1 Introduction

The increase of greenhouse gas emissions and climate warming have become a hot issue of global concern [1,2,3]. The global atmospheric concentration of greenhouse gas reached a record high in 2020, and the global concentration of carbon dioxide reached 413.2 ppm and will continue to rise in the future [4]. In recent decades, due to the increase in greenhouse gas emissions, global warming has turned into the main risk to survival and security of mankind and an obstacle to sustainable development, as it leads to environmental issues and disasters [5,6,7,8]. Urbanization is the process of population agglomeration to the city, which brings about a series of economic and social changes [9, 10]. It includes population urbanization, land urbanization and economic urbanization [5, 11, 12]. Urbanization affects the carbon dynamics of regional ecosystems by changing the economic production, lifestyles, and land use [13,14,15,16]. Cities are the core of human economic activity and energy consumption, and more than 50 per cent of the world's population lives in them [17]. Based on the United Nations, the global urbanization rate of 56% in 2021 is projected to increase to 68% in 2050, representing an increase of 2.2 billion people in the global urban population [18]. Cities are strongly associated with more than 67 per cent of global energy use and more than 71 per cent of greenhouse gas emissions [19]. Urban areas generate greater than 60 per cent of the world's material wealth and also give off 75% of the world’s CO2 emissions [20]. The Academy for World Watch has described calls cities "key" to effectively tackling climate change [21]. Global urbanization will continue as population and economy grow [22]. Global urbanization rate of 56% in 2021 is projected to increase to 68% in 2050, representing an increase of 2.2 billion urban population in global [18]. Therefore, it is of great practical significance to explore the relationship between urbanization and CO2 emissions and propose effective carbon emissions reduction measures to reduce carbon emissions and promote sustainable urban development.

Existing studies have assessed the relationship between urbanization and CO2 emissions from two perspectives: quantitative relationship and spatial relationship. Currently, a large number of scholars have utilized statistical methods to analyze the quantitative relationship between urbanization and carbon emissions. Such as the ordinary least squares (OLS) [23, 24], an artificial neural network [25], and Panel Autoregressive Distributed Lag model (ARDL) [26, 27]. Urbanization and carbon emissions, as well as their interplay, often vary depending on the regional economic and social levels, exhibiting significant spatial heterogeneity [7, 28]. Quantitative research maybe overlook the spatial characteristics of the relationship between urbanization and carbon emissions. In terms of spatial relationships, spatial Durbin model is popular method. For example, Chen et al. [29] used the spatial Durbin model to analyzing the effect of population, land and economic urbanization and carbon emissions in China. Zhang et al. [30] explored the effect of population and land urbanization on CO2 emissions in China from 2005 to 2014. Previous studies have provided a solid basis for exploring the relationship between CO2 emissions and urbanization, but there have been some shortcomings. Due to the influence of geographical location and socio-economic development, the relationship between urbanization and CO2 emissions exhibits significant spatial heterogeneity. Analyzing its spatial relationship only on global scale would overlook regional linkages and fail to provide more accurate information to stakeholders [31]. Hence, our core scientific question is what is the spatial relationship between urbanization and carbon emissions, as well as their multiscale characteristics. This will provide decision-making information for large-scale urban development and planning.

Geographically weighted regression (GWR) method is an effective way to assess the spatial relationship between urbanization and carbon emissions at multiple scales spatially. GWR method can build up the spatial relationship between research object and associated influencing factors by constructing the local regression function [32]. GWR takes into account the local effects of relationships between variables and explains the spatial heterogeneity of variables. It can fit the local spatial relationship between independent and dependent variables well. Compared to the quantitative methods [24, 25, 33] and spatial Durbin model [20, 29] used in existing studies, the GWR method can effectively analyze the spatial heterogeneity of the relationship between urbanization and CO2 emissions. It can effectively reveal the characteristics of spatial relationships, providing reference for the formulation of macro urban development plans. However, there is still a lack of research on the multiscale study of the spatial linkages between population, land and economic urbanization and CO2 emissions by using GWR model.

