Coupling coordination and spatial network characteristics of carbon emission efficiency and urban green innovation in the Yellow River Basin, China

Abstract

Carbon emission and sustainable development have attracted global attention. Promoting urban green innovation (UGI) in the Yellow River Basin (YRB) will help in lowering the intensity of carbon emissions and improve the safety and sustainability. A SBM-DEA model was constructed to measure carbon emission efficiency (CEE) and the degree of coupling and coordination with UGI was calculated in 73 prefecture-level cities in the YRB. The spatial association network of CEE coupled with UGI is constructed by using a modified gravity model, social network analysis and the quadratic assignment procedure (QAP), to analyze spatial potential energy, network characteristics and clustering characteristics. The study found that: (1) The coupling coordination degree of CEE and UGI in the YRB shows fluctuating growth, mutual promotion and continuous coordinated development. (2) The spatial linkage between CEE and UGI is gradually close, and the potential energy of the spatial linkage increases year by year, with obvious spatial spillover effect, indicating that the radiation and influence between cities are gradually increasing. In contrast to the middle stream, the upstream and downstream regions show a higher percentage of spatial potential energy in the entire network, and their network structure is more intricate and robust. (3) The clustering patterns of the three major urban clusters are examined using the block model, exploring the positioning and functions of various cities in these urban conglomerations, which includes the net spillover, net benefit, two-way spillover and broker plate, so as to strengthen the connection and coordinated development between cities. (4) Factors such as spatial adjacency, industrial structure, population density, digital economy and urbanization level, and energy intensity significantly impact the spatial association network, along with temporal and regional heterogeneity. Therefore, tailored policies are needed in the YRB to strengthen collaboration between CEE and UGI, fostering the development of a circular economy and promoting sustainable development.

Introduction

Climate change and environmental intensification, carbon emissions, and sustainable development have become focal topics worldwide. As one of the most populous countries globally, China estimates that by 2030, over 60% of its population will be urban¹. This acceleration in urbanization and rapid economic growth has increasingly highlighted existing urban carbon emission issues. In response, China officially integrated the “Dual Carbon Plan” into its overall layout for ecological civilization construction in 2020^2,3, emphasizing the urgent need to shift from traditional extensive development to circular economy models⁴. This involves actively promoting industrial innovation and upgrades, fostering pollution reduction and carbon emission decreases, with plans to reduce carbon emissions by 45% from the 2005⁵.

The Yellow River Basin (YRB), located in central China and spanning nine provinces and autonomous regions including Qinghai, Gansu, and Shaanxi, is China’s second-largest river system. The YRB is one of China’s most important economic regions⁶. As of 2020, the regional’s GDP accounted for 25.26% of China’s GDP, showcasing its significant strategic position in terms of economy, agriculture, and ecology⁷. However, extensive development have exposed the YRB to increasing carbon emissions, severe environmental pollution, land degradation, and ecological vulnerability, placing immense pressure on the basin’s water resource carrying capacity⁸, and posing a substantial threat to the region’s green and low-carbon development. Thus, studying the present condition of urban carbon emissions and the measures for carbon reduction in the YRB not only aids in addressing climate change and enhancing environmental quality but also promotes the sustainable growth of the economy and urban development, accelerating the achievement of the “Dual Carbon Goals.”

The Chinese government has been actively promoting an ecological civilization, introducing a range of policy regulations and management measures to foster urban green innovation (UGI). Advancements encompass developing eco-cities, enhancing investments in science and technology education, promoting renewable energy, encouraging clean production, and developing cutting-edge technologies⁹. Existing research identifies UGI as a key factor in achieving sustainable development to reduce resource consumption, lessen environmental pollution, and simultaneously stimulate economic growth¹⁰. CEE is a crucial indicator as it evaluates emissions, reflecting the level of socio-economic development under the constraints of dual carbon goals and resource input. This indicator objectively reflects the development level of a region’s green and low-carbon economy¹¹. Therefore, studying the association between CEE and UGI in the YRB can offer important perspectives for China’s ecological civilization construction and urban carbon reduction policies. Moreover, as the YRB is a vital economic and agricultural region in China, the issues of urban carbon emissions and green innovation in this area impact the social and economic development of the region directly. Researching the experiences in this region can offer valuable lessons for addressing sustainable development issues in similar areas.

Furthermore, with the continuous improvement of transportation infrastructure and the flow of elements between cities, further discussion on the spatial synergy and spillover effects of urban CEE is warranted. This can enhance regional competitive advantages and promote coordinated regional economic development. Based on this understanding, this paper examined 73 prefecture-level cities in the YRB from 2006 to 2020. By measuring the coupling and coordination between CEE and UGI in the YRB, constructing a spatial association network, and studying the spatial spillover effects and the co-evolution of coordinated development in various areas of the basin.This study makes three marginal contributions: (1) It measures the coupling and coordination between CEE and UGI in the YRB from both a holistic and localized multi-scale perspective, moving beyond typical analyses at the national, provincial, or single-city level. (2) It constructs a modified gravity model combined with social network analysis (SNA) to build a spatial association network (SAN) for the coupling and coordination between CEE and UGI in the YRB. (3) It examines the evolutionary characteristics and influencing mechanisms of the spatial connections, exploring spatial spillover effects and regional linkages. This study is organized into six sections. Following the introduction, Sect. 2 presents a review of relevant literature. Section 3 introduces the models and methods employed, while Sect. 4 focuses on analyzing specific spatial relationships, evolutionary characteristics, and influencing factors. Section 5 provides a discussion of the findings, and Sect. 6 concludes the conclusion and policy implications.The conceptual framework is illustrated in Fig. 1.

Fig. 1

Conceptual framework.

Literature review

Measurement and spatial analysis of CEE

The measurement of CEE encompasses both single-factor and total-factor productivity. focuses solely on the amount of CO₂ produced per unit of GDP, neglecting a comprehensive examination of the population’s resource input¹². In contrast, total-factor productivity takes into account economic factors alongside carbon emission constraints and additional costs. CEE serves as a typical indicator that assesses the level of green economic development based on resource input under these constraints¹³. Common measurement methods include the slack-based measure (SBM) model considering undesirable outputs, stochastic frontier analysis, the Malmquist-Luenberger index model. Some researchers have improved on these traditional methods based on introducing polar coordinate approach¹⁴, proposing a radial measurement (RSBM) method. Others have utilized machine learning techniques^15,16, establishing carbon emission intensity prediction models according to factor analysis result, providing scientific theoretical support for carbon reduction policy formulation. Additionally, scholars have examined the spatial differentiation of CEE from macro-level perspectives, exploring the relationship between carbon emission intensity and economic levels across various development stages^17,18. Furthermore, scholars have adopted methods from a provincial level, using multi-regional input-output models to determine bilateral carbon emission intensities, advocating differentiated management approaches, and promoting inter-provincial carbon compensation¹⁹.

At the industry level, scholars have measured and studied CEE within specific sectors, such as the new energy vehicle industry. Findings indicate that capital allocation positively affects CEE in the automotive sector, and technological innovation has a cumulative effect²⁰. Consequently, carbon emissions in dominant industries such as electricity, construction, and agriculture, are critical factors. Researchers thus analyze the development trends, and influencing factors of CEE under different industry forms^21,22. Hence, studying CEE in high-carbon emission industries is meaningful for the overall energy-saving and emission-reduction efforts.

