\right)}\left(2\right),X^{\left(0\right)}\left(3\right){,\dots ,X,}^{\left(0\right)}\left(n\right)\right]}^T \\ {\left[au\right]}^T={\left[B^{T}B\right]}^{-1}{B}^{T}Y \end{gathered}$$

(10)

Establish a prediction model based on the differential equation model:

$${\stackrel{\wedge }{X}}^{\left(1\right)}(k+1)=\left(X^{\left(0\right)}(1)-{\displaystyle \frac{u}{a}}\right)e^{-ak}+{\displaystyle \frac{u}{a}}$$

(11)

Accumulate a reduction in parameter a to obtain the actual predicted value:

$${\stackrel{\wedge }{X}}^{\left(0\right)}(k+1)={\stackrel{\wedge }{X}}^{\left(1\right)}(k+1)-{\stackrel{\wedge }{X}}^{\left(1\right)}(k)$$

(12)

Herein, k = 1, 2, …, n. In order to ensure the accuracy of the prediction results, error testing was conducted. In the posterior error test, the mean of X⁽⁰⁾(k) is determined by the following formula:

$$\overline{X}={\displaystyle \frac{1}{n}}\sum _{k=1}^n{X}^{\left(0\right)}(k)$$

(13)

The mean of ε⁽⁰⁾(k) is calculated using the following formula:

$$\overline{\epsilon }={\displaystyle \frac{1}{n}}\sum _{k=1}^n{\epsilon }^{\left(0\right)}(k)$$

(14)

Therefore, the mean squared error ratio is listed as follows:

$$\begin{gathered} Z={\displaystyle \frac{S_2}{S_1}} \\ {S}_1^2={\displaystyle \frac{1}{n}}\sum _{k=1}^n{\left[X^{\left(0\right)}\left(k\right)-\overline{x}\right]}^2 \\ {S}_2^2={\displaystyle \frac{1}{n}}\sum _{k=1}^n{\left[X^{\left(0\right)}\left(k\right)-\overline{\epsilon }\right]}^2 \end{gathered}$$

(15)

In the formula, S₁ is the standard deviation of the original data, and S₂ is the residual standard deviation.

3.5. Decomposition Method of Decoupling

The LMDI decomposition method is a residual-free analytical approach that quantifies the contribution of individual driving factors to changes in carbon emissions. It is an improvement of the original approach [51] on the basis of Kaya’s identity [52], which can clearly identify and quantify the contribution of each influencing factor to the change in the target variable. According to the relevant reports [53,54,55,56], this study intends to select influencing factors from four aspects, including energy, economy, industry, and population.

The formula for the LMDI decomposition method is listed as follows:

$$C=\sum _{\mathrm{i}}({\displaystyle \frac{C_i}{E_i}}\times {\displaystyle \frac{E_i}{E_t}}\times {\displaystyle \frac{E_t}{Y}}\times {\displaystyle \frac{Y}{G}}\times {\displaystyle \frac{G}{P}}\times P)=\sum _{\mathrm{i}}C{I}_{\mathrm{i}}\times E{S}_{\mathrm{i}}\times EI\times I\times E\times P$$

(16)

Among them, C_i shows the carbon emissions generated by the i-th energy source; E_t is the total energy consumption; E_i means the consumption of the i-th type of energy; G represents GDP; P reveals the total population of the region; CI_i indicates the i-th energy carbon emission coefficient; ES_i expresses the energy structure of the i-th type of energy; EI stands for energy intensity; I denotes the ratio of the added value of the industrial sector to GDP, which is also the scale indictor of a specific industry; and E presents the regional real per capita GDP, which refers to economic growth.

The additive effect of the LMDI method refers to decomposing the change in carbon emissions from year s to year t to obtain the contribution of different factors to carbon emissions. The formula for the additive effect of the LMDI method is listed as follows:

$$\Delta \mathrm{C}=C_t-C_s=\Delta \mathrm{C}\mathrm{I}+\Delta \mathrm{E}\mathrm{S}+\Delta \mathrm{E}\mathrm{I}+\Delta I+\Delta E+\Delta P$$

(17)

In the above formula, ∆C represents the change in carbon emissions, while C_t and C_s are the carbon emission changes in the t-th and s-th year, respectively. ∆CI means the carbon emission coefficient effect, and ∆ES indicates the energy structure effect, whereas ∆EI expresses the energy intensity effect. ∆I represents the industrial structure effect, and ∆E denotes the economic growth effect, while ∆P shows the population size effect.

The decomposition formulas for each influencing factor in the above formula are listed as follows:

$$\Delta CI=\sum _i{\displaystyle \frac{C_{it}-C_{is}}{ln{C}_{it}-ln{C}_{is}}}ln{\displaystyle \frac{{CI}_t}{{CI}_s}}$$

(18)

$$\Delta ES=\sum _i{\displaystyle \frac{C_{it}-C_{is}}{ln{C}_{it}-ln{C}_{is}}}ln{\displaystyle \frac{{ES}_t}{{ES}_s}}$$

(19)

$$\Delta EI=\sum _i{\displaystyle \frac{C_{it}-C_{is}}{ln{C}_{it}-ln{C}_{is}}}ln{\displaystyle \frac{{EI}_t}{{EI}_s}}$$

(20)

$$\Delta I=\sum _i{\displaystyle \frac{C_{it}-C_{is}}{ln{C}_{it}-ln{C}_{is}}}ln{\displaystyle \frac{I_t}{I_s}}$$

(21)

$$\Delta E=\sum _i{\displaystyle \frac{C_{it}-C_{is}}{ln{C}_{it}-ln{C}_{is}}}ln{\displaystyle \frac{E_t}{E_s}}$$

(22)

$$\Delta P=\sum _i{\displaystyle \frac{C_{it}-C_{is}}{ln{C}_{it}-ln{C}_{is}}}ln{\displaystyle \frac{P_t}{P_s}}$$

(23)

To further understand the contribution of various influencing factors to the decoupling elasticity index between energy carbon emissions and socioeconomic development, Equation (18) is substituted into Equation (2). Then, the decomposition formula for the contribution of various influencing factors of carbon emissions to decoupling is listed as follows:

$$t={\displaystyle \frac{(\Delta CI+\Delta ES+\Delta EI+\Delta I+\Delta E+\Delta P)/C_s}{(Y_t-Y_s)/Y_s}}=t_{\mathrm{C}\mathrm{I}}+t_{\mathrm{E}\mathrm{S}}+t_{EI}+t_I+t_{\mathrm{E}}+t_P$$

(24)

In the above formula, t_CI, t_ES, t_EI, t_I, t_E, and t_P are the contribution of the six driving effects to the decoupling elasticity index, respectively.