The objective of this study is to take China as an example and utilize the GWR method to access the spatial relationship between population, land and economic urbanization and CO2 emissions on multiple scales. Firstly, we utilized the spatial statistical model to quantify and evaluate the spatial pattern of China’s CO2 emissions in 2019. Then, we constructed the urbanization index based on the statistical data, and spatially quantified the economic urbanization, land urbanization and population urbanization of China in 2019. Finally, we adopted the GWR model to analyze the linkage between land, population and economic urbanization and CO2 emissions on the national, economic zone and provincial scales. Our main contribution in this study is to answer the spatial relationships between land, population, economic urbanization, and carbon emissions at different scales, and to identify their characteristics. Compared to existing single-scale studies [21, 34], we conducted multiscale analyses at national, economic zone, and provincial levels. In contrast to econometric methods [25, 30], we utilized the GWR model to better characterize the spatial relationships between urbanization and carbon emissions. This study will help identify the alignment between urbanization and carbon emissions, providing insights for the coordinated development of urban land, population, and economy.

2 Study area and data

2.1 Study area

The study takes China as the study area. We took the prefecture-level cities as the basic analysis unit. Due to the availability of data, we analyzed 31 provinces except Hong Kong, Macao, and Taiwan, including 407 prefecture-level cities (Fig. 1). The relationship between urbanization and carbon emissions exhibits strong spatial heterogeneity [27, 35]. To identify the characteristics of the relationship between urbanization and carbon emissions at different spatial scales, we analyzed the spatial relationship between urbanization and CO2 emissions at the national, economic zone and provincial levels. Economic zones are divisions delineated by the Chinese government based on the characteristics of regional economic and social development, closely associated with the level of urbanization. Based on the report "Strategies and Policies for Coordinated Regional Development", there are eight economic zones: Eastern Coastal China (ECC), Northern Coastal China (NCC), Middle Reaches of the Yellow River (MRYLR), Northeast China (NEC), Southern Coastal China (SCC), Northwest China (NWC), Middle Reaches of the Yangtze River (MRYTR) and Southwest China (SWC).

Fig. 1
figure 1

Study area

Full size image

China has undergone speedy urbanization in recent years. From 1992 to 2016, China's urban land expand from 1.59 thousand km2 to 8.80 thousand km2, an increase of 7.21 thousand km2, accounting for 20.9% of global growth [36, 37]. The urbanization rate of population increased from 19.4% in 1980 to 51.27% in 2011, with the urban population firstly surpassing the rural. By 2021, the urbanization rate of population in China has risen to 64.72%, surpassing the global average of 56% [38]. China's total national economic output has been increasing its share of the world's total economic output, and the added value of the secondary and tertiary economies has also continued to grow [39]. At the same time, China's carbon emissions also exhibited speedy increase over the past 30 years. From 1978 to 2007, carbon emissions rose by about 3.5 times, from 0.37 billion tons to 1.68 billion tons. In 2009, China overtook the United States as the largest emitter of carbon dioxide. In 2021, China emitted more than 11.9 billion tons of carbon dioxide, or 33% of the global total [40]. In sum, China provides a good case study for exploring the linkage between urbanization and CO2 emissions.

2.2 Data

The data used in this study include carbon emissions, socioeconomic statistics, and geographic information auxiliary data (Table 1). Monthly CO2 emissions data for 2019 were from the ODIAC Fossil Fuel Emission Dataset. The spatial resolution is 1 km, and the unit is tons of carbon per pixel (ton carbon/cell). Economic and social statistics such as built-up area, percentage of secondary and tertiary industries, urban population, and total population were from the China Statistical Yearbook 2020. The administrative boundaries of the entire country, provinces, and cities were taken from the Resource and Environmental Science Data Platform of the Chinese Academy of Sciences.