In terms of CEE drivers, scholars primarily have explored factors such as the digital economy²³, which demonstrates a direct nonlinear impact on CEE. Correspondingly, the relationships between factors like clean energy technology innovation, market segmentation and potential^24,25, have been crucial for promoting carbon reduction and green sustainable development. From a methodological standpoint,, due to the spatial distribution features of CEE, scholars have employed spatial Durbin models to analyze driving factors, including their spatial spillover effects on neighboring cities. For instance, researchers have used the STIRPAT and spatial econometric models to analyze factors affecting CEE²⁶, constructing frameworks based on population, economic, and technological characteristics. Findings reveal significant spatial spillover effects of urban CEE in China. Additionally, studies on low-carbon pilot cities show that recommended policies not only improve CEE but also have positive spillover effects on neighboring cities²⁷. Further, models such as Tobit have been employed to analyze impact factors such as resource richness, industrial structure, and urbanization²⁸. Findings suggest that industrial transformation is benefit to sustainable development. Overall, the process is complex, with the concentration of urban population and economic growth having a significant positive impact. Ultimately, environmental regulations can directly reduce carbon emissions through indirect methods such as technological efficiency²⁹.

For scholars adopting a network perspective, focusing on themes such as CEE, or the construction industry, the construction of SANs based on regional or provincial perspectives is instructive. These studies delve into the spatial connections of carbon emissions by analyzing network structure, centrality features, clustering features, and influencing factors, leading to significant recommendations for carbon reduction and sustainable development^30,31. Thus, employing SNA to assist in understanding the spatial connections and coordinated development of urban sustainable development is trending. Measurement methods provide a comprehensive view of the interconnectivity of different regions or entities concerning CEE, offering insights for policy development and strategic planning for sustainable growth.

Urban green innovation

UGI has been proposed as a critical topic in the discussion on sustainable urban development, attracting extensive research interest for its potential to propel low-carbon, eco-friendly, and intelligent growth in cities. Such innovation aims to balance economic growth with reduced carbon emissions and lessened environmental stress, thereby enhancing urban adaptability and resilience³². In the existing literature, measuring UGI primarily involves three approaches:

• Calculating total factor productivity based on a city’s resource inputs and innovation outputs³³.

• Building comprehensive indices to assess urban innovation performance through factors like innovation results and environments³⁴.

• Utilizing green patent applications as a measure, considering both their quantity and quality³⁵.

At China’s provincial level³⁶, existing research have explored the coupling between urban innovation and ecological efficiency³⁷, and analyzes the evolution and factors influencing UGI in cities. Some scholars using spatial econometric models have assessed the effect of UGI on the quality of economic and its spatial spillover effects, highlighting its practical importance in fostering high-quality urban development and transformation³⁸. Furthermore, studies adopting a micro-perspective on the relationship between corporate UGI and regional sustainable development have contributed to regional energy conservation, emissions reduction, and technological advancement^39,40. By adopting a spatial network viewpoint, researchers have constructed UGI correlation networks by employing a complex network analysis that demonstrates the evolution and characteristics of the UGI ecosystem⁴¹. Analyses of factors such as the proportions of the tertiary industry, technological expenditure, and sulfur dioxide emissions indicate that the strength of network connections needs further enhancement, while correlation networks should evolve toward a more complex structure. These findings have also led to targeted policy recommendations aimed at facilitating regional innovation synergy and promoting spatially coordinated UGI.

Relationship between UGI and CEE

UGI serves as a vital tool for promoting sustainable economic development, reducing carbon emissions, and protecting environment⁴². UGI enhances resource efficiency, reduces carbon emissions, and improves environmental quality. Ultimately, green technological innovation positively impacts CEE⁴³ and is crucial in driving its enhancement. Research conducted in 22 developed economies indicates that UGI not only improves resource efficiency but also reduces carbon emissions, thereby fostering sustainable development⁴⁴. Environmental regulation and industrial upgrading also play significant roles⁴⁵, as improvement typically coincides with cleaner air and a better environment. This is especially true since research has shown that air quality pollution, due to the energy crowding-out effect, inhibits urban technological innovation, leading to negative consequences⁴⁶. In response, cities have implemented various strategies to improve CEE, such as carbon emission trading systems and smart city pilot policies. These policies encourage enterprises and research institutions to study environmental protection technologies and innovations, thereby elevating the level of urban innovation, and highlighting mutual interaction and influence⁴⁷. By adopting low-carbon measures and promoting UGI, cities can balance economic growth with environmental sustainability, thus enhancing urban competitiveness, improving residents’ quality of life, and contributing to a sustainable future.

This brief literature review highlights past studies that have predominantly focused on macro-level national or provincial analyses, concentrating mainly on CEE measurement and its influencing factors. However, a notable gap exists in multi-scale approach research, both overall and specific, examining the coupling and spatial connections between CEE and UGI, particularly within the YRB. Additionally, discussions on the associations between CEE and UGI, as well as their spatial effects, primarily revolve around models such as Moran’s I, which explore spatial associations and spillover effects. Therefore, this research addresses the gap by analyzing the SAN coupling CEE with UGI, examining the evolution of spatial associations and spillover issues. It also investigates the clustering characteristics of the YRB and its intra-regions. Exploring the factors influencing the SAN will be significant for managing pollution in the YRB and will provide a foundation for supporting its green and high-quality development.

Materials and methods

Measuring carbon emission efficiency and urban green innovation

Carbon emission efficiency

In present research, the measurement indicators for the desired resource inputs were green economic outputs and unintended total CO₂ emissions of cities in the YRB. The CEE input and output indicators can be seen from Table 1. Based on related literature, four types of resource input indicators were selected: labor, capital, land, and energy⁴⁸. The total number of on-duty staff members represented labor input. For lacking official city statistics on urban capital, the total fixed asset investment of the population was used as a substitute. The base year was set as 2006, and the perpetual inventory model was used for calculation. Land input was represented by the urban built-up area. Additionally, due to the absence of data on the consumption of coal, gasoline, and natural gas in the city yearbooks, we used the total consumption of urban electric power resources for energy input, considering data accuracy and reasonableness. Unlike previous studies, we added an environmental indicator—urban green space—to the expected output indicators, in addition to the economic indicator of regional GDP. The presence of green spaces within the city significantly influences the overall urban carbon sink, facilitating the absorption of CO₂ emissions within the urban environment. It was a critical indicator reflecting the improvement of urban CO₂ air concentration and key in measuring the achievement of green transformational development.

Table 1 CEE input and output index.

Regarding relevant research^49,50, the present research employs the slack-based measure and data envelopment analysis (SBM-DEA) to calculate CEE. This model is a non-oriented and non-radial method that does not mandate a fixed proportion between inputs and outputs. It incorporates undesirable outputs while also differentiating among decision-making units (DMUs) with an efficiency value of 1, thereby inhibiting the loss of information from efficient units. The specific model construction is:

$$\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\rho\:=min\frac{1+\frac{1}{m}\sum\:_{i=1}^{m}\frac{{\omega\:}_{i}^{-}}{{x}_{ik}}}{1-\frac{1}{{r}_{1}+{r}_{2}}\left(\sum\:_{s=1}^{{r}_{1}}\frac{{\omega\:}_{s}^{e}}{{y}_{sk}^{e}}+\sum\:_{q=1}^{{r}_{2}}\frac{{\omega\:}_{q}^{ue}}{{y}_{qk}^{ue}}\:\:\:\right)}$$

