2.1 Study area
CZT urban agglomeration is located in the central and eastern part of Hunan Province, including Changsha, Zhuzhou and Xiangtan, with a total area of 28,000 km2 (Fig. 1). It is located in the hilly valley, with high terrain in the east and west, and low in the middle. The landform types in the region are diverse, with basins and hills interlaced, towns and villages interlaced, and Xiangjiang River passing through, forming a good ecological background. With a solid industrial base, complete infrastructure and complete industrial chain, it is one of the important manufacturing bases in China and has strong strength and advantages in automobile, electronics, chemical industry and other industries. In 2020, CZT urban agglomeration has a GDP of CNY 1.76 trillion (about 255.3 billion US dollars), 42% of Hunan Province. The population is 16.67 million, 25% of Hunan Province.
CZT urban agglomeration is the core of Hunan's economic development and an important part of the Triangle of Central China. As an experimental zone for reforming comprehensive facilities for the construction of resource-saving and environment-friendly society in China, with the national "carbon peak and carbon neutral" goal, CZT urban agglomeration should further collaborate to manage the environment based on the basis of common ecological environment of the three prefecture-level cities. This could promote the harmonious and low-carbon sustainable development of nature, economy and society.
2.3 Methods
This study is divided into three parts (Fig. 2): 1) According to the main functions of land use types, the study area was divided into production space, living space and ecological space. 2) Carbon flows were calculated according to land use change and carbon metabolism density change, and the carbon metabolism of PLE space in the study area was analyzed by ecological utility network analysis. 3) Using the PLUS model, the influencing factors of urban carbon metabolism were explored from the perspective of land use change.
2.3.1 Determination of PLE Space
The PLE space was classified based on the main function of the land (Fig. 3). This study was guided by the multifunctionality of production, living and ecology of land (Xie et al. 2021), and combined with the carbon emission or sequestration attributes of PLE space and data accessibility, the CZT urban agglomeration was divided into production space, living space and ecological space. Production space was the space used to produce materials needed for human activities, divided into agricultural production space and industrial production space, included cultivated land (carbon emission and carbon sequestration) and industrial land (carbon emission). Living space was the space covered wholly or partially by buildings or structures that could provide cultural recreation for human beings, had a secondary category that was residential living space, included urban land and rural settlements (both carbon emission). Ecological space was the space that was relatively little used by humans and could directly or indirectly improve the ecological environment, divided into forest ecological space, grass ecological space, water ecological space and other ecological space, included forest land, grassland, watershed and unused land (all for carbon sequestration).
2.3.2 Calculation Methods and Model Construction
(1) Carbon Sequestration
Carbon sequestration mainly came from ecological space, which was calculated by coefficient method, the calculation formula was as follows (Feng et al. 2022):
$${S}_{e}=\sum _{i=1}^{n}{A}_{i}\times {k}_{i}$$
1
Where Se was the ecological space carbon sequestration; n was the number of land use types in ecological space; Ai was the area of the ith land use type; ki was the carbon sequestration coefficient of the ith land use type, and watershed was further divided into river, lake and reservoir (Table 2).
Table 2
Carbon sequestration coefficients of different land use types / (kg C / m2 / a)
Land use type | coefficient | Source | Land use type | coefficient | Source |
Cultivated land | 0.0007 | (Zhang et al. 2016b) | River | 0.0250 | (Walsh 1991) |
Forest land | 0.0657 | (Fang et al. 2007) | Lake and Reservoir | 0.0390 | (Meybeck 1993) |
Grassland | 0.0596 | (Zhao 2012) | Unused Land | 0.0005 | (Peng et al. 2016) |
(2) Carbon Emission
Carbon emission of cultivated production space was mainly from agricultural production activities and carbon dioxide emission from livestock respiration. Agricultural production activities included agricultural fertilizer, agricultural machinery use and agricultural irrigation process. According to the current status of livestock rearing in the study area, pig, cattle, sheep and poultry were considered. The calculation formulas were as follows (Xia et al. 2018a):
\({E}_{cu}={E}_{a}+{E}_{l}\) | (2) |
\({E}_{a}={G}_{f}\times A+{A}_{crop}\times B+{W}_{m}\times C+{A}_{i}\times D\) | (3) |
\({E}_{lr}=\sum _{i=1}^{x}{F}_{i}\times {b}_{i}\) | (4) |
Where Ecu was the carbon emission from cultivated land; Ea was the carbon emission from agricultural production activities; Elr was the carbon emission from livestock respiration; Gf was the fertilizer application; Acrop was the crop sowing area; Wm was the total power of agricultural machinery; Ai was the irrigated area; A, B, C and D were carbon emission conversion coefficients; x was the livestock type; Fi was the number of livestock kept of the ith species; bi was the carbon emission coefficient of livestock respiration of the ith species (Table 3).