4. Results

4.1. The Evolution Characteristics of Carbon Emissions in the Years 2006–2022

In order to more intuitively demonstrate the evolution characteristics of carbon emissions in 2006–2022, a time development chart of the total carbon emissions in the YREB was drawn (Figure 4). It is evident that the total carbon emissions of the YREB show a periodic fluctuation and upward trend and are relatively stable. Specifically, carbon emissions increased from 2780.69 Mt in 2006 to 4385.04 Mt in 2022, reflecting the energy and environmental pressures faced by the YREB. This also highlights the challenges of energy conservation, emission reduction, and environmental protection policies, as well as the urgency of international cooperation and global climate governance.

The following analysis will be conducted in stages. During the rapid growth phase (2006–2011), the carbon emissions of the YREB increased from 2780.69 Mt to 4001.99 Mt, with a growth rate of up to 43.9%. During the relatively stable stage (2012–2015), the carbon emissions remained relatively stable with minimal fluctuations. This may be related to the government’s focus on improving energy efficiency and implementing environmental policies, such as promoting clean energy and strengthening energy-saving management. At the same time, economic growth during this period may rely more on technological progress and industrial structure optimization, rather than simply increasing energy consumption. During the period of fluctuating growth (2016–2022), the carbon emissions once again showed an increasing trend, but the growth process was accompanied by some fluctuations. Especially after 2019, there has been a significant increase in carbon emissions, which may be related to economic recovery, increased energy consumption, and possible policy adjustments.

Continuing with the analysis of specific important time points, in 2011, the total carbon emissions continued to grow to 400.19923 Mt, reaching the first peak of this period. In 2014, carbon emissions slightly decreased to 396.3359 Mt, which was the only year of decline. In 2019, the total carbon emissions once again exceeded 400 Mt, reaching 412.72823 Mt. This may be related to factors such as economic recovery, increased energy consumption, and changes in industrial structure. Finally, in 2021 and 2022, the total carbon emissions reached 432.5704 Mt and 438.50404 Mt, respectively, which were the highest values during this period. This may be related to economic growth, increased energy consumption, and possible policy adjustments. At the same time, it also reflects that the YREB still faces significant challenges in promoting green and low-carbon development.

4.2. Results of Carbon Emission Energy Proportion in the Years 2006–2022

From Figure 5, coal, as the main force in energy consumption, accounts for as much as 64.30%. This data not only highlights the important role of coal in the energy structure of the YREB but also reflects its tremendous support for economic and social development. However, with the increasing awareness of environmental protection and the development of clean energy technology, the future role of coal may be challenged. Crude oil, as another important fossil fuel, accounts for 9.15%. Crude oil is not only the main fuel source in the transportation industry but also an important source of chemical raw materials. With the growth of the global economy and population, the demand for crude oil may continue to increase, but this also brings environmental pressures and geopolitical risks. Coke is a product of high-temperature treatment of coal, mainly used in industries such as steel and chemical, accounting for 10.53%, indicating that coke plays an irreplaceable role in industrial production. The production and consumption of coke also reflect the process of industrialization and changes in industrial structure. The proportion of four energy sources, namely gasoline, petrol, diesel, and fuel oil, is relatively low, but with the increase in car ownership and the acceleration of urbanization, the demand for gasoline and petrol may continue to grow. However, with the development and popularization of new energy vehicle technology, this trend may be affected to some extent. Diesel is mainly used for equipment such as heavy vehicles, ships, and generators and still holds an irreplaceable position in specific fields such as logistics and shipping. With the growth of global trade and the development of the logistics industry, the demand for diesel may remain stable or slightly increase. In specific situations such as power shortages or ship transportation, fuel oil still has certain application value. Natural gas, as a clean and efficient energy source, accounts for 4.10%, indicating that its position in energy consumption has not been given sufficient attention. With the strengthening of environmental policies and the development of clean energy technologies, natural gas may become an important component of the future energy structure. From the above analysis, it can be seen that the current energy structure is still dominated by fossil fuels, especially coal and crude oil. However, with the increasing awareness of environmental protection and the development of clean energy technology, this structure may gradually change. In the future, clean energy sources like natural gas may gradually replace some fossil fuels, and the development of new energy vehicle technology will have a certain impact on the demand for gasoline and oil.

Figure 6 demonstrates significant regional variations in energy consumption structures across provinces. Eastern provinces such as Shanghai and Zhejiang show a stronger inclination towards adopting green technologies, primarily benefiting from three key advantages: greater technological and financial capabilities afforded by their higher economic development levels, lighter industrial structures, and stringent environmental policy constraints. A case in point is Shanghai, where clean energy accounted for 34% of total consumption in 2022, alongside the nation’s highest R&D investment intensity. In contrast, western provinces like Guizhou and Yunnan remain predominantly coal-dependent, constrained by both their local resource endowments and current industrialization stages and by inadequate infrastructure development and fiscal reliance on traditional energy sectors.