Table 1 Data resources
Full size table

3 Methods

3.1 Analyzing the spatial pattern of CO2 emissions

First, monthly CO2 emissions data were used to count the total CO2 emissions in 2019 (Fig. 2). Then, we quantified the CO2 emissions of the nation, economic zone, province and city using the zonal statistical method. According to Li et al. [35], we divided the CO2 emissions into five levels, indicating low CO2 emissions (less than 3 million t c), medium–low CO2 emissions (3–6 million t c), medium CO2 emissions (600–1100 t c), medium–high CO2 emissions (11–20 million t c) and high CO2 emissions (more than 20 million t c) respectively.

Fig. 2
figure 2

Flow chart

Full size image

3.2 Quantifying the urbanization

Urbanization is a dynamic evolutionary process involving many factors such as population, resources, and the environment. Population mobility is a manifestation of resources converging in cities, economic development is the primary goal of urbanization, and land use change is the result of population mobility and economic growth [25, 30, 34]. Achieving green or sustainable urbanization requires coordinated development between the population, economic, and resource urbanization subsystems [41, 42]. Referring to Liu et al. [25], we selected three indicators of population urbanization, land urbanization, and economic urbanization to represent urbanization. Land urbanization is the process of converting non-urban land into urban land. During this process, natural ecosystems will be occupied by urban land and urban construction will bring about large amounts of CO2 emissions [25, 43]. Population urbanization presents the course of the rural population continuing to transform and concentrate into cities [44,45,46]. With the agglomeration and increase of urban population, the change of people's lifestyle will lead to additional CO2 emissions [29]. Economic urbanization is the course of industrial structure change. With the continuous advancement of urbanization, agriculture will be transformed into industry or service industry. The development of non-agricultural industries will lead to more CO2 emissions [21]. The formula is as follows.

  1. 1.

    Land urbanization

    $${LU}_{i}=\frac{{UA}_{i}}{{TA}_{i}}\times 100\%$$
    (1)

    where \({LU}_{i}\) is the rate of land urbanization in region \(i\).\({UA}_{i}\) is the built-up area in region \(i\).\({TA}_{i}\) is the administrative area in region \(i\).

  2. 2.

    Population urbanization

    $${PU}_{i}=\frac{{UP}_{i}}{{TP}_{i}}\times 100\%$$
    (2)

    where \({PU}_{i}\) is the rate of population urbanization in region\(i\). \({UP}_{i}\) is the urban population in region\(i\). \({TP}_{i}\) is the total population in region \(i\).

  3. 3.

    Economic urbanization

    $${EU}_{i}=\frac{{UE}_{i}}{{TE}_{i}}\times 100\%$$
    (3)

    where \({ EU}_{i}\) is the rate of economic urbanization in region \(i\).\({UE}_{i}\) is the total output of the second and tertiary industries in region \(i\).\({TP}_{i}\) is the total output value in region \(i\).

3.3 Analyzing the linkage between urbanization and CO2 emissions

Referring to Tu et al. [32] and Song et al. [31], we evaluated the linkage between urbanization and CO2 emissions by the GWR model at the national, economic zone and provincial scales. GWR is a local modeling tool obtained by optimizing the Ordinary Least Squares (OLS) method. OLS reflects a single, homogeneous relationship in space by global regression model. OLS is calculated as follows,

$$y={\beta }_{0}+{\sum }_{k=1}^{n}{\beta }_{k}{x}_{k}+\varepsilon$$
(4)

where \(y\) is the explained variable, and \({x}_{k}\) are the explaining variables. \({\beta }_{0}\) is the intercept. \({\beta }_{k}\) is the local estimator. \(\varepsilon\) is the error. \(n\) is the amount of explaining variables.