(1)

$$\:\:\:\:\:\:\:\:\:\:\:\:\:\:s.t.\left\{\begin{array}{c}{x}_{ik}\ge\:{\sum\:}_{j=1,j\ne\:k}^{n}{x}_{ij}{\lambda\:}_{j}-{\omega\:}_{i}^{-}\:\:,\:\:\:\:\:i=1,...,m\\\:{y}_{sk}^{e}\ge\:{\sum\:}_{j=1,j\ne\:k}^{n}{y}_{sj}^{e}{\lambda\:}_{j}+{\omega\:}_{s}^{e}\:\:,\:\:\:\:\:i=1,...,{r}_{1}\\\:{y}_{qk}^{ue}\ge\:{\sum\:}_{j=1,j\ne\:k}^{n}{y}_{qj}^{ue}{\lambda\:}_{j}+{\omega\:}_{q}^{ue}\:\:,\:\:\:\:i=1,...,{r}_{2}\\\:1-\frac{1}{{r}_{1}+{r}_{2}}\left(\sum\:_{s=1}^{{r}_{1}}\frac{{\omega\:}_{s}^{e}}{{y}_{sk}^{e}}+\sum\:_{q=1}^{{r}_{2}}\frac{{\omega\:}_{q}^{ue}}{{y}_{qk}^{ue}}\:\:\:\right)>0\:\:\:\:\:\:\:\:\:\\\:{\omega\:}_{i}^{-}\ge\:0,{\omega\:}_{s}^{e}\ge\:0,{\omega\:}_{q}^{ue}\ge\:0\end{array}\right.$$

(2)

Here, $\:{\uprho\:}$ represents the CEE of each DMU, which in this context refers to different cities. The number of DMUs is denoted by n. The number of input factors is indicated by m, and $\:{\text{r}}_{1}$ and $\:{\text{r}}_{2}$ represent the desired and undesired outputs, respectively. $\:{{\upomega\:}}_{\text{i}}^{-}$​, $\:{{\upomega\:}}_{\text{s}}^{\text{e}}$​, and $\:{{\upomega\:}}_{\text{q}}^{\text{u}\text{e}}$ respectively mean the slack variables for input factors, desired and undesired outputs. The vectors $\:\text{x}$​, $\:{\text{y}}^{\text{e}}$​, and $\:{\text{y}}^{\text{u}\text{e}}$​ correspond to the matrices of input factors, desired, undesired outputs. The weight vector is represented byλ.

Measuring urban green innovation

Drawing on relevant research³⁵ and integrating the concept of green and sustainable development, a UGI index system was established. UGI is measured by two indicators: the quantity of UGI and the quality of UGI. These indicators collectively assessed a city’s efforts in green development and environmental protection. The quantity of UGI reflects the number of projects, initiatives, policies, or technologies generated in urban development, and demonstrates the city’s diversity and extent of efforts towards sustainable development. The quality of city UGI emphasizes the real output, effectiveness, and sustainability of projects or measures, and highlights the actual impact and long-term sustainability of innovations. Therefore, the UGI quantity is measured via the number of green patent applications filed in a given year, including the number of higher-innovation green inventions and models granted. The quality of UGI is gauged by the number of green patent applications authorized in that year; this approach enabled a comprehensive assessment of a city’s commitment and performance in fostering UGI.

Coupling coordination model

This model is an analytical tool widely used across various disciplines to examine and describe the interdependence and interaction between multiple systems. Its primary aim is to facilitate the understanding of how to achieve balance and collaboration among systems. The coupling relationship between CEE and UGI is complementary, and the coupling coordination association in them is instrumental in reducing carbon emissions, enhancing resource utilization efficiency, and promoting green and sustainable development.

Based on relevant research⁵¹, we constructed a coupling model for CEE and UGI systems. As shown in the following formula, the model aims to quantify the degree of interaction of these two systems:

$$ \:C = \frac{{2\sqrt {CEE \times \:UGI} }}{{CEE + UGI}},\,\,{\mkern 1mu} {\mkern 1mu} 0{\text{ }} { \leqq }{\text{C }} { \leqq }{\text{1}} $$

(3)

In the model, CEE and UGI represent the comprehensive levels of CEE and UGI, respectively, within the YRB. The coefficient k is a tuning parameter and is set to 2 as it only involves CEE and UGI, with C representing the coupling result. The higher the value of C, the greater the mutual influence between the two systems. While the index C reflects the strength of interaction in the systems, it does not reflect the overall coordination. Hence, to better assess the degree of harmony and integration between CEE and UGI, a coordination degree model is proposed. This model goes beyond just demonstrating the interaction intensity; it also evaluates how well these systems are aligned and function together toward common objectives.

$$\:D=\sqrt{C\times\:T}$$,\,\,{\mkern 1mu} {\mkern 1mu} 0{\text{ }} { \leqq }{\text{D}} { \leqq }{\text{1}} $$

(4)

$$ {\text{T}} = \alpha CEE + \beta UGI. $$

(5)

In the model, D denotes the coupling coordination degree, while T signifies the composite coordination index for CEE and UGI. The parameters α, $\:{\upbeta\:}$ are utilized to represent the respective contributions of CEE and UGI to urban sustainable development. Assuming that both factors contribute equally, α, $\:{\upbeta\:}$ are assigned identical values of 0.5. The magnitude of D serves as an indicator of the level of coupling coordination. This model delineates six distinct levels of coordination between CEE and UGI, ranging from extremely uncoordinated to high-quality coordination⁵². These levels are: (1) Extremely uncoordinated, $\:D\in\: (0, 0.2]$; (2) Moderately uncoordinated,$\:D\in\: (0.2, 0.4]$; (3) Slightly uncoordinated,$\:D\in\: (0.4, 0.5]$; (4) Basic coordination,$\:D\in\: (0.5, 0.6]$; (5) Moderate coordination,$\:D\in\: (0.6, 0.8]$ and (6) High-quality coordination $\:D\in\: (0.8, 0.1]$.

Modified gravity model

The determination of association between cities is a fundamental aspect of network analysis. The traditional Vector Autoregression (VAR) model is significantly affected by lag order. VAR-constructed networks fail to reflect the evolutionary trends of SANs⁵³. In this context, we constructed a modified gravity model. In this revised model, $\:{X}_{ij}$ represents the gravitational force in city i and j. $\:{P}_{i}$ and $\:{P}_{j}$ symbolize the total populations of cities i and j at year endpoint, respectively. $\:{C}_{i}$​and $\:{C}_{j}$​ denote the coupling coordination degrees of CEE and UGI for cities i and j. $\:{G}_{i}$ and $\:{G}_{j}$ refer to the GDP of cities i and j. This model is designed to measure the relative importance of CEE in cities i and j.