Carbon emission of industrial production space was mainly from energy activities and traffic operations, the calculation formulas were as follows (Feng et al. 2022):
\({E}_{in}={E}_{p}+{E}_{t}\) | (5) |
\({E}_{p}=\sum _{i=1}^{z}{C}_{i}\times {Q}_{i}\) | (6) |
\({E}_{t}={M}_{p}\times {K}_{p}+{M}_{b}\times {K}_{b}+{M}_{m}\times {K}_{m}\) | (7) |
Where Ein was the carbon emission from industrial production space; Ep was the carbon emission from energy activities; Et was the carbon emission from traffic operations; z was the energy consumption type; Ci was the ith energy consumption; Qi was the ith energy carbon emission coefficient; Mc, Mb, Mm were mileage of private cars, buses and motorcycles respectively; Kc, Kb, Km were carbon emission coefficients for mileage of private cars, buses and motorcycles (Table 3). Research showed that each car in China traveled an average of 15,000 km per year, and each motorcycle traveled an average of 4,000 km per year (Jia et al. 2010).
Carbon emission of residential living space was mainly from the human respiration and electricity consumption. The calculation formulas were as follows (Du et al. 2021):
\({E}_{r}={E}_{h}+{E}_{e}\) | (8) |
\({E}_{h}=\text{P}\times \text{H}\) | (9) |
\({E}_{e}=\text{V}\times \text{R}\) | (10) |
Where Er was the carbon emission from residential living space; Eh was the carbon emission from human respiration; Ee was the carbon emission from electricity consumption; P was the resident population in living space; H was the carbon emission coefficient of human respiration; R was the carbon emission coefficient for electricity consumption (Table 3).
Table 3
Carbon emission coefficients of PLE space
Type of space | Item | Coefficient | Unit | Source |
Cultivated production space | A | 857.54 | kg C / t | (West and Marland 2002) |
B | 16.47 | kg C / hm2 |
C | 0.18 | kg C / kW |
D | 266.48 | kg C / hm2 |
Pig | 301.125 | kg C / head / a | (Du et al. 2021) |
Cattle | 2920 | kg C / head / a |
Sheep | 37.25 | kg C / head / a |
Poultry | 12.775 | kg C / head / a |
Industrial production space | Raw Coal | 0.7559 | kg C / t | (Zhang et al. 2016b) |
Coal Washed | 0.7559 | kg C / t |
Coke | 0.8550 | kg C / t |
Coke Oven Gas | 0.3548 | kg C / m3 |
Nature Gas | 0.4483 | kg C / m3 |
Gasoline | 0.5538 | kg C / t |
Kerosene | 0.5714 | kg C / t |
Diesel Oil | 0.5921 | kg C / t |
Combustion Oil | 0.6185 | kg C / t |
Liquefied Petroleum Gas | 0.5042 | kg C / t |
Other petroleum products | 0.5857 | kg C / t |
Private car | 22.3 | kg C / 100km | (Xia et al. 2018a) |
Bus | 88.1 | kg C / 100km |
Motorcycle | 6.7 | kg C / 100km |
Residential living space | Human respiration | 79 | kg C / person / a | (Zhang et al. 2016b) |
Electricity consumption | 0.1223 | kg C / kWžh | (Du et al. 2021) |
2.3.3 Carbon Flow Calculation of PLE Space
The analysis of PLE space carbon metabolism allowed to probe into the influence of human activities and land use changes on carbon metabolism. In order to research the evolution of PLE space carbon metabolism capacity in CZT urban agglomeration more intuitively, the carbon emission or sequestration per unit area of various land use types in unit time was defined as carbon metabolism density (W). The product of the change of carbon metabolic density and land use change area at a certain stage was defined as carbon flow (f). The calculation formulas were as follows (Xia et al. 2018b):
where i and j were different PLE space land use types; Wj and Wi were the net carbon flow density of PLE space land use types j and i; Vj and Vi were the net carbon flows of PLE space land use types j and i; Sj and Si were the areas of PLE space land use types j and i; fij was the carbon flow from carbon flow from land use type j to i; ΔW was the carbon metabolic density difference; ΔS was the area transferred from PLE space land use type j to i. If ΔW > 0, it meant that this was a positive carbon flow, which was realized as a decrease of carbon emission or an increase of carbon sequestration, which contributed to the carbon metabolic balance of the system in PLE space. If ΔW < 0, it meant that this was a negative carbon flow, which was realized as an increase of carbon emission or a decrease of carbon sequestration, which threatened the carbon metabolic balance of the system in PLE space.