Referring to Fotheringham et al. [47], the core function of the GWR model is as follows,

$${y}_{i}={\beta }_{0}\left({u}_{i}, {v}_{i}\right)+{\sum }_{k=1}^{n}{\beta }_{k}{\left({u}_{i}, {v}_{i}\right)x}_{ik}+{\varepsilon }_{i}$$
(5)

where \(\left({u}_{i}, {v}_{i}\right)\) is the coordinates of the sample point \(i. {y}_{i}\) is the dependent variable of the local regression equation at point \(i\). \({x}_{ik}\) are the independent variables of the local regression equation at point \(i\). \({\beta }_{0}\left({u}_{i}, {v}_{i}\right)\) is the intercept. \({\beta }_{k}\left({u}_{i}, {v}_{i}\right)\) are the locally estimated coefficients. \({\varepsilon }_{i}\) is the error. \(n\) is the amount of explaining variables. Due to the location to the sample point will impact the effective of the local regression equation, we used the bisquare weighting function to determine the spatial range of the local regression by referring to Yu et al. [48]. The function is as follows,

$${W}_{ik}={\left(1-{\left(\frac{{d}_{ik}}{b}\right)}^{2}\right)}^{2}$$
(6)

where \({W}_{ik}\) is the spatial weights. \({d}_{ik}\) is the distance between the sample point \(i\) and the proximity point \(k\). \(b\) is the bandwidth. When the bandwidth is larger, the local regression range is larger. When \({d}_{ik}\) is greater than \(b\), the spatial weight function is 0. Based on the Akaike Information Criterion, we can choose the optimal bandwidth.

The GWR model will output local R2, local regression coefficients, adjusted R2, and t-values. Adjusted R2 represents the overall fitness of the model and values range from 0 to 1. The higher adjusted R2 means the model is more adaptable. The regression equation for each sample point is assigned a local R2, local regression coefficient, and t-value, respectively. The t-value can examine the significance of the local regression coefficient of the independent variable. The local regression coefficient represents the positive and negative relationship between the explained and explaining variables. The local R2 means the fitness of the local regression model [32]. There are three hypotheses in our GWR model. First is a dataset containing hundreds of samples. Secondly, there is no multicollinearity between variables. Thirdly, there is spatial non-stationarity in the data. In this study, our model has 407 samples that met the requirements of a large dataset. After multicollinearity tests, the coefficient of variance inflation indexes of each variable is all less than 10, which indicates that no multicollinearity exists. The Moran's index for CO2 emissions of 0.2756 (p-value < 0.01) indicates that the relationship between samples varies with changes in spatial location, indicating spatial non-stationarity.

According to the outputs of the GWR model, we quantified the relationship between urbanization and CO2 emissions in China in 2019. First, we applied prefecture-level cities as the elemental unit of analysis for the local regression, took CO2 emissions in 2019 as the dependent variable, and entered economic urbanization, population urbanization and land urbanization in 2019 as single independent variables into the GWR model, respectively. Referring to Tu et al. [32], we identified regions with significant positive correlation between CO2 emissions and urbanization based on t-values (p < 0.05) and local regression coefficients. The linkage between urbanization and CO2 emissions is strong when local R2 of the areas is greater than 0.75.

Then in areas where the local R2 was greater than 0.75, the relationship between urbanization and CO2 emissions was strong. Finally, we investigated the relationship between urbanization and CO2 emissions at the national, economic zone and provincial scales, respectively. At each scale, the area with significant positive correlation between CO2 emissions and urbanization and the areas with strong relationship between them were counted.

4 Results

4.1 The spatial patterns of CO2 emissions in China in 2019

In 2019, nearly a quarter of the areas in China had medium or higher CO2 emissions. The areas with medium or higher CO2 emissions cover about 2.30 million km2, or 24.35% of the study area (Table 2). Among them, the areas of medium CO2 emissions, medium–high CO2 emissions and high CO2 emissions were 1.42, 0.85 and 0.04 million km2 respectively, or 14.99%, 8.99% and 0.37% of the study area (Table 2).