To construct the SAN via the coupling of CEE and UGI by the gravity model, the following formula is used:

$$\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:{\:\:X}_{ij}={K}_{ij}\frac{\sqrt[3]{{P}_{i}{C}_{i}{G}_{i}}\sqrt[3]{{P}_{j}{C}_{j}{G}_{j}}}{{D}_{ij}^{2}}, \,\, \:{K}_{ij}=\frac{{C}_{i}}{{C}_{i}+{C}_{j}}.$$

(6)

In this study, $\:{D}_{ij}$ refer to the distance in cities i and j. We focus on the SAN coupling CEE with UGI. The multifaceted and complex relationship between a city’s economy, and CEE is highlighted⁵⁴. The spatial association cannot be comprehensively measured using solely the geographic distance between cities. Hence, the concept of spatial economic distance is proper for our research. This metric, $\:{d}_{ij}$, represents the geographic distance of two cities, while $\:{g}_{i}$ and $\:{g}_{j}$ respectively represent the per capita GDP of cities i and j. The formula is:

$$\:{D}_{ij}=\frac{{d}_{ij}}{{g}_{i-}{g}_{j}}$$

(7)

In the SAN coupling CEE with UGI, cities serve as nodes in the network, while the connections between them in the coupled SAN are the network’s links. By constructing a matrix, the mean value of each row is taken as the critical value. Under condition of a value < the threshold value, it indicates no association, the value is 0. Or else, it signifies an association, and the value is 1. Additionally, the relative importance of the coupling coordination in two cities is quantified, resulting in a directed, asymmetric spatial association matrix $\:{K}_{ij}$. Furthermore, the concept of spatial contact potential energy$\:{\:\:X}_{i}$ is defined as the sum of city i spatial contacts with other cities in the YRB, reflecting the city’s overall spatial attraction and radiation capabilities. The calculation formula is:

$$\:{\:\:X}_{i}=\sum\:_{j=1}^{n}{\:\:X}_{ij}$$

(8)

Social network analysis method

SNA primarily targets static networks, analyzing the structure of social networks from the perspectives of graph theory and quantification. This paper evaluates the network structure, features, evolutionary trends of the network coupling CEE with UGI. Additionally, it utilizes block modeling to analyze the clustering characteristics and individual traits of urban agglomerations.

Network structural characteristics

The network characteristics of the coupling between CEE and UGI are analyzed using network density, hierarchy, efficiency as key parameters⁵⁵. The specific meanings and formulas of indicators can be seen from Table 2.

Table 2 Indicators and formulas for network structural characteristics.

Block model

The block model serves to analyze the city’s role and can describe the clustering status of the network based on the number of cities in the network and the relationships outside the block⁵⁶. The present research applied CONCOR (Convergent of Iterated Correlations) for iterative calculations. Assuming that there are different block types, and there are $\:{g}_{k}$ cities in each block $\:{b}_{k}$, the possible association in the block is $\:{g}_{k}({g}_{k}-1)$. Therefore, based on the above results, four distinct types of blocks emerge (as Table 3).

Table 3 Block model clustering evaluation.

QAP analysis

To explore influencing factors, multivariate regression models are typically used. However, due to potential collinearity among explanatory variables and errors introduced by geographical proximity between cities, we employ matrix regression analysis from SNA, specifically the QAP analysis method. QAP is employed to model regression relationships between matrices, offering a conservative estimation of standard deviation in imperfectly structured data⁵⁷. Based on the research theme, we selected geographical adjacency relations, industrial structure, digital economy level, and energy intensity as influencing factors⁵⁸. The equation is:

$$\:W=f(ADJ,IS,POPS,URB,DIG,EI)$$

(9)

In this context, $\:W$ refer to the spatial association matrix of the coupling between CEE and UGI, while $\:AD$ signifies the spatial adjacency matrix of the YRB. Due to shorter geographical distances between adjacent areas, leading to lower transportation and connection costs, the coupling of CEE and UGI is more likely to form spatial connections. These connections radiate and associate with surrounding areas, eventually creating a spatial network with multiple nodes and numerous links. This paper constructs a 0–1 adjacency matrix to characterize the differences in these adjacency relationships. According to similar research⁵⁹, the formula is:

$$ \:AD = \left\{ {\begin{array}{*{20}c} {1,\,\,\,i\,\,\,is\,\,\,adjacent\,\,\,to\,\,j} \\ {0,\,\,{\text{i}}\,\,\:is\:\,\,not\:\,adjacent\:\,to\:\,{\text{j}}} \\ \end{array} } \right.,{\mkern 1mu} {\mkern 1mu} {\mkern 1mu} \:\,i \ne \:j. $$

(10)

$\:IS$ means the Industrial Structure Difference Matrix, it is closely related to resource consumption and pollutant emissions. The secondary industry, mainly comprising steel, mineral resources, is a major source of pollution. A higher proportion of the secondary industry in a region increases environmental protection costs, which is harmful to energy saving and carbon reduction. The industrial structure impacts CEE and also reflects the level of urban industrial innovation. The ratio of the tertiary industry output value to the secondary industry is applied for this purpose. $\:POPS$ denotes the Population Density Difference Matrix, which is the size and density of a city’s population that directly affect its total energy consumption and carbon emissions. Population concentration can also lead to talent aggregation, furthering the city’s UGI level. This is measured by the ratio of the city’s total population at year-end to its land area. $\:DIG$ represents the Digital Economy Level Difference Matrix. As an emerging sector of the tertiary industry, the digital economy directly affects the optimization of industrial structure, and CEE. $\:URB$ stands for the Urbanization Difference Matrix. Here, the development of urbanization positively influences talent aggregation and urban innovation. Urbanization involves factors like infrastructure development, and energy consumption in cities, all of which have complex effects on the CEE. $\:EI$ signifies the Energy Intensity Difference Matrix, which enhances the level of UGI and helps reduce energy intensity, promote energy conservation, improve CEE, and foster a green, low-carbon transition. This is measured by the ratio of total energy consumption to GDP. Furthermore, the difference matrices are constructed based on the absolute differences in each city’s corresponding indicators for 2006 and 2020; all difference matrices undergo standardization.

Research area and data

The sample consisted of 73 prefecture-level cities in the YRB, spanning central China, a region rich in natural resources and cultural heritage. Adopting a holistic-regional, multi-scale analytical perspective, we divided the YRB into three areas: upper, middle, lower reaches (see the location map in Fig. 2). The upper reaches include cities such as Lanzhou, Xining, Yinchuan, and Hohhot, totaling 15 cities; the middle reaches comprise cities like Xi’an, Taiyuan, Yuncheng, and Baoji, totaling 22 cities; and the lower reaches encompass cities like Kaifeng, Jinan, and Qingdao, totaling 36 cities. The data spans from 2006 to 2020 and was sourced from official databases like China Environmental, Construction Statistical Yearbook. Green innovation data are derived from the China Innovation Patent Research Database. The study refers to related research for urban CO₂ emission data⁶⁰. Missing data for a few cities in specific years were supplemented using linear interpolation.

Fig. 2

Location map of the YRB and major cities in the three reaches (The map was drawn using ArcGIS 10.8 software, which can be accessed at https://www.arcgis.com/index.html).

