Spatial Analysis of Carbon Metabolism in Different Economic Divisions Based on Land Use and Cover Change (LUCC) in China

Abstract

Urbanization has greatly altered Earth’s surface form, and land use changes can lead to significant changes in carbon emissions. However, how these changes affect ecosystems remains unclear. Therefore, this study calculated the carbon absorption and emissions in 31 Chinese provinces using high-resolution (300 m) land use data. Subsequently, a carbon flow model was used to evaluate the carbon transfer that occurred from the changes in land use in every province between 2000 and 2020. The standard deviation ellipse analytic techniques were also employed to research the spatiotemporal evolution features of carbon flow in various economic zones. Furthermore, the flux and utility analysis approaches in ecological network analysis were used to quantitatively examine the interaction relationship between two carbon metabolism land uses. The results revealed that the continuous expansion of China’s construction land has reduced the area of agricultural land, resulting in industrial land (53.14%) and urban land (39.38%) being the main contributors to the total carbon emissions. Among them, the five eastern provinces of Hebei, Jiangsu, Zhejiang, Shandong, and Guangdong had carbon emissions of more than 100 million tons. From 2000 to 2020, the center of gravity of the carbon flow in construction land had shifted significantly from Henan Province to Gansu Province. The ecological relationship of exploitation and control dominated the two land use types. It is mostly found in Xinjiang, Qinghai, Gansu, Inner Mongolia, and Ningxia provinces. The findings could provide relevant policy implications for the Chinese government to mitigate carbon metabolism on land.

1. Introduction

As the social economy develops, increasing carbon emissions cause many ecological and environmental issues, such as prominent contradictions between humans and land, inefficient use of energy and resources, and extreme contamination of the environment. These issues have obstructed the urban economy’s sustainable development [1,2]. Urbanization commonly results in land use changes, which are the primary source of increased urban carbon emissions in cities. If land use is properly optimized, then urban carbon emissions can be effectively reduced, and the issue of global warming can be alleviated [3]. The transfer of materials and energy within an urban system is called urban metabolism [4]. Urban metabolism encompasses resource mining, production, consumption, and waste generation within a city. Baccini [5] originally introduced the concept of urban carbon metabolism in 1996. This concept describes the transfer of carbon between different compartments of the environment and urban system and involves carbon emissions and storage. The equilibrium between these processes establishes whether a city emits or absorbs carbon.

Generally, carbon metabolism research is based on urban carbon budget accounting. With the development of technology, carbon budget has evolved from field investigation and resource collection estimation [6] to model simulation combined with empirical data [7] and 3S technology [8]. Researchers aimed to achieve different research objectives by combining multiple methods and technologies, such as model and remote sensing [9], field investigation and model simulation [10], and field investigation and remote sensing [11], to make estimates highly accurate. Land usually serves as the carrier of all urban activities, and rapid urbanization certainly leads to land use change. Consequently, researching urban carbon metabolism from the standpoint of land use is conducive to urban environmental sustainability [12,13]. Bolin [14] first examined the relationship between land use and cover change (LUCC) and the terrestrial carbon cycle. He found that land change can increase the atmospheric CO₂ concentration. Moreover, Wang et al. [15] constructed the LUCC map of Dongying, a city in Shandong Province (in China). They found that the built-up area land grew sharply.

Urban carbon metabolism typically encompasses various types of metabolic agents and multipath influences, which form a complex network interaction. Therefore, constructing a network model can aid in comprehending the transfer of materials and energy among different components and can effectively assist in identifying the structural distribution and functional relationships of the entire urban carbon metabolism system [16]. Hannon [17] first used ecological network analysis (ENA) to model the internal structure of ecosystems and the interactions among various nutrient levels. Finn [18] refined Hannon’s method and introduced several metrics and indices for quantifying the ecosystem structure, including cycling index, average path length, and system flow, thereby greatly advancing ENA formation. Patten [19] formally proposed ENA and pointed out that this method serves as a means of identifying information related to the flow of matter and energy within systems, enabling the quantitative study of interactions among the participating members within the network. Some researchers quantitatively assessed the improvement of urban landscape ecological networks by comparing ecological networks with green space system planning [20]. Some also employed ENA to build a system of an urban spatial structure and then used this methodology to investigate the urban spatial structure [21].

However, previous studies mainly analyzed land urbanization, economic urbanization, and population urbanization in the course of urbanization [22,23]. Some studies have examined the spatiotemporal distribution pattern of urban carbon emissions. For instance, Gately et al. [24] evaluated the spatial distribution of CO₂ emissions from road transportation in the United States and found that CO₂ emissions from roads and population density grow nonlinearly. Liu et al. [25] used the city-size indices to determine the spatial association networks of CO₂. They also analyzed the spatial association networks of carbon emissions of various city sizes in the Yangtze River Delta from the macro, middle, and micro levels. The effect of changing land use types on carbon emissions has been the subject of several studies. For example, Sanquetta et al. [26] quantified the carbon loss from land use change from primary forest to cultivated pasture in the Amazon region. Qiu et al. [27] studied the spatial distribution of CO₂ emissions during the conversion from peatlands to croplands by simulating the coupling of peatland carbon and land use change in the Northern Hemisphere. However, how carbon emissions change because of high-resolution land use area transfer remains unclear. Xia et al. [28] took 13 cities in China’s Yangtze River Delta as examples. They also used panel data regression analysis to study the association between LUCC and city size growth and urban carbon metabolic rate. Wei et al. [29] investigated the relationship between the land use changes related to Beijing’s urban form development and the spatiotemporal variabilities in carbon flow. They considered the spatial change in carbon flow brought about by the land use change in a single city or region but not the carbon flow situation and ecological relationship among various land use changes across the nation. Moreover, how these changes and economic development interacted was unclear.