Table 2 CO2 emissions of economic zone in 2019
Full size table

CO2 emissions showed significant regional differences in China in 2019. The high emissions areas largely concentrated in the eastern region (Fig. 1). Among the eight economic zones, the area of medium or higher CO2 emissions in NCC, ECC and MRYLR was 0.28, 0.16 and 1.00 million km2, respectively, accounting for 76.61%, 74.34% and 59.79% of each economic zone, and higher than the national average of 24.35% (Fig. 3, Table 2). The area with moderate or above CO2 emissions in NWC, SWC and MRYTR accounted for no more than 10% of each economic zone (Fig. 3, Table 2).

Fig. 3
figure 3

CO2 emissions of economic zone in 2019

Full size image

4.2 The urbanization in China in 2019

Land urbanization displayed the pattern of southeast high and northwest low (Fig. 4). In 2019, land urbanization was 0.64% at national scale. Land urbanization in 233 cities was greater than the country level, accounting for 57.25% of the sum of cities in China. At the economic zone scale, land urbanization of the ECC, NCC and SCC were higher at 4.15%, 2.81% and 2.67% respectively, 6.48, 4.39 and 4.17 times the national level (Table 3). At the province scale, land urbanization in 22 provinces were higher than the national level, among which Tianjin and Shanghai both exceeded 10%, at 12.01% and 10.50%, respectively.

Fig. 4
figure 4

Spatial pattern of land urbanization in China in 2019

Full size image
Table 3 China’s urbanization in 2019
Full size table

The distribution of population urbanization was relatively scattered, with high values concentrated in the eastern coastal region and the north (Fig. 5). The population urbanization rate was 61.47% at national scale in 2019. Population urbanization in 153 cities was greater than the country level, accounting for 37.59% of the sum of cities in China. At the economic zone scale, ECC had the highest level of population urbanization at 73.02%, followed by the SCC, NCC and NEC at 69.52%, 64.26% and 63.15% respectively, all greater than the country average. NWC had the lowest population urbanization rate of 50.65% (Table 3). At the province scale, population urbanization in 11 provinces exceeded the national level, with Shanghai, Beijing and Tianjin exceeded 80%, 88.30%, 86.60% and 83.50%, respectively.

Fig. 5
figure 5

Spatial pattern of population urbanization in China in 2019

Full size image

The spatial distribution of economic urbanization was relatively uniform in 2019 (Fig. 6). The economic urbanization rate in China in 2019 was 91.73%. At the economic zone scale, the economic urbanization in ECC was the highest with 96.74%, followed by SCC, NCC and MRYLR with 94.77%, 94.27% and 91.86% respectively. SWC had the lowest economic urbanization rate of 81.19% (Table 3). At the province scale, economic urbanization in 14 was above the national level, with Shanghai and Beijing the highest, both at 99.70%.

Fig. 6
figure 6

Spatial pattern of economic urbanization in China in 2019

Full size image

4.3 The relationships between urbanization and CO2 emissions

In 2019, there was a clear spatial heterogeneity in the relationship between land urbanization and CO2 emissions in China (Fig. 7a). Land urbanization had a significant positive correlation with CO2 emissions in an area of 2.19 million km2, or 23.16% of the study area. At the economic zone scale, SCC, NEC, ECC and SWC had a larger share of land urbanization significantly and positively correlated with CO2 emissions, with 164.80, 378.62, 99.37 and 578.77 thousand km2 respectively, accounting for 49.58, 47.97, 47.50 and 42.44% of the corresponding economic zones (Fig. 7b). At the provincial scale, Shanghai and Jilin had a significant positive correlation between land urbanization and CO2 emissions of 5.59 and 152.77 thousand km2, respectively, or 89.19% and 80.02% of the total area of cities. In addition, Shanghai and Jilin were also the regions with the strongest linkage between land urbanization and CO2 emissions, with 44.71% and 38.55% of areas having a local R2 greater than 0.75, respectively (Fig. 7c).