The research findings confirm a pronounced technological divide between eastern and western regions. Eastern provinces exhibit lower energy consumption per unit GDP compared to their western counterparts, with particularly notable gaps in energy utilization efficiency and smart management systems. To facilitate systemic transformation, it is recommended to establish technology transfer platforms to promote the westward diffusion of mature green technologies from eastern regions while simultaneously improving ecological compensation mechanisms to balance regional development priorities. Special emphasis should be placed on supporting renewable energy infrastructure development and traditional energy substitution in western provinces.

4.3. The Decoupling Effect Analysis Between Carbon Emissions and Socioeconomic Development

How to achieve significant carbon reduction while ensuring socioeconomic development is the core issue that the government needs to address for the long-term sustainable development of the YREB. Based on

Table 2. Changes in the decoupling relationship between carbon emissions and industrial value added.

Sector	Upper Reaches		Middle Reaches		Lower Reaches		YREB
	Index	State	Index	State	Index	State	Index	State
2006–2007	0.3429	II	0.3768	II	0.4091	II	0.3902	II
2007–2008	0.2397	II	−0.0003	I	0.3158	II	0.2110	II
2008–2009	0.8121	VII	0.4123	II	0.6620	II	0.6753	II
2009–2010	0.1839	II	0.4098	II	0.4624	II	0.3725	II
2010–2011	0.2730	II	0.6877	II	0.6014	II	0.5431	II
2011–2012	0.3004	II	−0.4185	I	0.0785	II	0.0181	II
2012–2013	−0.0752	I	−0.6896	I	0.4693	II	−0.0189	I
2013–2014	−0.1408	I	−0.0219	I	−0.1820	I	−0.1292	I
2014–2015	−2.4445	I	−0.2945	I	0.4391	II	−0.4655	I
2015–2016	0.2092	II	0.1772	II	0.3240	II	0.2614	II
2016–2017	0.1315	II	0.2519	II	0.1580	II	0.1675	II
2017–2018	−0.0600	I	0.4507	II	−0.0212	I	0.0624	II
2018–2019	0.5305	II	0.3504	II	0.6269	II	0.5079	II
2019–2020	−0.2147	I	0.9504	VIII	0.0343	II	−1.9606	I
2020–2021	0.4462	II	0.2315	II	0.4257	II	0.3793	II
2021–2022	−0.4321	I	1.4628	VI	0.5311	II	0.4262	II

Note: The means of Roman numerals can be explained as follows: I is strong decoupling; II denotes weak decoupling; III means recessive decoupling; IV stands for strong negative decoupling; V expresses weak negative decoupling; VI represents expansive negative decoupling; VII shows expansive coupling; and VIII indicates recessive decoupling.

, from the perspective of the entire YREB, its decoupling status exhibits certain volatility and complexity across different years. During the inspection period, the entire YREB achieved a decoupling state of carbon emissions. However, it should be noted that 2019–2020 is an abnormal year for the entire YREB. During this period, the whole YREB area performed better than any reaches of the YREB, at −1.9606, indicating a strong decoupling state. While the economy grew, carbon emissions decreased, which was the most ideal state. This may be related to the official passage of the YREB Protection Law of the People’s Republic of China, which elevated the ecological environment protection work in the YREB to an unprecedented level. Through active response from governments at all levels, investment in ecological environment protection in the YREB has been increased, and a series of pollution control and ecological restoration projects have been implemented in the three regions. These measures have significantly improved the ecological environment quality of the YREB.

From the perspective of the upper reaches of the YREB, its decoupling status showed certain volatility and complexity from 2006 to 2022. From 2006–2007 to 2011–2012, the region was in a weak decoupling state (I). In the four time periods of 2012–2015, 2017–2018, 2019–2020, and 2021–2022, the decoupling status of the region was in a strong decoupling state (I); that is, the correlation between socioeconomic development and resource consumption, environmental pollution significantly weakened, and socioeconomic development was almost not dependent on the growth of these negative factors. However, starting from the 2008–2009 fiscal year, the decoupling status in the upper reaches of the YREB showed an expanding coupling (VII) state, where resource consumption, environmental pollution, and economic growth grew synchronously, placing significant pressure on the environment. This may be related to factors such as the special economic environment, policy adjustments, or natural disasters during that period.

From the perspective of the middle reaches of the YREB, the decoupling status during the investigation period also showed diverse characteristics, mainly in the weak decoupling state (I), maintaining a certain balance between socioeconomic development, resource consumption, and environmental pollution. In the years 2007–2008 and 2011–2015, the region achieved strong decoupling (I), and the correlation between socioeconomic development and negative factors significantly weakened. However, from 2019 to 2020, there were significant fluctuations in the decoupling status in the middle reaches of the YREB. Especially in the 2019–2020 and 2021–2022 fiscal years, the region experienced both recurrent decoupling and expansive negative decoupling, respectively. The former caused an economic recession and carbon emissions decreased at a similar pace, which was acceptable. The latter caused significant pressure on the environment.

From the perspective of the lower reaches of the YREB, weak decoupling (I) is the main trend. The region has maintained a certain balance between socioeconomic development, resource consumption, and environmental pollution, and the overall economic growth rate has exceeded the impact of these negative factors. In 2013–2014 and 2017–2018, the decoupling transition in the lower reaches of the YREB also showed strong decoupling (I). During this period, the decline in resource consumption or environmental pollution may have been very fast, placing relatively little pressure on the environment. At the same time, the economy maintained its growth. The possible reason is that during this period, the government strengthened environmental supervision, ensuring that enterprises complied with environmental regulations and reduced pollution emissions through strict law enforcement and inspections.

4.4. Spatial Evolution Characteristics of Carbon Emissions

4.4.1. Characteristics of Spatial Pattern Evolution

To further explore the regional evolution characteristics of the total carbon emissions in the YREB, this study selects the total carbon emissions from four time periods of 2006, 2011, 2016, and 2022 and uses ArcGIS 10.8 software to draw the spatial distribution map of the total carbon emissions in the YREB (Figure 7).