In response to the limitations of the previous research, 31 provinces in China during four periods (2000–2005, 2005–2010, 2010–2015, and 2015–2020) were selected as the subjects in this study. The carbon emissions and carbon absorption of every province were calculated based on high-resolution (300 m) national land use area data. Then, a carbon flow model was applied to evaluate the carbon transfer resulting from the changes in land use in various provinces during 2000–2020. In addition, the spatiotemporal evolution of carbon flows across several economic zones was explored using standard deviation ellipse (SDE) analysis. Finally, ENA was used to analyze the ecological relationship caused by land use type change. The research findings may offer a theoretical basis and guidance for land use planning and low carbon emissions in all provinces.

2. Literature Review

In the context of addressing global climate change, different economic zones in China have been studied in depth in terms of their carbon metabolism space due to their unique development paths and land use patterns.

Wei et al. [29], taking Beijing as an example, constructed a research framework integrating ecological network analysis (ENA) and the patch generation land use simulation (PLUS) model, and the results showed that the overall land use in Beijing has been a net carbon sink after 2010, which is mainly due to the industrial upgrading and optimization of the land structure between man-made and cultivated land.

Wang et al. [30] took Beijing–Tianjin–Hebei as an object and applied the InVEST model, combined with climate data and land use data, and found that the significant expansion of built-up area in 2015 led to a decrease in the regional carbon stock from 40.55 × 10⁸ t in 1990 to 40.04 × 10⁸ t, and the carbon density showed a northwestern–southeastern decreasing trend, with 134 counties and districts showing decreases in carbon stock and 26 counties and districts increases, and the urbanization level and carbon sequestration were significantly negatively correlated. Chen et al. [31] studied regional carbon metabolism from 2005 to 2020 by constructing a land use carbon emission network and social network analysis model (SNA) and found that the spatial distribution of the carbon emissions in the region was uneven, with Beijing and Tianjin as the high-emission centers, and the spatial correlation was complex, and the stability of the network and the synergistic emission reduction ability were weak. Huang et al. [32] analyzed the carbon emission influencing factors and predicted carbon emissions with Beijing–Tianjin–Hebei as the object, and used the STIRPAT model and genetic algorithm (GA) optimization based on the Extreme Learning Machine (ELM), and the study showed that the reduction in energy intensity and optimization of the energy consumption structure have a significant effect on the reduction in emissions.

From the perspective of urban agglomerations, Wei et al. [33] collected regional socio-economic governmental reports, including urban planning and carbon emission data from 75 cities in the Yangtze River Delta, Beijing–Tianjin–Hebei, and Pearl River Delta from 2005 to 2020, and utilized the Spatial Durbin Model (SDM) to explore the spatial and temporal evolution of the carbon emissions and the driving factors, which indicated that the carbon emissions demonstrated an upward trend and the spatial distribution was changing dynamically, and the economic, industrialization, and demographic factors were also important in this study. The study shows that carbon emissions are growing and their spatial distribution is changing dynamically, with economic, industrialization, and population factors playing a significant role.

Zhang et al. [34] studied the ecological network of the carbon metabolism of cropland in Nanchang and found that the vertical net carbon flow of the cropland showed an inverted U-shape from 2000 to 2020, and the vertical and horizontal net carbon flow of the cropland was negative in general, which was mainly due to the conversion of cropland to transportation and industrial land, and cropland accounted for the largest proportion of the carbon fluxes. Zhu et al. [35] studied the temporal and spatial coupling relationship between land urbanization and carbon emissions in Zhejiang Province from 1995 to 2015 and found that the carbon emissions and urban expansion rate in the province showed a trend of rising and then decreasing; i.e., with the advancement in urbanization, the scale and structure of the land use should be reasonably regulated to promote a decline in carbon emissions after reaching the peak and realize the low-carbon transition. The study by Ke et al. [36] on Shenzhen shows that, although its carbon emissions declined after 2010 due to energy efficiency optimization, the carbon emissions associated with land use change still increased, especially the high carbon emission intensity of Shenzhen’s transportation and industrial land use, which accounted for 85% of the carbon emissions, and the land use change resulted in a cumulative increase of 85% in the associated carbon emissions.

Wang et al. [37] investigated the combined effects of land use and land cover change on carbon metabolism in Gansu, integrating data from multiple sources and using ecological network analysis (ENA) to study the situation from 1995 to 2020, during which the carbon emissions increased by a factor of 3.18, with transportation and industrial land use being the main causes, and the ENA showed that the competitive relationship between the land use types was dominant and fluctuating, and the mean of the mutuality index was 0.72, and LUCC has a negative impact on carbon metabolism in general.

3. Methods

3.1. Research Area

China is the world’s second-largest economy [38,39,40]. Rapid economic growth has increased carbon emissions, and this increase has become a significant problem for the country’s continued economic progress. Therefore, the four major economic zones (northeast region, eastern region, central region, and western region) divided by the National Bureau of Statistics [41] were selected as the research subjects in this study to observe and reveal the regional evolution trends in carbon metabolism. The study utilized land use data from 31 Chinese provinces (excluding Hong Kong, Macao, and Taiwan) in 2000, 2005, 2010, 2015, and 2020, with an emphasis on carbon input and carbon metabolism (Figure 1).