Fig. 7
figure 7

The relationship between land urbanization and CO2 emissions in China. a Spatial pattern of the relationship; b percentage of area with significantly positive correlation at each economic zone; c percentage of area with significantly positive correlation at each province

Full size image

Population urbanization had a significant positive correlation with CO2 emissions in an area of 2.24 million km2, or 23.77% of the study area (Fig. 8a). The area with a strong relationship between population urbanization and CO2 emissions (local R2 greater than 0.75) was 493.75 thousand km2, or 5.23% of the study area. At the economic zone scale, ECC had the highest proportion of area with significant positive correlation between population urbanization and CO2 emissions, amounting to 173.73 thousand km2, accounting for 83.05% (Fig. 8b). At the provincial scale, Zhejiang had the highest proportion of area with significant positive correlation between population urbanization and CO2 emissions, amounting to 96.78 thousand km2, accounting for 94.91% of the regional area (Fig. 8c).

Fig. 8
figure 8

The relationship between population urbanization and CO2 emissions in China. a Spatial pattern of the relationship; b percentage of area with significantly positive correlation at each economic zone; c percentage of area with significantly positive correlation at each province

Full size image

The area with significant positive correlation between economic urbanization and CO2 emissions was 2.49 million km2, or 31.21% of the study area (Fig. 9a). The area with a strong relationship between economic urbanization and CO2 emissions (local R2 greater than 0.75) was 515.76 thousand km2, or 5.46% of the study area. At the economic zone scale, MRYLR had the highest proportion of area with significant positive correlation between economic urbanization and CO2 emissions, amounting to 1000.30 thousand km2, accounting for 59.76% of the economic zone area (Fig. 9b). At the provincial scale, Zhejiang had the highest proportion of area with significantly positive economic urbanization and CO2 emissions, amounting to 88.78 thousand km2, accounting for 87.06% of Zhejiang (Fig. 9c).

Fig. 9
figure 9

The relationship between economic urbanization and CO2 emissions in China. a Spatial pattern of the relationship; b percentage of area with significantly positive correlation at each economic zone; c percentage of area with significantly positive correlation at each province

Full size image

5 Discussion

5.1 Differences in the relationship between urbanization and CO2 emissions

At the national scale, the areas where land, population and economic urbanization were significantly and positively associated with CO2 emissions were 2.19 million km2, 2.24 million km2 and 2.95 million km2, respectively, or 23.16%, 23.77% and 31.21% of China. Among them, the areas with a strong positive coloration between land, population and economic urbanization and CO2 emissions (local R2 greater than 0.75) were 542.94, 493.75 and 515.75 thousand km2 respectively, or 5.75%, 5.23% and 5.46% of China. Overall, the regions with significant positive correlation were concentrated in the northeast, east and southwest of China, with the similar spatial characteristics of the Heihe–Tengchong Line. This finding is similar to other studies [20, 49]. Population concentration, industrial production and land use change in urbanization will all contribute to significant carbon emissions [25]. To the right side of the Heihe–Tengchong Line was an area with huge population concentration, high urbanization and good economic development. Comparing the spatial pattern of CO2 emissions in 2019, we found that carbon emissions were also significantly higher in the southeast China than in the northwest China (Fig. 2). This is the main reason why the regions with significant positive correlation between urbanization and carbon emissions were concentrated in the southeast China.