During the research period of this study, there were significant differences in the total carbon emissions of various provinces and cities in the YREB at different time periods and between provinces and cities, presenting an overall diversified development pattern centered on the lower reaches of the YREB. Specifically, in 2006, the carbon emissions of most provinces were at a low level of 200–400 Mt. Especially in Jiangxi Province and Chongqing City, the emissions were below 200 metric tons. And Jiangsu Province was the only province with emissions exceeding 400 metric tons. This may be closely related to the carbon emissions during this period and the level of socioeconomic development at that time. Some provinces with relatively backward economies also had relatively low carbon emissions.

In 2011, during this period, carbon emissions in most provinces increased significantly. The emissions in Jiangsu Province increased from over 400 Mt to over 800 Mt, while those in Zhejiang and Hubei provinces increased from over 200 Mt to over 400 Mt, demonstrating rapid economic growth and industrialization. Meanwhile, some economically underdeveloped provinces also saw a slow increase in their emissions, for example, Jiangxi. Even though provinces adjusted their industrial structure and optimized their energy consumption structure, the effects were not fully evident. In addition, the environmental policies and carbon emission regulations at that time may not have been strict enough, which promoted the growth of carbon emissions.

In 2016, the growth rate of carbon emissions slowed down, mainly represented by Jiangsu Province and Zhejiang Province. During this period, the emissions in Hubei Province even showed a significant downward trend, with only Anhui Province showing a significant upward trend in emissions. Some provinces are also actively adjusting their industrial structure, reducing high energy consuming and high emission industries, thereby reducing carbon emissions. In 2022, the trend of carbon emissions will become more complex. The emissions in Zhejiang and Jiangsu provinces are still relatively high, but the growth rate has significantly slowed down. The emissions in Guizhou Province, Yunnan Province, and Hubei Province are also increasing, but the growth rate is relatively moderate. It is worth noting that emissions in Shanghai are showing a downward trend in contrast to the emission reduction trends in other provinces. From the above analysis, it can be seen that there are significant differences in carbon emissions among different provinces in China at different time periods and between provinces. These differences are related to factors such as the socioeconomic development level, industrial structure, energy consumption structure, and environmental protection policies.

4.4.2. Characteristics of Spatial Distribution Trend of Carbon Emissions

Characteristics of spatial distribution trend: Using ArcGIS to measure the geographic distribution module, we selected the first-level standard deviation and drew the standard deviation ellipse and center of gravity transfer trajectory maps for 2006, 2011, 2016, and 2022 (Figure 8). From Figure 8, it can be seen that the overall carbon emissions distribution center of the YREB during the research period was located in Hubei Province. This is related to the early industrial development in Hubei Province, which was dominated by heavy industry. From 2006 to 2016, it moved towards the northeast direction, with the front part moving at a significantly slower speed than the back part, indicating that the carbon emissions in the middle and lower reaches increased faster than those in the upstream areas. The increase in the front and back parts was faster, mainly due to the rapid socioeconomic development, industrialization, and urbanization processes in various regions of the lower reaches of the YREB during this period, leading to a significant increase in resource consumption. The center of carbon emissions shifted towards the southeast from 2016 to 2022, and the speed of movement was relatively slow, reflecting a significant increase in carbon emissions in some cities of the provinces Anhui and Zhejiang in the southeast direction. The main reason is that the focus of socioeconomic development has shifted as some industries shifted, but energy efficiency utilization rates have not been high in recent years.

The overall level of the YREB during the research period showed a relatively small amplitude of standard deviation ellipse change, showing a stable “Southwest–Northeast” trend. According to

Table 3. Data related to standard deviation ellipses and center of gravity shift trajectory plots.

Year	Length/km	Area/km2	CenterX /km	CenterY /km	XStdDist/km	YStdDist/km	Rotation /° (Degree)	E Value
2006	4159.9492	875,510.3478	817.7390	3220.9766	942.5403	295.7344	150.9295	0.6862
2011	4063.0505	856,734.3180	824.9482	3231.9400	915.5577	297.9168	150.6828	0.6746
2016	4008.0156	837,272.2666	857.8856	3247.7039	902.2877	295.4305	150.4911	0.6726
2022	4019.8073	841,151.4118	864.6010	3240.7714	905.1973	295.8454	151.5102	0.6732

, the overall azimuth of the ellipse shows a decreasing to increasing trend, with the long and short axes shortened by 37.3430 km and −0.1110 km, respectively. The observed reduction in overall ellipticity suggests an increasingly concentrated and directionally oriented spatial pattern of carbon emissions, and the standard deviation ellipse area slightly decreased, also indicating that the spatial distribution of carbon emissions was in a clustered trend. This spatial pattern evolution demonstrated strong alignment with regional economic development strategies, i.e., the northeastward shift of the ellipse centroid reflects the radiating and driving effects of the YREB urban agglomeration as an economic growth pole, while the contraction of the ellipse area confirms the spatial agglomeration effect of carbon emissions under the policy orientation of “prioritizing ecological protection over large-scale development”.

4.4.3. The Characteristics of Local Distribution Trends of Carbon Emissions

This study uses GeoDa software (version 1.20.0.36) to conduct an in-depth analysis of carbon emissions in the YREB in 2006, 2011, 2016, and 2022. By observing Figure 9, it can be observed that the agglomeration characteristics of most cities remained stable, but the agglomeration types of some provinces underwent significant changes. And in most years, the L-H (low-high) cluster phenomenon dominated, while the H-H cluster phenomenon gradually increased over time. Specifically, some underdeveloped areas in the upper and middle reaches of the YREB mainly exhibit L-L cluster aggregation, with relatively low carbon emissions, such as Guizhou Province and Hunan Province. The lower reaches of the YREB have a large industrial scale and significant economic advantages, mainly exhibiting the characteristics of H-H cluster agglomeration, which is highly similar to the economic-level distribution characteristics of the region. A further analysis of the specific situation in each year was conducted.