The eastern area is mainly the coastal area of China and is characterized by flat terrain, low elevation, and high economic level. The region exhibits a high degree of industrialization, and its carbon emissions always remain at a significant level [42]. The central area has moderate economic development, dense population, and relatively high carbon emissions [43]. The western area has a relatively low economic level and a large area, but the population is small [44,45]. It features vast plateaus and a notable rise in carbon emissions. The northeast area exhibits slow upgrade and industrialization, resulting in low carbon emissions [46,47].

3.2. Data Sources

In this research, the land use data of all Chinese provinces for 2000, 2005, 2010, 2015, and 2020 were sourced from the Data Center for Resources and Environmental Sciences, Chinese Academy of Sciences [48]. The maps’ resolution in this paper is 300 m. All the study data were derived from Landsat TM images and generated through remote sensing interpretation. In ArcGIS 10.8, all the data were reclassified into six first-level categories according to the classification system of China’s Multiperiod Remote Sensing Land Use Monitoring Dataset [49,50]: agricultural land, forest land, grassland, water area and wetland, construction land (including urban land, industrial land, and highroad), and bare land.

The study data, such as energy data, household energy consumption, nonagricultural population, total power of agricultural machinery, traffic volume, and stock of pigs and sheep, were derived from China’s provincial statistical yearbooks [51] and China’s Economic and Social Big Data Research Platform [52]. Certain missing parts of data, such as the number of taxis in each province in 2000 and coal consumption in 2020, were substituted with data from adjacent years.

4. Accounting Methods and Model Construction

4.1. Calculation of Carbon Sources and Sinks

This article calculated the main carbon sinks and sources depending on the natural and artificial compartments of carbon metabolism in all study provinces: agricultural land (A), forest land (F), grassland (G), water area and wetland (W), construction land (C), and bare land (B).

Carbon sink coefficient was used to compute carbon sinks $C_i$ in ecological land [29]. The formula is as follows:

$$C_i=S_i\times k_i$$

(1)

where $S_i$ is the area of the i type of land (hm²), and $k_i$ is the carbon sink coefficient (kg/hm²/a) of the type i land (Table S1) [53,54,55].

The carbon sources of agricultural land, urban land, industrial land, and highway were calculated.

(1) Carbon sources of agricultural land mainly refer to direct carbon emissions by agricultural planting and livestock raising. The former mainly comprise emissions from fertilizer use, agricultural machinery, and farmland irrigation. Greenhouse gases, such as CH₄ and N₂O from livestock, were calculated using the emission coefficient published by IPCC. The emissions of CH₄ and N₂O were converted into standard carbon emissions for unified calculation using the conversion coefficients already known (1 ton CH₄ = 6.82 tons CO₂ and 1 ton N₂O = 81.27 tons CO₂). The following is the calculation formula [56]:

$$C_A=C_a+C_p+E=K_{2}F{+K}_{3}M+K_{4}S+E$$

(2)

$$E=\sum E_{{CH}_4}+\sum E_{N_{2}O}=6.28\times \sum N_i\times \left(K_5+K_6\right)\times {10}^{-7}+81.27\times \sum N_i\times K_7\times {10}^{-7}$$

(3)

where $C_A$ represents the carbon emissions from agricultural land (t), $C_a$ represents the carbon emissions from agricultural activities (tons, represented by t), $C_p$ denotes carbon emissions (t) in irrigation and $E$ is livestock carbon emissions (t), $F$ stands for the amount of fertilizer applied (10,000 t), $M$ represents the total power of agricultural machinery (10,000 kW), $S$ is the irrigated area (1000 hm²), $E_{{CH}_4}$ stands for the total emissions of CH₄ (t), $E_{N_{2}O}$ corresponds to the total emissions of N₂O (t), and $N_i$ is the number of the i type of livestock (head). $K_5$, $K_6$, and $K_7$ are the CH₄ emission coefficient of intestinal fermentation, CH₄ emission coefficient of manure treatment, and N₂O emission coefficient (kg/head/a) of type i livestock (Table S2), respectively [57,58,59,60,61,62].

(2) In urban land, the major carbon sources are the direct carbon emissions of residents living and breathing. The calculation formula is as follows [63]:

$$C_U=\sum E_1{K}_i+K_{1}P$$

(4)

where $C_U$ is urban land carbon emissions (t), $E_1$ is the total energy consumption of residents (10,000 tons of standard coal), $K_i$ is the carbon emission coefficient of various energy sources, and $P$ is nonagricultural population (10,000 people).

(3) In this research, the carbon sources of industrial land pertain only to the carbon emissions from direct energy consumption in industry, excluding those from raw material production, such as cement. According to data availability, nine major industrial energy consumption types were selected in this study (Table S3) [57,64]. The following is the calculation formula [56]:

$$C_I=\sum E_i{K}_i$$

(5)

where $C_I$ represents carbon emissions of industrial land (t), and $E_i$ is the total industrial energy consumption (10,000 tons of standard coal).

(4) As for the highway carbon sources, the study merely considered domestic carbon emissions, which mainly come from private cars, buses, taxis, and motorcycles. Meanwhile, the top-down approach was utilized to determine the transportation carbon emissions of buses. In particular, all types of energy were converted into standard coal consumption and then multiplied by the carbon emission coefficient. The following is the formula used for computation [56,65]:

$$C_H=K_8{M}_p+C_b+K_9{M}_t+K_{10}{M}_m$$

(6)

$$C_b=E{C}_i\times F_i\times K_i$$

(7)

where $C_H$ represents highway carbon emissions (t); $M_p$, $M_t$, and $M_m$ are the mileage (km) of private vehicles, taxis, and motorcycles, respectively. Calculations indicate that the average yearly driving of private cars, taxis, and motorcycles in China is 15,000, 120,000, and 4000 km/vehicle, respectively. Moreover, $C_b$ is the carbon emissions of buses (t), $E{C}_i$ is the physical amount of the i energy terminal consumption in the transportation industry (10,000 tons of standard coal), and $F_i$ is the conversion coefficient of standard coal.