At the economic zone scale, the strong relationship between population urbanization and CO2 emissions was concentrated in the southern coastal and eastern coastal economic zone. The population urbanization in these two regions is relatively high, especially in the eastern coastal economic zone. Consequently, the carbon emissions from urban areas for both livelihood and production purposes are also high, indicating a strong correlation. The strong relationship between land urbanization and CO2 emissions was mainly concentrated in the northeastern coastal and eastern coastal economic zone. Northeastern coastal and eastern coastal economic zone are characterized by both a higher quantity and proportion of urban land, which correlates with the area's high carbon emissions [34, 50]. The strong relationship between economic urbanization and CO2 emissions was mostly distributed in the northern coastal and southern coastal economic zone. The southern coastal economic zone is one of the most dynamic regions in China, with a large urban population and industrial base, which is also a key reason for the high regional carbon emissions. These spatial relationships only serve to illustrate the varying dimensions of urbanization and total regional carbon emissions. Per capita carbon emissions may exhibit different relationships. Economically developed areas often exhibit higher resource utilization efficiency, resulting in lower per capita emissions compared to less developed regions [35]. In the eastern coastal and southern coastal economic zone, the areas with a strong relationship between population urbanization and CO2 emissions were 23.39 and 72.58 thousand km2 respectively, accounting for 11.18% and 21.84% of each economic zone. In the northeastern coastal and eastern coastal economic zone, the areas with a strong linkage between land urbanization and CO2 emissions were 109.64 and 44.42 thousand km2 respectively, accounting for 13.89% and 21.23% of each economic zone. In the southern coastal and northern coastal economic zone, the area with a strong linkage between economic urbanization and CO2 emissions were 94.53 and 71.40 thousand km2 respectively, accounting for 28.44 and 19.32% of each economic zone.

At the provincial scale, Yunnan, Gansu and Inner Mongolia were the largest areas with significant positive correlation between land, population and economic urbanization and CO2 emissions, covering of 254.52, 257.48 and 698.32 thousand km2 respectively, accounting for 66.41%, 60.53% and 60.93% of the area of each province. The provinces with the strongest linkages between land, population and economic urbanization and CO2 emissions were Yunnan, Gansu and Sichuan respectively, with local R2 greater than 0.75 covering 75.44, 76.44 and 123.29 thousand km2 respectively, or 19.69%, 17.97% and 25.37% of the area of each province.

Overall, the linkage between urbanization and CO2 emissions varies widely spatially. It also indicated inequalities in the development of land, population or economic urbanization between different regions. Generally, in regions where the relationship between land, population, and economy aligns with urbanization, urban development tends to be relatively balanced [41, 42], meaning there are no issues such as excessive urban land expansion or low utilization of population resources. For example, Zhejiang had a high proportion of areas with significant positive correlation between land, population and economic urbanization and CO2 emissions. This indicated that the level of urbanization in Zhejiang is strongly correlated with its carbon emission intensity, with resource-benefit efficiency matching its environmental impact. For example, the proportion of areas with significant positive correlation between land urbanization and CO2 emissions was very high in Jilin, but the correlation between population and economic urbanization and CO2 emissions was not significant. This partially indicated that population and economic urbanization in the region was less developed than land urbanization, and the regional carbon emissions are mainly caused by land urbanization.

5.2 Implications

The harmonized progress of land, population and economic urbanization can effectively reduce regional CO2 emissions. Our study displayed that there are significant differences in the spatial relationship between land, population and economic urbanization and CO2 emissions in different regions. For example, CO2 emissions are positively correlated with regional land urbanization in Jilin, while population and economic urbanization is less correlated with CO2 emissions. This suggests that the development of regional 'land-population-economy' urbanization is unbalanced, and there may exist the problem that the land urbanization level is not suitable for regional urban population and economic development. This also means that rational planning and control of land urbanization can effectively reduce regional carbon emissions. For cities where there is a strong correlation between land urbanization and carbon emissions (such as Jilin, Liaoning, and Guangxi), there may be a problem of excessive urban land sprawl. In order to promote the coordinated development of urban land, population, and economy, these areas can increase the intensity of land use, improve land use efficiency, rather than expand the urban land area and boundaries.

There are also cities with weak correlation between population urbanization and carbon emissions, such as Hainan and Heilongjiang. Land and economic urbanization in those regions showed a significantly positively linkage with CO2 emissions, while population urbanization did not. The carbon emissions in these areas may be higher due to changes in production or land use. To promote low-carbon urban development, these regions can optimize industrial structure, improve resource utilization, or adjust land use layout. For example, increasing ecological land, controlling industrial land. There are high percentages of areas where land, population and economic urbanization are significantly correlated with CO2 emissions in Guangdong and Zhejiang. These regions can coordinate urban development in terms of land, population and economy, so that urban land scale, population size and economic volume can match each other and systematically optimize urban carbon emissions.