Results

Temporal and spatial evolution analysis of coupling coordination

Based on a temporal evolution perspective, between 2006 and 2020, the CEE of the YRB showed an “N”-shaped fluctuating upward trend, characterized by an initial increase, a decrease, and then another increase (see Fig. 3). Overall, the CEE in the YRB increased from 0.461 to 0.569, with an annual growth rate of 0.72%. UGI showed a phase-based improvement, with the level of UGI rising from 0.255 to 0.407, an annual increase rate of 1.01%. The coupling coordination between CEE and UGI experienced fluctuating growth, progressing through three stages: moderate disharmony, minor disharmony, and initial coordination. From 2008 to 2014, CEE exhibited a downward trend, while UGI increased. However, the coordination between the two remained weak during this period. Notably, in 2013 and 2014, carbon emission efficiency reached its lowest point, and the upward momentum of green innovation also stalled, further intensifying the imbalance between the two. Beginning in 2015, the coordination between CEE and UGI showed significant improvement. CEE gradually rebounded from its low point, while UGI exhibited a stable growth trend. This trend can be linked to China’s national policies on carbon neutrality and ecological civilization, where substantial investments have been made in promoting green innovation.This indicates that the mutual promotion of CEE and UGI has been continuously coordinated and developed, achieving a certain degree of optimization. However, there were differences between the basin’s upper, middle, and lower reaches. The coupling coordination is higher in the lower reaches than in the upper and middle reaches, with the middle region only evolving from moderate to minor disharmony (see Fig. 4). Cities with higher coupling coordination primarily include Qingdao, Xi’an, Jinan, Zibo, Hohhot, and Zhengzhou. These cities focus on introducing advanced technology and attracting high-tech talents, emphasizing the quality of economic development and environmental policy guidance, which is conducive to promoting industrial structure adjustment and nurturing an innovative environment. They effectively leverage market mechanisms and policy guidance to enhance the green innovation levels of enterprises and universities, thereby continually improving CEE. Conversely, cities with lower coupling coordination, such as Qingyang, Ulanqab, Lvliang, Changzhi, and Pingliang, lack universities and research institutes and are deficit in nurturing and attracting high-tech talents. These cities have weaker capabilities in scientific and technological innovation. Moreover, their industrial structures are relatively backward, relying mainly on resources like coal and non-ferrous metals. The issues associated with extensive industrial practices persist, with a high proportion of secondary industries and a need to strengthen industrial pollution emission treatment capabilities, leading to problems of high pollution and emissions.

Fig. 3

Trends in CEE and UGI in the Yellow River Basin from 2006 to 2020.

Fig. 4

Spatio-temporal evolution of coupling coordination between CEE and UGI in the Yellow River Basin for 2006, 2013, and 2020 (The map was drawn using ArcGIS 10.8 software, which can be accessed at https://www.arcgis.com/index.html).

Spatial connection analysis of coupling coordination

Evolution analysis of the spatial association network

By utilizing an improved gravity model and SNA, we constructed a spatially directed association network for CEE and UGI coupling in the YRB. The SAN and topological network maps were created using software tools like ArcGIS and Gephi. From 2006 to 2020, the total number of spatial associations in the YRB increased from 999 to 1036 with the continuous improvement of transportation facilities, the flow of talent, various factors, and industrial interactions between cities. This gradual intensification of spatial associations (see Fig. 5), also exhibited a clear spatial spillover effect. The increase in the number of spatial associations indicates an enhancement in the rationality of resource allocation and the fluidity of resources and factors within the YRB. This is conducive to the collaborative improvement of regional green and low-carbon initiatives and helps to strengthen the stability of the SAN. In this network, the influence on cities is not confined to their internal region but is also impacted by external factors from other regions. Moreover, none of the cities in the network are completely isolated and unaffected by others. This indicates that the coupling of CEE and UGI between cities in the YRB has direct or indirect effects, forming complex network associations. The increasingly robust spatial association and spillover effects identified parallel the policy push for regional cooperation under initiatives such as the Yellow River Basin Ecological Protection and High-Quality Development Plan. The focus on spillover effects among cities highlights the impact of policies like the Regional Coordinated Development Strategy, which encourages resource sharing and innovation collaboration between provinces to balance development across different parts of the YRB. In the entire network, cities like Jinan, Zhengzhou, Qingdao, Baotou, Hohhot, Taiyuan, and Ordos have closer connections with other cities and are central to the network. These cities promote clean production, renewable energy and a circular economy, focusing on the cultivation and introduction of advanced and high-tech technologies. They vigorously pursue pollution reduction and carbon emission reduction efforts, influencing surrounding cities to enhance their CEE. Conversely, cities like Baoji, Jincheng, Nanyang, and Yangquan are on the periphery of the network with sparser connections to other cities.

Fig. 5

Spatial topology structure of the coupling of CEE and UGI in the Yellow River Basin.

In the YRB, the number of internal spatial associations within the upper, middle, and lower reaches has gradually increased over time, with the strength of these connections also growing annually. This evolution has led to the SAN becoming more complex, as depicted in Fig. 6. The area with the most spatial associations is the lower reaches, followed by the upper reaches, with the middle reaches having the least linkages. In the upper reaches, the central cities including Ordos, Wuhai, Yulin, Baotou, and Yinchuan, radiate outward from these central cities. Cities like Lanzhou and Xining also emerge as important nodes in the network, while peripheral cities mainly include Ulanqab and Bayannur. The number of network associations in the upper reaches increased from 101 in 2006 to 114 in 2020, indicating a growing trend that enhances the network’s cohesiveness. In the middle reaches, the number of network relationships increased from 93 to 104, with provincial capitals Xi’an and Taiyuan serving as the main central cities and other cities positioned on the network’s periphery. The SAN in this region exhibits a “dual-core” structure. The lower reaches experienced an increase in network associations from 406 to 441, shifting from a dual-core to a multi-centric. This evolution makes the network more complex and stable, with closer associations. Central cities in the lower reaches include places such as Jinan, Qingdao, Binzhou, Zhengzhou, and Yantai, while cities like Rizhao, Pingdingshan, and Bengbu remained on the network’s periphery. The increased complexity and centrality of the network in the lower reaches reflect a more intensive and interconnected regional development.

Fig. 6

Spatial linkage intensity of coupling CEE and UGI in the upper, middle, and lower reaches of the Yellow River Basin (The map was drawn using ArcGIS 10.8 software, which can be accessed at https://www.arcgis.com/index.html).

Potential analysis of spatial connections

Based on Formula (8), we calculated and analyzed the spatial contact potential energy of CEE in the YRB for 2006, 2013, and 2020 (see the results in Table 4). The potential energy of coupling coordination directly reflects the strength of a city’s spatial associations with other cities. Higher potential energy indicates stronger radiation and attraction effects between the city and others and stronger intensity of spatial associations. From 2005 to 2020, the gravitational strength of the coupling between CEE and UGI gradually increased. The spatial contact potential energy in the YRB rose from 15,181,011.19 to 52,785,565.77, an increase of 247.70%. This significant increase in spatial contact intensity indicates a strengthening of the spatial spillover effect and increasing interactions and connections between cities. The potential energy values of cities in all three reaches increased annually. The lower reaches showed the highest values and proportions of spatial potential energy, indicating that this area is a precursor for the coordinated development of low-carbon growth and GI in the YRB. Although the middle reaches temporarily lag in total potential energy, their growth rate leads to the upper and lower reaches, and their share of overall potential energy is increasing. This suggests that the middle reaches are gradually leveraging their geographical advantages and playing a supportive and connecting role in coordinated development. Cities with higher ranks in spatial contact potential energy, such as Qingdao, Jinan, Binzhou, Yinchuan, Baotou, Taiyuan, and Xi’an, have higher potential energy values and account for a larger proportion in the spatial potential energy of the YRB. These cities prioritize the introduction and cultivation of advanced technology, focus on enhancing UGI performance, emphasize quality economic development and the implementation of pollution reduction and carbon emission reduction, and actively promote green and low-carbon development. This contributes to the coordinated development of CEE and UGI in the YRB and regional collaborative improvement.

Table 4 Spatial correlation potential energy of CEE in the Yellow River Basin in 2006, 2013, and 2020.