4.2. Carbon Flow Accounting

A horizontal carbon flow model can be constructed according to the net carbon source or carbon exchange rate (defined as carbon metabolic density) per unit area of each land type and the amount of land converted between different land types. The calculation formula is [29,56]:

$$\Delta \mathrm{W}=W_i-W_j={\displaystyle \frac{V_i}{S_i}}-{\displaystyle \frac{V_j}{S_j}}$$

(8)

$$f_{ij}=\Delta W\times \Delta S$$

(9)

where $f_{ij}$ indicates the carbon flows from j to i; $\Delta W$ is the difference in carbon metabolic density of various land use types (kg/hm²); $W_i$ and $W_j$ are the net carbon flow density (kg/hm²) of i and j, respectively; $V_i$ and $V_j$ are the net carbon flows (kg) of i and j, respectively; $S_i$ and $S_j$ are the area (hm²) of i and j, respectively; and $\Delta S$ is land use transfer area (hm²). The horizontal carbon flow calculation results can determine the direction of carbon transfer and the type of land use that dominates carbon flows. $f_{ij}$ > 0 indicates a positive horizontal carbon flow represented by a reduction in carbon emissions or an increase in carbon sinks, contributing to the balance of carbon metabolism. $f_{ij}$ < 0 indicates a negative horizontal carbon flow manifested as an increase in carbon emissions or a decrease in carbon sinks, aggravating the disorder of carbon metabolism.

4.3. Gravity Center Changes with Carbon Flows

The spatial evolution of the center of gravity of the carbon flow of six types of land can be evaluated in period t using the barycentric coordinates $(x_i,y_i)$ of each province represented by the longitude ($X_t$) and latitude ($Y_t$). The following is the formula used for computation [29]:

$${\mathrm{X}}_{\mathrm{t}}={\displaystyle \frac{\sum f_{ij}{x}_i}{\sum f_{ij}}},Y_t={\displaystyle \frac{\sum f_{ij}{y}_i}{\sum f_{ij}}}$$

(10)

SDE is a spatial statistical approach that uses the center, major axis, minor axis, and azimuth angle as fundamental parameters to quantitatively describe the general characteristics of the research objects’ spatiotemporal evolution. The calculation formula is [66]:

$$\mathrm{M}\left(\overline{\mathrm{X}},\overline{\mathrm{Y}}\right)=\left|{\displaystyle \frac{\sum _{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{w}}_{\mathrm{i}}{\mathrm{x}}_{\mathrm{i}}}{\sum _{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{w}}_{\mathrm{i}}}},{\displaystyle \frac{\sum _{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{w}}_{\mathrm{i}}{\mathrm{y}}_{\mathrm{i}}}{\sum _{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{w}}_{\mathrm{i}}}}\right|$$

(11)

$${SDE}_x=\sqrt{{\displaystyle \frac{\sum _{i=1}^n{\left(x_i-\overline{x}\right)}^2}{n}}},{SDE}_y=\sqrt{{\displaystyle \frac{\sum _{i=1}^n{\left(y_i-\overline{y}\right)}^2}{n}}}$$

(12)

$$\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{\theta }={\displaystyle \frac{\left(\sum _{\mathrm{i}=1}^{\mathrm{n}}{{\mathrm{w}}_{\mathrm{i}}}^2{{\mathrm{x}}_{\mathrm{i}}}^2-\sum _{\mathrm{i}=1}^{\mathrm{n}}{{\mathrm{w}}_{\mathrm{i}}}^2{{\mathrm{y}}_{\mathrm{i}}}^2\right)+\sqrt{\left(\sum _{\mathrm{i}=1}^{\mathrm{n}}{{\mathrm{w}}_{\mathrm{i}}}^2{{\mathrm{x}}_{\mathrm{i}}}^2-\sum _{\mathrm{i}=1}^{\mathrm{n}}{{\mathrm{w}}_{\mathrm{i}}}^2{{\mathrm{y}}_{\mathrm{i}}}^2\right)+4\left(\sum _{\mathrm{i}=1}^{\mathrm{n}}{{\mathrm{w}}_{\mathrm{i}}}^2{{\mathrm{x}}_{\mathrm{i}}}^2{{\mathrm{y}}_{\mathrm{i}}}^2\right)}}{2\sum _{\mathrm{i}=1}^{\mathrm{n}}{{\mathrm{w}}_{\mathrm{i}}}^2{{\mathrm{x}}_{\mathrm{i}}}^{\prime }{{\mathrm{y}}_{\mathrm{i}}}^{\prime }}}$$

(13)

where $M\left(\overline{X},\overline{Y}\right)$ is the center of gravity coordinate; ${SDE}_x$ and ${SDE}_y$ represent the major axis and minor axis of the SDE, respectively; $\theta$ is the azimuth of the ellipse; $x_i$ and $y_i$ are the spatial coordinates of the element i; and $\overline{x}$ and $\overline{\mathrm{y}}$ are the coordinate deviations of the mean center of gravity.