The relationship between economic urbanization and carbon emissions is the closest. Our research indicates that the correlation between economic urbanization and carbon emissions is the highest, which also suggests that economic development is the main source of carbon emissions. This conclusion is similar to some other studies in developing countries, such as Malaysia [33] and India [23]. Furthermore, most studies indicate that there is heterogeneity in the relationship between urbanization and carbon emissions, which may be a kind of inverted U-shaped statistical relationship [24, 51]. Therefore, for regions where urbanization is still rapidly developing, more attention should be paid to the carbon emissions brought about by industrial development. Optimizing industrial structures and improving resource utilization efficiency are key to low-carbon development in these regions.

5.3 Future perspectives

In this study, we utilized a geographically weighted regression model to assess the spatial linkage between urbanization and carbon emissions at multiple scales. We evaluated the linkage between land, population and economic urbanization and CO2 emissions at the national, economic zone and provincial scales respectively, providing an extensive investigation of the relationship between urbanization and carbon emissions. Compared with former research, we quantified the spatial relationship between urbanization and CO2 emissions at three spatial scales, provided a more complete picture of the effect of urbanization on CO2 emissions, and intuitively reflected the characteristics in the linkages between population, land and economic urbanization and CO2 emissions. We found significant spatial heterogeneity in the association between China’s urbanization and CO2 emissions in 2019. The stronger linkage between population urbanization and CO2 emissions was concentrated in the southern coastal and eastern coastal economic zones, the stronger linkage between land urbanization and CO2 emissions was concentrated in the northeastern and eastern coastal economic zones, and the stronger colorations between economic urbanization and CO2 emissions was concentrated in the southern coastal and northern coastal economic zones.

There are still some shortcomings in this study. We have only assessed the spatial statistical relationship between urbanization and CO2 emissions using the GWR model, ignoring the underlying mechanism of influence. The basic statistical unit of this study is the prefecture-level city, which neglects the information at the sub-prefecture level. We mainly analyzed the spatial relationship between land, population, and economic urbanization and carbon emissions, overlooking other relevant indicators, such as green urbanization and carbon emissions per capita or per unit of GDP. The statistical data used in this study is from 2019, a single year lacking analysis of temporal changes in relationships. However, its impact on exploring the spatial relationship between urbanization and carbon emissions is minimal. In the future, we can use higher resolution and long-time remote sensing data to obtain high precision spatio-temporal data on urbanization. Additionally, we can systematically and multi-prospectively analyze the mechanism of the impact of urbanization on CO2 emissions through coupled system dynamics models.

6 Conclusions

We explored the spatial linkage between land, population and economic urbanization and CO2 emissions at national, economic zone and provincial scales by a geo-weighted regression model. The results displayed that the regions with significant positive correlation between land, population and economic urbanization and CO2 emissions were concentrated in the northeast, east and southwest regions, with the similar spatial characteristics of the Heihe–Tengchong Line. The linkage between urbanization and CO2 emissions is obvious spatial heterogeneity. The strong coloration between population urbanization and CO2 emissions was mostly located in the eastern and southern coastal economic zones. The stronger relationship between land urbanization and CO2 emissions was concentrated in the northeastern and eastern coastal economic zones. The stronger coloration between economic urbanization and CO2 emissions was concentrated in the southern and northern coastal economic zones. There was no consistent spatial coloration between land, population and economic urbanization and CO2 emissions. The uneven development of land, population and economic urbanization would lead to more CO2 emissions. We suggest that local governments should be paid for the coordinated development of land, population and economy to avoid additional CO2 emissions resulting from the mismatch between land, population and economy in cities.