Characteristics analysis of the spatial association network

From the perspective of network density (see Fig. 7), the network density of the SAN coupling CEE and UGI in the YRB showed a slight upward trend from 2006 to 2020. The overall network density increased from 0.192 to 0.197, with the order being upper reaches > lower reaches > middle reaches. Network density reflects the closeness of members within a network. The average network density of the middle reaches’ city cluster is only 0.249, indicating fewer connections between cities, which is not conducive to collaborative development among them. Regarding network efficiency, the overall trend for the entire network and the upper and middle reaches is similar, showing a fluctuating downward trend. This indicates an increase in effective relationships within the network, accompanied by a gradual decrease in redundant and ineffective connections, ultimately enhancing network stability. However, the network efficiency of the city cluster in the middle reaches remains relatively high, indicating a disparity in urban innovation capabilities and the implementation of pollution reduction and carbon emission reduction policies between Taiyuan and Xi’an, the dual-core cities, and other cities. The spatial disparity is still significant, and the network structure is relatively fragile, displaying instability. In the lower reaches, the network efficiency fluctuates upward. This indicates that the disparity in the level of UGI between the Shandong Peninsula coastal area and other cities in the lower reaches is gradually widening, as is the disparity in the coordinated development of GI and CEE. This leads to an increase in redundant relationships within the network and a looser association network, which is unfavorable for the coordinated development of green and low-carbon initiatives in the lower reaches.

Fig. 7

Trend graph of network density and network efficiency of the SAN coupling CEE and UGI in the YRB from 2006 to 2020.

From the perspective of network hierarchy, as illustrated in Fig. 8, the overall network hierarchy in the YRB, as well as in the upper and lower reaches, shows an inverted U-shape. This pattern is characterized by an initial fluctuating rise followed by a decline. This indicates that with the flow of factors and talent, and regional interlinkages, the cities within the network mutually leverage each other’s strengths, with subordinate and peripheral cities progressively shifting towards the network’s core, assuming pivotal roles as significant nodes. This diminishes the disparity in network hierarchy within regions, which is beneficial for the coupled and coordinated development of regional CEE and UGI, leading to synergistic regional growth. However, the network hierarchy of the city cluster in the middle reaches is also showing a gradual decline, yet it remains the highest compared to the upper and lower reaches. This suggests that the middle reaches maintain a network structure centered around the dual cores of Xi’an and Taiyuan and are evolving towards a more complex network. The higher network hierarchy in the middle reaches indicates significant differences in the status of cities within the network, with strict hierarchical and affiliation relationships in the connections between cities. Cities in peripheral and subordinate positions have fewer interactions with others, suggesting a need to strengthen inter-regional linkages and the circular economy policy, enhance talent exchaanges, and factor flows. Gradually optimizing this imbalanced network structure will foster synergistic development within the region, resulting in leading to a collective enhancement of CEE and UGI levels.

Fig. 8

Trend graph of network hierarchy of the SAN coupling CEE and UGI in the Yellow River Basin from 2006 to 2020.

Block model clustering analysis

We employed block modeling to further analyze the spatial clustering characteristics of the coupling between CEE and UGI in the YRB, as well as the individual characteristics of cities. Drawing on the block model and other research methodologies proposed by Wasserman and Faust (1994), the SAN of CEE and UGI coupling was divided into four categories based on the number and direction of spatial connections: Net Outflow Block, Net Beneficiary Block, Bidirectional Spillover Block, and Brokerage Block. We used the CONCOR iterative method to perform this analysis. It set the maximum partitioning depth to 2 and the convergence criterion to 0.2. The partitioning results are depicted in Fig. 9.

Fig. 9

Cluster analysis relationship diagram of the association network coupling CEE and UGI in the YRB.

Block I in the spatial association network represents the Net Spillover Block. Cities in this block have a greater proportion of expected internal relationships than actual internal relationships. Additionally, these cities emit more connections to other blocks in the entire network than they receive. Cities such as Lanzhou, Yinchuan, Hohhot, and Yulin in the upper reaches; Xi’an, Baoji, and Luoyang in the middle reaches; and Jinan, Qingdao, Yantai, and Zibo in the lower reaches fall into this category. These cities are characterized by their technological research and development, which is more advanced than in other cities. They actively pursue industrial transformation and place significant emphasis on the introduction of high-tech. This indicates that these cities have a high degree of coupling coordination between CEE and UGI, exerting an influential and leading role in other cities. The identification of net spillover cities like Lanzhou, Xi’an, and Jinan mirrors the reality of these cities being hubs for technology and innovation. These cities have already been targeted by national policies such as the Belt and Road Initiative and the Western Development Strategy, which have incentivized them to lead in green innovation and technological development. Moreover, policies promoting industrial upgrading and high-tech development in these regions enhance the spillover effects observed in the SAN.In the SAN, they primarily play a role in creating spatial spillover effects.

Block II in the SAN is characterized as the Bidirectional Spillover Block. Cities in this block have a higher actual proportion of internal relationships than expected, indicating active connections both emitted to and received from other blocks in the network. This block involves a two-way interaction, where cities influence and are influenced by others. Cities like Baotou, Wuhai, and Ordos in the upper reaches, Taiyuan and Jiaozuo in the middle reaches, and Zhengzhou, Jining, and Weihai in the lower reaches are typical of this role. These cities have created favorable environments for technological innovation. They are proactive in attracting foreign investment and advanced technology, which drives high-quality development and comprehensive policymaking in pollution reduction and carbon emission reduction. Moreover, these cities hold advantageous geographical positions within their regions, leading to increased cooperation and communication with other cities. They not only foster technological innovation, energy saving, and carbon reduction within their jurisdictions but also influence and inspire other cities to do the same. By actively absorbing best practices and simultaneously exerting a significant impact on other cities, they play a pivotal role in both receiving and contributing to the overall progress of the region in sustainable development and technological advancements.

Block III in the SAN is identified as the Broker Block. In this block, the expected proportion of internal relationships is greater than the actual proportion, with cities mainly acting as bridges and intermediaries in the network. They facilitate connections between cities within the network, playing a crucial role in linking different parts of the network. Cities such as Baiyin and Dingxi in the upper reaches, Tongchuan and Weinan in the middle reaches, and Weifang, Binzhou, and Dongying in the lower reaches fall within this category. These cities are instrumental in the SAN, assisting in the scientific distribution and efficient utilization of various resources and factors. They play a significant role in maintaining the stability and connectivity of the entire network. In the context of complex network relationships, these cities serve a vital function in bearing and facilitating interactions and exchanges among different cities.

Block IV in the SAN is recognized as the Net Beneficiary Block. Cities in this block have a lower expected proportion of internal relationships compared to the actual proportion, and they receive more relationships than they emit. These cities are typically on the receiving end of the network dynamics. Cities like Bayannur and Ulanqab in the upper reaches, Yangquan, Changzhi, and Jinzhong in the middle reaches, and Xinxiang, Anyang, and Zhoukou in the lower reaches are good examples. These cities are often characterized by lower levels of economic development and tend to learn production techniques and high-tech practices from other cities. They generally have fewer universities or research institutions serving as bases for high-tech cultivation, leading to limited independent research and development capabilities. Furthermore, these cities often rely on industries such as mining, coal, and chemicals, which are associated with high emissions and pollution. As part of the Net Beneficiary Block, these cities mainly receive influences and initiatives from other cities. To progress, they need to accelerate the optimization of their industrial structures and the cultivation of new industries.

QAP regression results analysis

Utilizing UCINET software, we performed QAP regression analysis with 8000 random permutations. The results of the standardized regression coefficients are displayed in Table 5. Moreover, in 2006 and 2020, all variables passed the significance test. This underscores the comprehensive explanatory power of the six constructed difference matrices of influencing factors regarding the spatial association relationship between CEE and UGI coupling. For detailed regression results, refer to Table 5.