4.4. ENA Method and Ecological Relationship Judgment

Ecological network analysis (ENA) was used to analyze the flow of matter and energy in an ecosystem, whose basic units are divided into compartments and pathways [67]. $T_i$ is equal to all flows out of or into $i$ plus or minus $x_i$, where x is the change in carbon stock of land use type $i$. If carbon emissions are reduced or sinks are increased, $x_i>0$, $T_i$ is equal to all flows out of $i$ plus $x_i$; if carbon emissions increase or carbon sinks decrease, $x_i<0$, $T_i$ equals all flows into $i$ minus $x_i$. The carbon flux of each land use participating in carbon metabolism can be determined by applying the flux analysis method. Carbon flux is defined as the difference between each land use’s carbon flux in all horizontal directions and state variables. The state variable refers to the difference between the quantity of carbon inflow and outflow of each form of land use. After the carbon flux is obtained, it is standardized using ENA to generate the horizontal carbon flow matrix D, which can describe the direct relationship between two carbon metabolism compartments brought about by horizontal carbon flow. The complete utility matrix U is calculated based on the horizontal carbon flow matrix D to comprehensively consider the direct and indirect interactions between the two carbon metabolism compartments. The relationships between all nodes in the matrix are also examined to obtain the comprehensive interaction between the two carbon metabolism compartments under the overall ecological network effect. The calculation formula is

$$d_{ij}=\left(f_{ij}-f_{ji}\right)/T_i$$

(14)

$$D_{ij}=\left[\begin{array}{cc}{\displaystyle \frac{f_{jj}-f_{jj}}{T_j}}& {\displaystyle \frac{f_{ji}-f_{ij}}{T_j}}\\ {\displaystyle \frac{f_{ij}-f_{ji}}{T_i}}& {\displaystyle \frac{f_{ii}-f_{ii}}{T_i}}\end{array}\right]$$

(15)

$$U=\left(U_{\mathrm{i}\mathrm{j}}\right)={\mathrm{D}}^0+{\mathrm{D}}^1+{\mathrm{D}}^2+\cdots +{\mathrm{D}}^{\mathrm{n}}={\left(\mathrm{I}-\mathrm{D}\right)}^{-1}$$

(16)

where the unit matrix of $I$ is the self-feedback effect of each land use type in carbon flow transfer, $U_{ij}$ is an element of the matrix $U$, $n$ denotes the type of land use, and $T_i$ is the carbon flux of land $i$ [18].

On the basis of the complete utility matrix $U$, we can determine the ecological relationship between the two types of land use. In theory, the ecological network has nine types of ecological relationships (

Table 1. The ecological relationships between components of the network.

Matrix Notation	Positive (+)	Neutral (0)	Negative (−)
Positive (+)	Mutualism (+,+)	Commensalism (+,0)	Exploitation (+,−)
Neutral (0)	Commensalism host (0,+)	Neutralism (0,0)	Amensalism (0,−)
Negative (−)	Control (−,+)	Amensal host (−,0)	Competition (−,−)

Note: values in brackets represent the value in the first column of the table in matrix U(j) followed by the value in the first row of the table in matrix U(i) for flow f_ij.

). However, only four are common: competition, mutualism, and exploitation and control. The competition relationship signifies that the rivalry between the two land use types diminishes the utility for both parties. Mutualism symbiosis suggests that, through interdependent changes, the two land use types enhance their respective utilities. The relationship characterized by exploitation and control implies that one land use type benefits at the expense of the other, where one’s utility increases while the other’s decreases.

5. Results

5.1. Time and Space Changes in Land Use

Based on the analysis of land use data from 2000 to 2020, agricultural land in eastern China has progressively been converted into construction land. Notably, economically developed municipalities and provinces such as Beijing, Tianjin, Jiangsu, and Shanghai have experienced continuous outward expansion of their urban areas (Figure 2a). In western regions, particularly Xinjiang, the area of water area and wetland has gradually increased, while bare land has decreased by 2.66 × 10⁶ hm².

Figure 2b illustrates that the conversion between agricultural land and grassland to other land types was the most significant in China during the period from 2000 to 2020. Specifically, agricultural land is mainly transformed into forest land, grassland, and construction land. From 2005 to 2010, the transfer area of agricultural land to forest land reached 10.23 × 10⁶ hm². Grassland is mainly converted to agricultural land, forest land, and bare land. From 2005 to 2010, the transfer area of grassland to forest land reached 13.42 × 10⁶ hm². Additionally, the transfer area between grassland and bare land is generally high, with an average of about 6.60 × 10⁶ hm² of bare land being reclaimed into grassland every year, indicating China’s emphasis on ecological protection and restoration and improving land use efficiency. Over the study period, forest land was primarily converted to agricultural land. From 2010 to 2015, 4.61 × 10⁶ hm² of forest land was converted into construction land, accounting for 33.58% of the total transfer area. During 2000–2005, the amounts of water area and wetland transferred to grassland and bare land were 5.53 × 10⁶ hm² and 2.16 × 10⁶ hm², respectively. Between 2005 and 2020, there was minimal transition between these land use types. From 2010 to 2015, 4.19 × 10⁶ hm² and 4.37 × 10⁶ hm² of construction land were transferred to agricultural land and forest land, respectively.