The spatial analysis revealed that in 2006, the L-H cluster hotspots were predominantly located in the Anhui and Shanghai regions. The L-L cluster carbon emissions are mainly concentrated in regions such as Hunan. Due to relatively backward socioeconomic development, these regions have correspondingly lower levels of carbon emissions. In 2011, L-L clusters appeared in Guizhou Province. The region can rely on its unique climate conditions and natural resource advantages to vigorously develop its tertiary industry, such as tourism, in order to achieve a green transformation and sustainable development of the economy. In 2016, the L-H cluster in Anhui Province transformed into an H-H cluster area. This means that with the development of the secondary industry and the acceleration of urbanization in surrounding cities, it promoted the economy. Especially in some regions, traditional energy sources such as coal are still dominant, and the combustion of these energy sources produces a large amount of greenhouse gases such as carbon dioxide, exacerbating the pressure of climate change. By 2022, Guizhou Province was successfully separated from the L-L cluster area. As one of the important urban clusters in the YREB, the region plays a crucial role in responding to the national “dual carbon” goals. By strengthening energy structure adjustment, promoting industrial transformation and upgrading, and improving energy utilization efficiency, Guizhou Province has achieved significant results in reducing carbon emissions.

5. Influencing Factors of the Carbon Emissions

The LMDI decomposition method is an improvement of the original approach [51] on the basis of Kaya’s identity [52], which can clearly identify and quantify the contribution of each influencing factor to the change in the target variable. According to relevant reports [53,54,55,56], this study intends to select influencing factors from four aspects, including energy, economy, industry, and population.

5.1. Decomposition Results of the Corresponding Influencing Factors

Based on the extended Kaya model and LMDI method, this study decomposes the corresponding influencing factors of carbon emissions in the YREB through Equation (17) and calculates the contribution values of each influencing factor through Equations (18)–(24), as showed in Figure 10. The overall cumulative emission change in the YREB from 2006 to 2022 is estimated as 1604.9658 Mt. The promoting effect of the indicators ES, E, and P on carbon emissions is positive, while the indicators EI and I contribute to the carbon emissions reduction. The ranked order of emission reduction contribution is EI > I > ES > P > E. The decomposition analysis quantified the relative contributions of key influencing factors are listed as follows.

(1) The cumulative effect for the energy structure of ES (170.9519 Mt, 10.64%) is positive. At present, the energy structure of the YREB is still dominated by fossil fuels, including coal and oil, while the utilization rate of clean energy such as solar energy, wind energy, hydropower, etc., is relatively low. The widespread combustion of fossil fuels has led to significant emissions of greenhouse gases such as carbon dioxide, which has undoubtedly hindered the progress of carbon reduction efforts.

(2) The reduced estimation in the energy intensity of EI (−4539.9762 Mt, −28.87%) is a key factor leading to negative growth in carbon emissions, and its emission reduction effect is particularly significant. When energy intensity (EI) shows a negative value, it means that the energy consumption per unit of output is decreasing and energy efficiency is improving. Along with the rapid socioeconomic development in the Yangtze River Delta region, the significant improvement in energy efficiency and the positive transformation of economic structure have jointly promoted the continuous improvement of energy use efficiency. This series of efforts not only contributes to environmental protection but also lays a solid foundation for sustainable socioeconomic development.

(3) The reduced estimation in the added ratio of I (−780.1523 Mt, −48.61%) is also an important emission reduction factor second only to energy intensity (EI). Its negative cumulative effect reveals that the YREB is currently supporting more efficient industrial structure development with less carbon emissions, indicating that the region’s industrial structure is gradually moving towards a new stage of low-carbon and environmental protection. There is a dynamic evolutionary relationship between the adjustment of industrial structure and carbon emissions, with both upward and downward trends towards natural carrying levels.

(4) The corresponding estimation of the economic growth effect on energy intensity for the real per capita GDP of E (6403.8666 Mt, 399%) is similar to that of the indicator P, and the estimated E is the primary indicator for high carbon emissions in most provinces. It is worth noting that the more developed regions in the lower reaches of the Yangtze River benefit more significantly in this regard compared to the other two regions. This also leads to regional differences in carbon emission development, which may affect the emission reduction potential and responsibility of each region. The overall E value of the YREB is positive, which means that energy consumption and carbon emissions increase synchronously during the process of economic growth.

(5) The positive cumulative benefits of P (350.4758 Mt, 21.84%) indicate that carbon emissions are still increasing with population growth. It is necessary to continue to increase emission reduction efforts under the global population growth. The population effect carbon emission contribution order of the lower reaches of the Yangtze River with a larger population size is similar to that of the entire YREB, while the carbon emission contribution values of the upper and middle reaches of the Yangtze River are lower than that of E. The increase in population is often accompanied by the demand for energy, the acceleration of urbanization, and the increasing scarcity of land resources, which in turn leads to an increase in carbon emissions. Therefore, the positive value of population size effect has to some extent suppressed the progress of carbon reduction.

5.2. Decomposition of Carbon Emissions’ Influencing Factors

By using the LMDI factor decomposition method to calculate the changes in carbon emissions in the YREB from 2006 to 2022, as shown in Figure 11, the impact and direction of energy structure effect, energy intensity effect, economic scale effect, output scale effect, and population scale effect on energy carbon emissions in various provinces and cities were obtained. Overall, the economic growth effect shows the most significant carbon increasing effect, while the energy intensity effect and industrial institution effect both exhibit significant carbon reducing effects. The overall effectiveness of the energy structure effect and population size effect is relatively small.

From a regional perspective, the upper reaches, as a crucial ecological barrier and energy supply base in China, have similarities in carbon emissions with the overall trend of the economic belt and also exhibit unique differences. It is worth noting that the energy structure effect has a relatively more significant impact on carbon emissions in the region, not only effectively reducing carbon emission intensity but also promoting the green development of the regional economy. The population size has a negative impact on carbon emissions in the upper reaches of the YREB in certain specific years; that is, population growth does not lead to a synchronous increase in carbon emissions but rather helps to reduce carbon emissions. This achievement is mainly due to the active response of the upper reaches of the YREB to policy calls, relying on abundant natural resources and resource-based industries as the mainstay, gradually reducing dependence on fossil fuels.