The spatial adjacency matrix ADJ, from the years 2006 to 2020, showed positive standardized regression coefficients for the YRB as well as for its upper, middle, and lower reaches. These coefficients were significant at 5% level. It demonstrates that geographic proximity positively impacts the spatial association relationship between CEE and UGI coupling. It also suggests that the coordinated development of CEE and UGI in the YRB exhibits significant spatial spillover effects and radiating impacts.

From 2006 to 2020, the IS underwent a shift from negative to positive overall, with the upper and lower reaches of the YRB showing a consistent direction. This change suggests that initially, as differences in industrial structures expanded among cities, variations in their capacities for technological innovation, energy conservation, and emission reduction emerged. These variations inhibited the formation of a network coupling CEE and UGI. It implies that cities with similar type of industry are more inclined to construct association relationships in terms of CEE and UGI. With the proposed circular economy framework, cities will accelerate the transformation of cleaner production, enhancing investments in research and development, and cultivating talent. These efforts have not only improved the level of UGI but also positively influenced surrounding cities. The industries in these regions have created momentum and linkages, gradually favoring the strengthening of spatial connections. However, the coefficient for the middle reaches has consistently been positive but small. This indicates that there is room for improvement in the rationality of the industrial structures of cities in this region. The dissemination and learning of advanced production technologies and concepts have been less effective, and barriers to inter-city cooperation persist. These factors contribute less to the formation and stability of spatial association relationships.

From 2006 to 2020, the POPS showed a change in impact coefficients from negative to positive in the overall YRB and the lower reaches. In the upper and middle reaches, the impact coefficients also increased. This demonstrates that as population density continues to expand, the flow and concentration of talent have been strengthened. The circulation of advanced knowledge has intensified with the mobility of talent, leading to tighter inter-regional linkages. Additionally, the increasing proportion of highly educated individuals and the rising rate of higher education have driven continuous improvements in the innovation levels of surrounding cities. This progression aids in the formation of the SAN coupling CEE and UGI.

For the URB in 2006 and 2020, the YRB as a whole and its upper, middle, and lower regions all exhibited positive standardized regression coefficients, which were significant at the 1% level. This reveals that as urbanization levels have continuously risen, differences in urbanization have become increasingly pronounced. Variations in green infrastructure development, pollution and carbon reduction management, and urban technological research and investment have expanded across different areas. Leading cities in energy conservation and emission reduction are progressively influencing others, radiating advanced technologies and stimulating investments in neighboring cities. Together, they are fostering favorable business environments and industrial chain development, thereby enhancing the coupling coordination of CEE and UGI, and strengthening the intensity of spatial association relationships.

From 2006 to 2020, the DIG for the entire YRB, as well as its upper, middle, and lower regions, showed positive standardized regression coefficients that were significant at the 1% level, with a gradually increasing impact coefficient. This indicates that initially, the development level of the digital economy in the YRB was relatively low, with significant regional disparities in digital economy levels. The widening of these disparities was not conducive to the formation of a spatial association network. However, with the continuous development of emerging technologies, such as artificial intelligence and the Internet of Things within the digital economy, these advanced technologies are being integrated into production and daily life. This integration is beneficial for enhancing resource utilization efficiency and fostering innovation in clean production technologies, thereby promoting improvements in CEE. Furthermore, the continuous enhancement of the digital economy level facilitates smoother information flow between regions, which is beneficial for the close connection between CEE and UGI coupling. This result shows that The Digital China Strategy has accelerated urban technological advancement, which is evident in the impact of digital economy levels on the SAN.

In 2006, the EI regression coefficients for the YRB and upper reaches were significantly negative. This suggests that during the earlier stages, the region was still in an extensive development phase with considerable disparities in energy intensity and relatively high energy consumption, which were not conducive to the formation of a SAN for CEE and UGI coupling. By 2020, the standardized regression coefficients for the entire YRB, as well as the upper, middle, and lower reaches, turned positive and were significant at the 1% level. This change indicates that with the gradual advocacy for clean energy and clean production, energy intensity has been optimized. Leading cities have been promoting the reduction of pollution and carbon emissions, as well as energy efficiency. These practices have been increasingly adopted by surrounding cities, leading to a reduction in energy intensity and strengthening the spatial connections in CEE and UGI coupling.

Table 5 QAP regression results in 2006 and 2020. The significance ofthese symbols should be clearly stated, indicating their meaning in the context of the data presented.

Discussion

This study focused on the coupling and spatial connections between CEE and UGI in the YRB to promote the development of circular economy. Green initiatives are instrumental in harmonizing environmental, social, and economic development, and provide substantial support for policy formulation and management. By integrating these approaches, the study contributes to a holistic strategy for addressing ecological and economic challenges in the YRB, setting a path for circular economy progress.

The contributions of the paper are varied. First, this paper is grounded in the theories of green sustainable development and urban economics and CEE measurement systems are based on green economic indicators. Urban green spaces are beneficial for air purification, absorbing greenhouse gases like CO₂, and also impact the total urban soil carbon sequestration, thereby aiding in the improvement of urban CO₂ concentrations. They are also crucial indicators for measuring progress in circular economy and low-carbon economy. Moreover, employing modified gravity model and SNA methods, and constructing SANs at multiple scales from a comprehensive local standpoint, facilitated the analysis of spatial structure and clustering characteristics. This approach provides a nuanced understanding of how green development is interwoven with economic growth, offering insights for policy-making and strategic planning and contributing to further research on regional collaborative development in the YRB.

Secondly, the structure of the association network in the entire YRB, including its upper, middle, and lower reaches, has gradually become more stable and complex. This indicates a dynamic feedback relationship between CEE and UGI, which is continuously interacting and strengthening. The upper reaches are evolving into a networked radial structure, while the middle reaches are characterized by a dual-core structure centered around the provincial capitals of Xi’an and Taiyuan. This evolution suggests that the development of clusters is more pronounced in the upper and middle reaches, and the connections between cities are significantly influenced by spatial and geographical distances. Compared to the upper and middle reaches, the lower reaches exhibit stronger and tighter spatial connections. The region, especially the Shandong Peninsula and areas around Zhengzhou is evolving from a single aggregation to a multi-centric structure. This multi-centric network structure is more conducive to inter-city interactions, facilitating the flow of talent and industry agglomeration. It has a positive impact on enhancing the coupling of CEE and UGI and on regional collaborative development⁶¹. Meanwhile, the lower reaches have the largest share of potential energy in the entire network, indicating that the SAN is predominantly formed around this region as the main area of advantage. It influences and drives the development of the middle and upper reaches, aiding in the coordinated development of CEE and UGI. This gradual evolution towards a multipolar and stable network structure has significant implications for regional development strategies and policy formulations towards sustainable urban development.

Thirdly, focusing on the drivers of the association network, the study constructed difference matrices to investigate the impact of factors such as geographical adjacency, industrial structure, population density, urbanization level, digital economy level, and energy intensity on the SAN coupling CEE and UGI. The influence of these driving factors on the SAN exhibits both temporal and spatial heterogeneity. The difference matrices of URB, POPS, and IS are identified as the main influencing factors on the SAN. This underscores the importance of enhancing talent aggregation and mobility, strengthening technological exchanges and industrial planning cooperation between cities, actively promoting industrial transformation, and increasing research investments. These efforts are crucial for achieving coordinated development in CEE and UGI, as well as for implementing effective emission reduction strategies. These findings align with the results by Liao⁶², highlighting the significance of these factors in practical implementations aimed at green and low-carbon development and environmental management.