5.2. Changes in Carbon Emissions and Carbon Absorption

During 2000–2020, forest land (90.04%) accounted for the largest proportion of carbon uptake. This finding indicates that it was the most significant land for the carbon sink function. Forest land was followed by water area and wetland (5.25%) and grassland (3.39%) (Figure 3). In the northeast area, the figure was 0.24 × 10⁸ t. In the three provinces in the area, Heilongjiang (0.15 × 10⁸ t) ranked second nationwide in carbon absorption. However, Jilin and Liaoning exhibited relatively high levels. The situation differed in the central area, where the carbon absorption was highly concentrated, totaling 0.31 × 10⁸ t. Among all the central provinces, Hunan, Hubei, and Jiangxi absorbed high amounts of carbon. Compared with the water area and wetland in other provinces, that in Shanxi exhibited the least absorption. However, its grassland took in more carbon than the grassland in other provinces. On the contrary, the western area exhibited a surprising absorbing capability, accounting for 53.94% of the national carbon absorption. In particular, the area of forest land in Yunnan accounted for the highest proportion. Thus, the province’s carbon absorption capacity, reaching 0.18 × 10⁸ t, ranked first in the country. Two other western provinces, Inner Mongolia and Sichuan, also had substantial absorption. However, Ningxia had the least carbon absorption, comprising merely 0.15% of the total in the western area. In the eastern area, the carbon absorption was 0.26 × 10⁸ t. Among the provinces in the area, Guangdong, Fujian, and Zhejiang exhibited relatively high carbon absorption, whereas Shandong, Jiangsu, Shanghai, Tianjin, and Beijing absorbed a relatively small amount of carbon.

Industrial land (53.14%) and urban land (39.38%) are the major contributors to total carbon emissions in China from 2000 to 2020 (Figure 3). In particular, the carbon emissions in the northeast increased from 2.85 × 10⁸ t in 2000 to 8.37 × 10⁸ t in 2020. As one of the first centers of heavy industry in China [68], Liaoning (50%) contributed to the largest share of carbon emissions among the three provinces. However, Jilin (20%) and Heilongjiang (30%) contributed a small amount of carbon emissions. In the central area, the emissions were highly collected, with Henan and Shanxi exhibiting high-level emissions at 3.11 × 10⁸ t and 2.75 × 10⁸ t, respectively. In Henan, the industrial land emitted almost the same quantity of carbon as the urban land. In Shanxi, the industrial land contributed 49.05% of the province’s total carbon emissions. By contrast, Jiangxi released the lowest carbon emissions, comprising 8% of the total in the central area. The carbon emissions in the western area were increasing yearly. From 2000 to 2005, the carbon emissions in the western area were less than those in the central area. However, they were higher than those in the central area from 2010 to 2020. In the entire area, industrial land yielded 50.95% of the overall emissions. This finding shows that the proportion of heavy chemical industry in the western area was strengthened. In Inner Mongolia, Shaanxi, and Xinjiang, the carbon emissions from industrial land use exceeded 1 × 10⁸ t from 2015 to 2020. From 2000 to 2020, Sichuan consistently experienced high carbon emissions from urban land, reaching 1.46 × 10⁸ t in 2020. By contrast, Qinghai and Ningxia yielded relatively low carbon emissions. The latter province exhibited similar quantities of emissions from urban and agricultural land emissions. Tibet had no reports of carbon emissions from industrial land because of the lack of statistics on energy consumption. The eastern area released the highest carbon emissions out of the four economic areas, with a total of 36.25 × 10⁸ t. Nonetheless, the eastern area’s provinces varied significantly in terms of their carbon emissions. For instance, the carbon emissions of five provinces, including Hebei, Jiangsu, Zhejiang, Shandong, and Guangdong, exceeded 1 × 10⁸ t. By contrast, Hainan emitted the least carbon, accounting for merely 1.14% of the total in the region.

5.3. Carbon Flow Time and Space Distribution

The positive transfer of carbon flow (positive carbon flow) implies an increase in carbon sequestration or a decrease in carbon emissions. By contrast, the negative transfer of carbon flow (negative carbon flow) means increased carbon emissions or decreased carbon sequestration. Table S4 shows the change in the carbon flow paths between different lands during the study period. The conclusion is that a negative net carbon flow was found between 2000 and 2020, suggesting that land use changes negatively affect carbon metabolism, increase carbon emissions, and unbalance the carbon cycle. The carbon flows of dominant pathways during 2005–2010 and 2010–2015 were more concentrated than those during other periods. Throughout these two periods, the carbon flow between agricultural land and construction land comprised the highest proportion of the total carbon flow, ranging from 76.79% to 78.71%. However, this proportion decreased from 66.61% to 71.86% of the total carbon flow during 2000–2005 and 2015–2020. Additionally, the main pathways during 2005–2010 and 2010–2015 exhibited similar carbon flow amounts and directions, with the major positive carbon flow pathways being C–A (64.93–72.29%) and C–W (9.54–21.92%) and the major negative carbon flow pathways being A–C (78.45–79.64%) and G–C (10.01–16.69%). However, the major negative carbon flow pathways during 2000–2005 and 2015–2020 remained the same as those during the previous periods, but differences were found in the major positive carbon flow pathways. For example, the positive carbon flow in A–F comprised 9.79% of the whole carbon flow during 2000–2005, whereas the C–G pathway accounted for 19.35% of the whole carbon flow during 2015–2020. This finding suggests that the country’s policy of converting agriculture back to grasslands and forests has produced the desired outcomes.