The middle reaches are one of the important engines of China’s socioeconomic development. The energy structure effect shifted from having a negative contribution to carbon emissions to a positive contribution in 2021–2022. During the period of 2006–2011, the industrial structure effect continued to show positive values, indicating that the optimization and upgrading of the industrial structure in the region had a positive impact on the growth of carbon emissions. During this period, the middle reaches of the YREB actively responded to the national call to accelerate the transformation of the socioeconomic development mode and vigorously promoted industrial structure adjustment and upgrading. But the effect is not obvious, and it is in a process of continuous exploration. From 2011 to 2022, the industrial structure effect showed negative values.

In the lower reaches, the absolute impact of economic growth on carbon emissions is greater compared to other regions, especially from 2006 to 2011, when this trend was particularly evident. This is mainly attributed to the fact that downstream regions, as the forefront of socioeconomic development, often experience rapid economic growth accompanied by large-scale energy consumption and industrial production activities, leading to an increase in carbon emissions. Although the industrial structure effect is also negative in most years, its absolute value is relatively small but has not significantly affected the carbon emission level of the region and has not made an initial effective contribution to emission reduction. This may be related to the relatively high proportion of heavy industry or high-energy-consuming industries in the downstream industrial structure. The population density effect significantly increases carbon emissions. With the increase in population density, the pressure on the use of roads and transportation networks increases.

5.3. Robustness Testing

To eliminate dimensional differences and enhance the comparability of influencing factors, this study standardized all explanatory variables using the Z-score before incorporating them into a two-way fixed effects model (Model) [57]. Regarding the direction of effects, EI, I, and E all showed significant impacts at the 1% level, with ES, E, and P demonstrating significant positive driving effects on carbon emissions, while EI and I exhibited significant inhibitory effects. In terms of effect magnitude, E showed the most pronounced positive promotion effect on carbon emissions, whereas EI displayed the most outstanding carbon reduction effect.

To ensure the reliability of the regression results, this study employed five methods for the robustness testing (

Table 4. Resulted indicators of the benchmark regression and robustness tests.

Explanatory Variable/Indicator	Model I	Model II	Model III	Model IV	Model V	Model VI
ES	0.072	0.057	0.047	0.082	0.072	0.024
	(0.049)	(0.050)	(0.055)	(0.056)	(0.048)	(0.081)
EI	−0.693 ***	−0.542 ***	−0.496 ***	−0.690 ***	−0.693 ***	−0.218 **
	(0.053)	(0.056)	(0.062)	(0.061)	(0.052)	(0.087)
I	−0.167 ***	−0.143 ***	−0.155 ***	−0.142**	−0.167 ***	−0.170 **
	(0.051)	(0.051)	(0.052)	(0.061)	(0.050)	(0.083)
E	0.729 ***	0.630 ***	0.561 ***	0.736 ***	0.729 ***	0.186 **
	(0.055)	(0.053)	(0.054)	(0.068)	(0.055)	(0.090)
P	0.027	0.051	0.057	0.001	0.027	0.099
	(0.049)	(0.048)	(0.052)	(0.062)	(0.049)	(0.079)
Urban and time fixed effects	YES	YES	YES	YES	YES	YES
N	175	175	175	143	175	165
adj.R2	0.627	0.540	0.463	0.608	—	0.079

Note: ** p < 0.01, *** p < 0.001.

). First, Model II and Model III applied 1% and 2% percentile winsorization, respectively [58]. Second, Model IV excluded the municipality samples (i.e., Shanghai and Chongqing) with the special policy authorities. Third, considering the continuous limited characteristics of the explained variables [59], Model V adopted the Tobit model for alternative regression. Finally, to mitigate endogeneity issues caused by bidirectional causality [59], Model VI introduced one-period lagged explanatory variables for regression analysis [60]. All the tested results confirmed the good robustness of the core findings.

5.4. Gray Prediction Results of Carbon Emission

Through detailed calculations, the carbon emission level of the YREB from 2006 to 2022 was determined. Subsequently, the gray prediction model was used to predict the carbon emission from 2023 to 2030, providing necessary data support for the evaluation of carbon peak measures. By conducting a posterior difference test, the results showed that the mean squared error ratio Z = S2/S1 < 0.35 for all predicted years confirmed that the predictions met the highest level of accuracy standards. Based on the data from the basic gray prediction model, the carbon emission of the YREB is expected to be only 2805.8423 Mt by 2030, and the growth rate is 2.45%, lower than the average growth rate of over 2.96% from 2006 to 2022. This significant change not only reflects the firm determination of the YREB in addressing climate change but also demonstrates its remarkable achievements in promoting green and low-carbon transformation.

The base of carbon emissions in upstream areas is relatively small, and the growth rate is 0.90%, which means that the growth rate of carbon emissions has been particularly slow and will soon reach zero growth (

Table 5. Gray predictions of carbon emission difference (unit: Mt).

Sector	Upper Reaches	Middle Reaches	Lower Reaches	YREB
2023	1107.9426	1012.8994	2369.4073	4485.4716
2024	1117.9283	1027.6861	2427.3298	4566.3426
2025	1128.0040	1042.6887	2486.6684	4648.6718
2026	1138.1705	1057.9103	2547.4575	4732.4852
2027	1148.4287	1073.3541	2609.7327	4817.8098
2028	1158.7793	1089.0233	2673.5302	4904.6727
2029	1169.2232	1104.9213	2738.8874	4993.1018
2030	1179.7613	1121.0514	2805.8423	5083.1252
Z	0.1006	0.1484	0.2561	0.2028

). As an important hub of the YREB, the reduction in carbon emissions in the middle reaches plays a crucial role in the green development of the entire economic belt. Forecast data shows that the carbon emissions growth rate in the middle reaches also shows a downward trend year by year (Table 5). The downward trend of carbon emissions in the lower reaches of the YREB is relatively slow, which may be related to the large economic volume and relatively difficult industrial structure adjustment in the downstream areas.