Fourthly, while this study quantitatively measures CEE and UGI in the YRB and analyzes the structure of its spatial network, we acknowledge certain limitations. Specifically, our research primarily focuses on the global and local network structures and their evolution, without delving deeply into the characteristics of individual network nodes, such as centrality and core-periphery structure. In future research, these individual network characteristics could be incorporated to provide a more comprehensive analysis. By including these indicators and utilizing advanced data collection techniques, future studies could offer more robust data support, enhancing both the accuracy and depth of the research. Additionally, expanding the range of indicators may yield more nuanced insights into the dynamics of CEE and UGI in other regions, ultimately contributing to more effective policy-making and strategic planning for sustainable development and regional collaborative development.

Conclusion and policy implications

Conclusion

This study constructed a SAN for CEE and UGI coupling in the YRB by utilizing an improved gravity model and SNA methods. The study explored the intricate connections and spatial spillover effects across the basin, revealing the following key findings:

• (1) The coordination degree of CEE and UGI coupling in the YRB showed a fluctuating upward trend. The coordination type progressed through three developmental stages: moderate disharmony, minor disharmony, and initial coordination. This indicates that CEE and UGI mutually promoted and continuously coordinated development, leading to a certain degree of optimization. This indicates that the relationship between CEE and UGI in the YRB is not static; rather, it has evolved dynamically over time, with these two factors mutually reinforcing each other. The spatial distribution of coordination also presented a distinctive pattern of “high on both sides, low in the middle,” pointing to the varying levels of development across the upper, middle, and lower reaches of the YRB. This suggests that while certain areas have progressed, others still face challenges in aligning UGI with CEE.

• (2) The analysis of the network structure revealed that the spatial association of CEE and UGI coupling in the YRB became increasingly close, with the potential energy of spatial connections growing annually and exhibiting clear spatial spillover effects. This suggests a steady increase in the radiating influence and impact among cities. Compared to the middle reaches, the upper and lower reaches have a higher proportion of spatial potential energy in the entire network, with a more intricate network structure. The network efficiency level in the middle reaches is relatively high, indicating a dual-core network structure centered around Xi’an and Taiyuan that has not yet developed to a complex structure. This result shows that the connections among cities in the middle reaches are fewer, which is not conducive to regional collaborative development. This insight emphasizes the need for policy attention to foster a more balanced regional network that can promote greater collaboration and coordination between regions.

• (3) The block model divides the SAN coupling CEE and UGI into net spillover blocks, net beneficiary blocks, bidirectional spillover blocks, and broker blocks, based on the number and direction of spatial connections. Cities such as Lanzhou, Yinchuan, Hohhot, and Yulin in the upstream region; Xi’an, Baoji, and Luoyang in the midstream region; and Jinan, Qingdao, Yantai, and Zibo in the downstream region are classified as net spillovers. These cities lead in technological research and development investments, actively pursue industrial transformation, and focus on introducing high-tech innovations. This indicates that these cities demonstrate a higher level of coordination in CEE and UGI coupling, exerting both an output and a driving effect on other cities, thereby playing a significant spatial spillover role within the network. This block model analysis offers an innovative perspective by highlighting how specific cities serve as regional leaders, driving broader green development through spatial spillover effects, while others act as beneficiaries or intermediaries within the network.

• (4) The QAP regression results illustrate that factors such as spatial adjacency relations, industrial structure, population density, digital economy level, urbanization level, and energy intensity all significantly impact the SAN. Furthermore, temporal and regional heterogeneities exist across the YRB’s upstream, midstream, and downstream regions. Differences in urbanization, population density, and industrial structure are key driving factors for CEE and UGI coupling within the SAN. This analysis highlights the regional heterogeneities and temporal variations across the YRB, offering a more detailed understanding of how these factors influence CEE and UGI coordination in different parts of the basin. The identification of these driving factors points to specific levers that policymakers can use to enhance the effectiveness of green innovation and carbon emission efficiency initiatives.

Policy implications

The research on the YRB indicates a continual evolution and increasing stability in the SAN that couples CEE with UGI. This network varies across the upper, middle, and lower reaches of the river, reflecting the unique characteristics and challenges of each region. From the perspective of this spatial network, several policy implications are recommended:

The study revealed a fluctuating upward trend in the coordination between CEE and UGI, advancing from moderate disharmony to initial coordination. This highlights the need for policies that foster mutual reinforcement between green innovation and carbon efficiency. Developing a comprehensive policy framework that includes carbon emission trading rights, pollution rights, and ecological compensation mechanisms is essential. Additionally, strengthening ecological protection policies will provide a solid foundation for long-term sustainable development. To incentivize green innovation, the government should create tax incentives for businesses that demonstrate high CEE and active UGI engagement. Facilitating the carbon trading market will motivate enterprises to develop low-carbon technologies, forming a market-driven carbon reduction system. Establishing a real-time carbon emission monitoring network, using advanced technologies such as big data and remote sensing, will allow precise tracking of emissions, thereby providing a scientific basis for government and business decisions.

Secondly, the SAN coupling CEE and UGI in the YRB reveals significant differences across its upper, middle, and lower reaches due to varying natural, geographical, and socio-economic conditions. Tailored policy recommendations for each region are crucial. The upper reaches of the YRB play a critical role in water conservation and serve as an ecological barrier, and embrace the accompanying responsibilities. This includes enhancing ecological protection and restoration efforts, and implementing large-scale ecological projects like afforestation and soil and water conservation works to reduce carbon emissions from natural ecosystems. Policies for ecological compensation should be developed to encourage farmer participation in these efforts, along with appropriate compensation. The development and deployment of renewable energy sources, such as small hydropower stations and wind energy, should be actively promoted to reduce reliance on traditional coal energy and improve energy efficiency. Furthermore, modernizing agriculture by guiding farmers towards green planting models like organic and precision agriculture can enhance carbon benefits. In the midstream areas, where socio-economic activities are more intensive, policies should focus on traditional industries to transition towards clean production methods and encourage enterprises to adopt energy-efficient technologies. The establishment of greening projects in industrial parks to enhance the environmental standards of industrial areas is recommended. Increasing support for green technology, and R&D, encouraging innovation in clean energy and environmental conservation fields, and elevating the level of industrial greening are also critical steps. Additionally, strengthening the construction of urban green spaces and green infrastructure to enhance urban carbon absorption and humidity regulation capabilities is essential, along with promoting the greening of public transportation systems. In the downstream areas, emphasis should be placed on technical training and research support to help businesses and farmers understand and adopt green technologies. Strengthening cooperation with research institutions and fostering industry-academia-research-application collaborations are key to driving green technology innovation and professional talent development in these regions. Policies to support the transition of traditional industries to greener practices, and encourage the adoption of clean production technologies, are also vital.

The study identified growing spatial connections and spillover effects, with more robust network structures in the upper and lower reaches of the YRB. Policies should be tailored to enhance these regional connections, particularly by promoting regional collaboration in green technology innovation. Therefore, the collaborative enhancement of CEE and UGI in the YRB is a complex endeavor that requires cooperation and coordination among regions spanning the upper, middle, and lower reaches. It is imperative to actively establish a cross-regional collaborative mechanism among these areas, setting up liaison offices or committees responsible for coordinating carbon emission management and environmental protection efforts. Gradually, the three major urban clusters within the basin should work towards establishing a joint defense and control mechanism, building an information-sharing platform to facilitate the exchange of information and resources between the upstream and downstream areas. Joint research and project cooperation should be encouraged to foster the exchange of green technologies and experiences throughout the basin, collaboratively enhancing CEE and achieving green and low-carbon development.