Between 2000 and 2015, the centers of carbon flow for grassland and bare land were predominantly located in Gansu and Inner Mongolia in western China. However, between 2015 and 2020, there was a significant shift in the center of gravity of the carbon flows in both regions (Figure 4a). Specifically, the center of gravity for grassland gradually moved southeastward into Shaanxi (Figure 4d), while that for bare land shifted from a southeastern orientation (7°) during 2010–2015 to a northeastern orientation (10°) during 2015–2020 (Figure 4g). These changes may be attributed to the enhanced rational planning of land use types in western China [69]. Additionally, the center of gravity for carbon flow from construction land continuously shifted northwestward (37°) toward Gansu between 2015 and 2020 (Figure 4f). Other types of land, including agricultural land, forest land, and water area and wetland, remained primarily concentrated in the central regions of Henan and Hubei (Figure 4b,c,e). This distribution is likely associated with the rapid economic development and high carbon emissions in the eastern and central regions [70].

The SDE distribution features of carbon flow in China’s four economic areas are shown in Figure 5. From 2000 to 2020, notable changes were observed in the spatial distribution patterns, areas, directions, and diffusion trends of the carbon flow centers in the four economic regions (Figure 5a). From a spatial perspective, the carbon flow centers in the northeast and east areas shifted toward the northeast direction (Figure 5b,e), moving approximately 0.76 × 10⁵ and 0.78 × 10⁵ km along the α direction, respectively. The carbon flow centers in the central and western areas shifted toward the southeast direction (Figure 5c,d), moving approximately 1.22 × 10⁵ and 2.79 × 10⁵ km along the α direction, respectively. This phenomenon may be attributed to the increasing carbon flow in the western area from 2005 to 2015. The major and minor axes of the SDE in the western area were closer than those in the three other economic regions. This finding indicates a highly concentrated carbon flow distribution in the northeast, central, and eastern areas. Furthermore, the flatness of the carbon flow’s SDE decreased, for example, from 0.87 to 0.80 in the northeast area and from 0.65 to 0.54 in the central area. This result suggests that the spatial dispersion of carbon flow in China from 2000 to 2020 tended to decrease.

5.4. Ecological Relationships

The ecological relationships of China’s land use types in the carbon metabolism system mainly include mutualism, exploitation and control, and competition (Figure 6). As illustrated in the table, the ecological relationship between 2000 and 2010 was predominantly characterized by a restrictive dynamic. In contrast, the period from 2010 to 2015 was marked by a mutualistic dynamic, with the predatory dynamic being virtually absent.

The main ecological relationships of the various land types from 2000 to 2020 were exploitation and control (39.17%), competition (36.67%), and mutualism (24.16%) (

Table 2. Distribution of ecological relationships.

Land Use Types	A	F	G	W	C	B
Mutualism relationships	2	9	5	1	8	4
Exploitation and control relationships	7	10	7	10	5	8
Competition relationships	11	1	8	9	7	8

). This finding indicates a significant conflict regarding the ecological relationships during land use change and competition for carbon storage among the land use types. The ecological relationship of exploitation and control was relatively distributed in forest land and water area and wetland, accounting for 42.55% of the whole land area. This finding indicates that the degradation of these land types could significantly impact the carbon metabolism in each province. Competition mainly existed in those land use types with carbon emissions and low carbon sinks. In particular, agricultural land, grassland, water area and wetland, and construction land contributed 25.00%, 18.18%, 20.45%, and 18.18% of the competition relationship, respectively. This result shows that fierce competition existed between carbon emissions, low carbon sink land, and other land use types, affecting the balance in carbon metabolism. Mutualism was mainly distributed in forest land and bare land, which comprised 58.62% of all the types of land.

Within the entire country, the ecological relationship area gradually increased during 2000–2005, 2005–2010, and 2010–2015, signifying an unstable changing trend in land use in various provinces before 2015. This type of land area gradually decreased during 2015–2020, indicating a relatively steady state of land use (Figure 7). In terms of space, the ecological relationship of exploitation and control is mainly concentrated in Gansu, Ningxia, and Shanxi provinces and the western regions of Heilongjiang and Jilin provinces. From 2000 to 2005, the ecological competition in Heilongjiang Province was concentrated in the west and then gradually replaced by the exploitation and control relationship. The southern part of Ningxia was dominated by a mutualism relationship during 2000–2005, but the exploitation and control ecological relationship expanded rapidly during 2005–2020 and extended to Gansu, Shaanxi, and Shanxi provinces. The ecological competition between Shandong, Jiangsu, and Tibet provinces has been increasing and concentrated in the north of Jiangsu and the west of Tibet during 2010–2015. The change in the ecological relationship between the two provinces will have an impact on their system effectiveness.

Throughout the study periods, the three ecological relationships of mutualism, exploitation and control, and competition were distributed uniformly but sporadically in the northeast area. Mutualism and competition exhibited a rising trend during 2000–2015, followed by a subsequent decrease during 2015–2020 (Figure 6 and Figure 7). In the central area, the ecological relationships were mainly characterized by competition and exploitation and control. They were distributed sporadically. Among all the central provinces, Anhui was an exception; it did not exhibit conspicuous changes in ecological relationships. On the contrary, exploitation and control and competition continuously increased from 2000 to 2015 in the other provinces. Then, they gradually decreased during the subsequent 5 years (Figure 7). The ecological relationships appeared to be scattered in the western area. Among all the western provinces, Xizang, Yunnan, Guizhou, and Guangxi were dominated by competition. Xinjiang, Qinghai, Gansu, Inner Mongolia, and Ningxia were characterized by exploitation and control. The western part of Sichuan was dominated by mutualism, whereas the eastern part was dominated by exploitation and control and competition (Figure 7). In the western directly administered municipality of Chongqing, the ecological relationships shifted from being characterized by exploitation and control and competition to the dominance of mutualism during 2010–2015. Similar to the central area, the eastern area was characterized by competition and exploitation and control. These relationships also exhibited a sporadic distribution.