2.3.4 Ecological Utility Network Analysis Method and Ecological Relationship Determination of PLE Space
Ecological utility network analysis was a modeling method used to analyze the flow of matter and energy among compartments in an ecosystem. In the process of researching metabolism through the ENA method, various indicators had been developed to characterize the ecological features of subjects, pathways, and networks as a whole, such as fluxes, synergistic indices and trophic levels (Chen and Chen 2012; Xia et al. 2016; Zhang et al. 2014), which provided an effective method for quantitative dissection of system structure and function. In this study, the flux analysis and utility analysis methods of ENA were used to research the interaction of carbon metabolism in PLE space and the influence of the evolution of PLE space on the ecosystem. The effective utilization matrix (D) could reflect the direct interaction among carbon flows, and the overall utility matrix (U) could be obtained based on the effective utilization matrix. The overall utility matrix could be used to obtain the integral mode of action among the compartments in the ecological network, and then explained the integrated action relationship of two compartments under the overall utility network. The calculation formulas were as follows (Fath and Patten 1998):
\({d}_{ij}=\frac{{f}_{ij}-{f}_{ji}}{{T}_{i}}\) | (13) |
\({D}_{ij}=\left[\begin{array}{cc}\frac{{f}_{jj}-{f}_{jj}}{{T}_{j}}& \frac{{f}_{ji}-{f}_{ij}}{{T}_{j}}\\ \frac{{f}_{ij}-{f}_{ji}}{{T}_{i}}& \frac{{f}_{ii}-{f}_{ii}}{{T}_{i}}\end{array}\right]\) | (14) |
\(U=\left({u}_{ij}\right)=\sum _{t}^{e}{D}^{t}={(I-D)}^{-1}\) | (15) |
where dij was the effective utilization rate of carbon flow among compartments; fij was the carbon flow from land use type j to i; Ti was the carbon flux of land use type i; uij was an element of matrix U; e was the number of PLE space land use types; t was the number of compartments for carbon flow exchange; I was the unit matrix, which indicated the self-feedback of each type of PLE space land use types in the process of carbon flow exchange.
Theoretically, there were nine ecological relationships in ecological networks (Table 4), but only four are common: exploitation, control, mutualism and competition. The essence of exploitation and control relationships were the same, both indicated that one compartment took energy from another, resulting in energy gain of one compartment and energy loss of the other. Mutualism relationship indicated that both compartments increased energy in the process of interaction. Competition relationship indicated that two compartments competed with each other, resulting in energy loss of both (Zhang et al. 2011).
Table 4
Classification of ecological relationships
Matrix Symbols | Positive (+) | Neutral (0) | Negative (-) |
Positive (+) | (+, +) Mutualism | (+, 0) Commensalism | (+, -) Exploitation |
Neutral (0) | (0, +) Commensalism Host | (0, 0) Neutralism | (0, -) Amensalism |
Negative (-) | (-, +) Control | (-, 0) Amensal Host | (-, -) Competition |
Notes: “+” meant positive; “−” meant negative; “0” means neutral.
In order to more specifically and accurately describe the impact of land use changes in PLE space on the development of the entire regional system, this study used an overall utility function M to quantify the ecological relationships among the compartments. The calculation formula was as follows (Zhang et al. 2010):
$$M=\frac{{M}_{+}}{{M}_{-}}$$
16
Where M+ was the number of carbon flows greater than 0 in the matrix U; M− was the number of carbon flows less than 0 in the matrix U. When M > 1, it meant that the role of PLE space evolution on regional carbon metabolism was positive, and the larger M was, the stronger the positive role. When M < 1, it meant that the role of PLE space evolution on regional carbon metabolism was negative, and the smaller M was, the stronger the negative role.
2.3.5 Analysis of Driving Factors of Carbon Metabolism in PLE Space
The PLUS model was a cellular automata model based on raster data, which could be used to simulate land use and land cover change at patch scale (Liang et al. 2021). In this study, whether the driving factors in the simulation process were effective was judged by comparing the results of land use predicted by PLUS model with the accuracy of land use in 2020. For the first present, land expansion and land demand data of the study area from 2000 to 2020 were extracted by Markov model. Then input the data of land expansion and driving factors, and got the land expansion analysis strategy (including the contribution value of driving factors and development potential map of different land use types). Finally, input the land use data in 2000 and the development potential map to simulate the land use data in 2020. Referring to existing researches (Liang et al., 2021; Zhao et al., 2022; Qin et al., 2023), 16 driving factors were selected in this paper: temperature, precipitation, elevation, slope, soil type, NPP, NDVI, population, GDP, nighttime light, distance to railway, distance to highway, distance to primary road, distance to secondary road, distance to tertiary road, and distance to government site (Fig. 4). The simulation for 2020 was compared with the realistic land use data, and the Kappa value was 0.74, and the model accuracy was 86.94%, which satisfied the study requirements. Therefore, the driving factors of land use change of PLUS model were valid.