Therefore, in the future, the focus of emission reduction in the YREB Delta Economic Belt should be placed on the downstream areas. By integrating the gray prediction analysis model, not only has the research and analysis method of carbon emissions in the YREB been improved, but the effectiveness evaluation of the model has also been supported, providing a strong policy support framework for local governments to promote the sustainable development of the YREB. The prediction results demonstrate strong consistency with the driving effects of E and EI identified in the LMDI decomposition, confirming the upstream regions’ potential to achieve carbon neutrality ahead of schedule through industrial structure optimization.

6. Discussion and Policy Recommendations

6.1. Discussion

In terms of the temporal characteristics of carbon emissions, there are significant differences in the total carbon emissions between different time periods and provinces and cities, which is consistent with the view that carbon emissions in Chinese urban agglomerations show significant regional differences [61]. In the direct carbon emissions of the YREB, coal is the main force in energy consumption, accounting for as much as 64.30%. This is closely related to the growth of the economy at the cost of consuming traditional energy. This means that although the government has invested in renewable energy, which can help reduce air pollution, the Chinese economy still heavily relies on the traditional energy sources [62]. This development pattern not only exacerbates climate change risks but also leads to severe resource misallocation, particularly as high-emission regions continue to expand their coal-fired power capacity. The pressing priority is to establish a regionally coordinated renewable energy accommodation mechanism that can transform western China’s clean energy advantages into tangible emission reduction outcomes across the entire basin through both grid infrastructure upgrades and electricity market mechanism innovations.

In terms of influencing factors, it was found that the largest contributor to the increase in carbon emissions is E, while the main reason for reducing carbon emissions is EI, which is consistent with the fact that carbon emissions in the lower reaches of the YREB are significantly higher than those in the other two regions. From a regional perspective, for the upper reaches of the YREB, the impact of energy structure on carbon emissions is relatively more significant. Technological innovation can optimize the energy consumption structure by promoting industrial structure upgrading [63], thereby helping to reduce carbon emissions. Although the population size effect may also reduce carbon emissions to some extent, its effect is not significant. The energy structure transformation in the middle reaches of the YREB will become an effective driving force for carbon reduction with the upgrading of technological means. This is related to the important role that production technology plays in reducing carbon emissions through energy-saving and energy substitution technologies and is an effective tool for controlling carbon emissions [64].

In contrast, the lower reaches of the YREB have a developed economy and the fastest growth in carbon emissions, and all its provinces and cities are facing huge pressure to reduce emissions [65]. Compared with the other two regions, the population size benefits in the lower reaches of the YREB contribute more to carbon emissions, while the carbon emission efficiency is relatively low [66]. This makes the implementation of carbon reduction tasks more difficult. However, we must acknowledge the fact that regional imbalances and inconsistencies remain significant. Specifically, the center of carbon emissions is clearly concentrated in Hubei Province in the middle reaches of the YREB, with its standard deviation ellipse showing a “Southwest–Northeast” trend. The upper and middle reaches of the YREB have long showed an L-H aggregation state, while H-H aggregation is mainly concentrated in the lower reaches of the YREB, which is quite similar to the distribution of economic levels. With the rapid development of the economy in the lower reaches of the YREB, carbon emissions have also increased sharply, leading to increasingly obvious differences between the east, west and center, forming a distribution pattern of “high in the east and low in the west and high in the north and low in the south”. The regional disparities primarily stem from the asynchronous development stages across different areas: the eastern region has transitioned to an innovation-driven development phase with service and high-tech industries as the dominant sectors, while the central region remains in the middle-to-late stage of industrialization characterized by a substantial traditional manufacturing base, and the western region continues to rely heavily on resource-intensive industries. To address these disparities, it is imperative to develop a tailored policy framework that strengthens incentives for low-carbon technology innovation in the eastern region, facilitates the green transformation of traditional industries in the central region, and implements ecological compensation coupled with clean energy substitution in the western region, all supported by enhanced cross-regional collaborative governance mechanisms.

Regarding the overall decoupling status, most of the time, it is in a weak decoupling stage, which means that the low-carbon development level of the YREB still needs to be further improved. Further observation reveals significant differences in decoupling status among different regions. Especially for provinces in the upper and lower reaches of the YREB, they often face a dilemma: either pursue rapid economic growth at the expense of the environment, or reduce carbon emissions at the expense of economic growth, which is far from the ideal state. This difference is an important driving force for the evolution of spatial distribution over time. In addition, this study also found that most provinces within the same region have similar decoupling states, which means that the carbon emissions situation between provinces is similar, and there may be commonalities and correlations in emission reduction mechanisms. Local governments should actively strive to improve and enhance the regional industrial structure, advocate for the construction of an integrated regional market, strengthen capital supervision and collaborative utilization, and improve the efficiency of regional capital allocation in the YREB in order to enhance the regional carbon emission reduction effect [67].

This study aimed to address the core dilemma facing the YREB in achieving its dual carbon goals and how to reconcile rapid socioeconomic development with effective carbon emission control. Our spatiotemporal decoupling analysis revealed the inherent unevenness in regional carbon emissions, while the decomposition of driving factors identified key variables influencing emission growth and reduction, providing a theoretical foundation for predictive modeling. The gray prediction results of 2030 emission differentials directly reflected the dynamic interplay of these factors, i.e., the synergistic effects of economic slowdown and industrial restructuring that would drive the emission reductions, although downstream regions might lag due to the persistent dominance of economic scale effects. This correlation analysis not only validated the scientific rigor of our decomposition results but also provided quantitative support for designing regionally differentiated mitigation policies.