6. Discussion

Researchers have experimented with several approaches to study urban carbon metabolism and help cities with the difficult challenge of cutting carbon emissions. Compared with analyzing carbon transfers caused by land transformation, the urban carbon metabolism method can provide additional information and construct a comprehensive framework to follow urban carbon flows [71,72].

This study considered 31 Chinese provinces and analyzed the changes in their carbon emissions. The findings demonstrate that, from 2000 to 2020, the carbon emissions of China’s provinces were linearly correlated with economic growth. However, among the four economic zones, the eastern area ranked first in carbon emissions. The northeast and western areas were increasingly becoming carbon-exposed every year throughout this fast economic expansion. Thus, they have become significant areas of concern for national carbon emissions. In recent years, carbon emissions associated with land use have become a research hotspot. However, previous studies mostly focused on a single province [65,73,74] or developed areas, such as the Beijing–Tianjin–Hebei and Yangtze River Delta urban agglomerations [75,76,77]. For instance, the research by Lv et al. [78] revealed that most of the regions with significant carbon emissions are found in Shanghai and the surrounding cities. Carbon emissions typically exhibit a declining tendency from the eastern to the western areas [79,80]. However, limited research has been conducted on China’s western region and undeveloped provinces. Only Zhang et al. [81] and Cai et al. [82] organized relevant studies and found that, during their selected period, the carbon emissions were effectively reduced in provinces such as Qinghai, Inner Mongolia, Yunnan, Guangxi, and Heilongjiang, whereas the carbon release of Sichuan increased by 43.14%. Different from the case in the abovementioned research, the data from 31 Chinese provinces were classified in this study into four economic regions for further analysis. The research results are highly comprehensive and can provide an in-depth understanding from a macro perspective.

In this study, a carbon flow model was also constructed in the horizontal direction by calculating carbon sources and sinks in the vertical dimension to analyze the direction and changes in carbon metabolism resulting from land use changes in every Chinese province from 2000 to 2020. The results indicate that negative carbon flows can be primarily attributed to the conversions between different land types, such as from agricultural land to construction land, grassland to construction land, and from water area and wetland to construction land. The increasing emissions from construction land contributed to the imbalance in carbon metabolism in each province. Some researchers employed ENA to investigate the carbon metabolism of agricultural land; they discovered negative carbon flows predominantly because of the transfer of agricultural land to transportation and industrial land [34,83] This research classified the land use types in the carbon metabolism of the Chinese provinces based on a first-level classification [49]. This research also consolidated agricultural, highway, and industrial lands into construction land, thereby making this land type generate substantial negative carbon flows.

Finally, the ecological network utility analysis approach was used to assess the ecological relationships among various compartments and their spatial distribution changes resulting from carbon flows across the 31 provinces of China. The findings reveal that the ecological relationships in each province were mainly characterized by exploitation and control. Moreover, the degradation of forest land and water area and wetland had a significant impact on carbon metabolism. An ENA conducted by some researchers indicates that, in Beijing, exploitation and control serves as the primary ecological relationship that promotes economic growth; this conclusion is consistent with the results of this paper [84]. By contrast, other researchers employing the same method to investigate the ecological relationships associated with land use type changes in Zhejiang have identified the competitive relationship between water area and wetland and forest land [85]. This finding diverges from the outcomes presented in this paper. The inconsistency can be attributed to the conversion of natural land, such as water area and wetland and forest land, into agricultural lands in the context of implementing agricultural land protection policies. This transformation has consequently exacerbated local carbon imbalances.

7. Conclusions

Based on the carbon accounting methods, this study focuses on 31 provinces in China as subjects. This study uses high-resolution (300 m) land use data to present the overall situation and spatial distribution of carbon metabolism in four economic zones during 2000–2005, 2005–2010, 2010–2015, and 2015–2020. It also reveals the spatiotemporal evolution of carbon flow related to land use changes through ENA to investigate the intricate ecological relationships among different compartments of the carbon metabolism system. The conclusions are as follows:

From 2000 to 2020, the carbon emissions of all the Chinese provinces were greater than the carbon absorption. They continued to increase to 84.63 × 10⁸ t. The major contributors are emissions from industrial land (53.14%) and urban land (39.38%). Hence, those provinces with high carbon emissions, including Jiangsu, Shandong, Guangdong, Hebei, Henan, Liaoning, and Inner Mongolia, require strict management. Furthermore, the carbon flow between construction land and agricultural land is the largest, with the center of gravity of carbon flow in construction land shifting significantly to the northwest (37.14°) during 2015–2020. Thus, the west’s logical distribution of land use types must be strengthened. The ecological relationship of the carbon metabolism of various land types in China between 2000 and 2020 is dominated by the relationship of exploitation and control. In particular, the degradation of forest land and water area and wetland severely disrupts the carbon balance of ecosystems, highlighting the urgent need to protect these ecological assets.

There are also certain limitations in this study. First, the data from adjacent years had to be used as substitutes because of the lack of statistics on the energy consumption of a small portion of households and industries. This approach may lessen the accuracy of carbon emission calculations. Second, this study classified the main subjects of land use related to carbon metabolism in Chinese provinces only at a first-level classification, resulting in insufficient precision in identifying ecological relationships. Therefore, future efforts should be further categorized to recognize ecological relationships accurately and offer more assistance for strategic planning.