Although this study provided a systematic analysis of carbon emissions in the YREB through multi-model integration, several limitations needed to be acknowledged. First, this research primarily relied on statistical yearbooks, which might not fully capture the current micro-level dynamics of regional carbon emissions. Second, the carbon emission indicator system only covered direct emissions from energy consumption, omitting supply chain emissions and carbon emissions from land use changes. This potentially underestimated the systemic impact of urban carbon footprints. Additionally, the accuracy of the gray prediction model for long-term trends might be affected by external factors like abrupt policy changes.

Future research could deepen this exploration in the following directions. First, it should integrate enterprise-level energy big data with remote sensing monitoring data to construct a real-time, multi-scale carbon emission monitoring system. Second, it should develop a comprehensive carbon emission accounting framework considering both energy-related and non-energy-related carbon emissions. Third, it should create hybrid prediction models, incorporating scenario analysis, machine learning, and other methods to enhance responsiveness to exogenous shocks, such as policy interventions and technological breakthroughs.

6.2. Policy Recommendations

In order to achieve the goals of carbon peak and carbon neutrality as soon as possible, the results of this study propose the following suggestions.

Firstly, the government should balance the development of the economy and carbon emissions. To address the prevalent weak decoupling status in most regions of the YREB, it is essential to promote industrial restructuring and low-carbon technological innovation in energy-intensive sectors to achieve a win–win scenario for economic growth and environmental protection. Some key measures, including improving resource efficiency, advancing green technology R&D and application, establishing a robust carbon emission monitoring and evaluation system, should be employed to regularly track mitigation progress and ensure policy implementation. Additionally, fostering green lifestyles and enhancing public engagement will create a collective societal effort toward emission reduction in the YREB.

Secondly, the government should establish Regional Collaborative Mitigation Mechanisms (RCMMs). For economically advanced downstream regions (e.g., Jiangsu, Zhejiang), priority should be given to clean energy substitution and technological upgrades in high-emission industries through the RCMM, leveraging their economic strengths to achieve deep decarbonization. For midstream regions (e.g., Hubei), they should optimize the energy structures and strengthen the cross-regional technical collaboration through the RCMM, aligning with the industrial relocation trends. For upstream ecologically sensitive areas (e.g., Yunnan, Guizhou), the government can capitalize on natural resource advantages by expanding hydropower and the other renewables while implementing the policy compensation mechanisms to balance development and conservation. Furthermore, a regional coordination framework—such as an interprovincial carbon monitoring platform and an integrated carbon market—should be established to address spatial disparities in emissions.

Finally, the government should optimize the regional energy structure and technological support. It is critical to reduce the coal dependency and scale up the degree in wind, solar, and hydropower adoption. Tailored technical pathways should be developed based on regional characteristics. For instance, Sichuan could focus on wind power development, midstream and downstream regions might prioritize industrial energy efficiency technologies, etc. Special support should be directed toward those areas with high mitigation potential but limited technical capacity.

7. Conclusions

The main conclusions of this study are developed as follows:

(1) The total carbon emissions of the YREB showed a periodic fluctuation and upward trend and were relatively stable. The spatial differences in carbon emissions among provinces and cities were evident, presenting an overall diversified development pattern centered on the lower reaches of the YREB. The center of carbon emissions was located in Hubei Province in the middle reaches of the YREB, and the standard deviation ellipse showed an obvious “Southwest–Northeast” trend. Most provinces have L-H clustering, and H-H clustering increases over time. The spatiotemporal heterogeneity of carbon emissions in the YREB suggested that basin-scale mitigation policies required a differentiated management system of “core–transition–peripheral zones.” This tiered governance model offers new insights for climate governance in transboundary river basins, such as the Nile and Mekong. Coal was the main source of energy consumption, accounting for as much as 64.30% in the YREB. The high coal-dominant energy structure served as a warning for developing countries undergoing rapid industrialization, highlighting the urgent need to prioritize clean energy transitions.

(2) The entire YREB achieved carbon emission decoupling, mainly in a weak decoupling state, maintaining a certain balance between the socioeconomic development, resource consumption, and environmental pollution. From a regional perspective, the decoupling status of the three regions was similar to the overall situation except for a few years. These findings provided a practical model for emerging economies seeking to decouple economic growth from carbon emissions. They demonstrated that through policy interventions and technological innovation, developing countries might achieve relative decoupling during mid-to-late industrialization stages. The YREB experience offered valuable lessons for other rapidly industrializing river basins in Southeast Asia.

(3) For the variables representing influencing factors of the YREB, the indicators ES, E, and P had a positive promoting effect on carbon emissions, whereas the indicators EI and I contributed to emissions reduction. The contribution order was EI > I > ES > P > E. From a regional perspective, the situation in the upper and middle reaches of the YREB was similar to the overall situation, but the impact of energy structure on carbon emissions was relatively more significant in the region. For the lower reaches of the YREB, the absolute impact of economic growth on carbon emissions was greater than the other regions. The quantitative analysis of driving factors provided a reference for prioritizing low-carbon transition policies in river basins at similar development stages globally. It particularly confirmed that EI and I should be core strategies for developing countries, while economically advanced regions must be assumed with greater mitigation responsibilities.

(4) Based on the data from the basic gray prediction model, the carbon emission growth rate of the YREB was 2.45%, lower than the average growth rate of over 2.96% from 2006 to 2022. There was hope of achieving carbon neutrality in the upper and middle reaches carbon emissions, while the downward trend in lower reaches areas was relatively slow and requires increased measures. The prediction results demonstrated that implementing differentiated mitigation strategies at the basin scale yields significant effects, offering an actionable model for global climate governance. For Belt and Road Initiative countries, this “zone-specific approach” could be adapted to set tiered emission reduction targets in economically diverse basins, balancing equity and feasibility. This study further suggested that developed countries supporting mitigation in developing nations should focus their technology transfer and financial support on downstream industrial clusters.