Changes in Wuhan’s Carbon Stocks and Their Spatial Distributions in 2050 under Multiple Projection Scenarios

Abstract

Urbanization in the 21st century has reshaped carbon stock distributions through the expansion of cities. By using the PLUS and InVEST models, this study predicts land use and carbon stocks in Wuhan in 2050 using three future scenarios. Employing local Moran’s I, we analyze carbon stock clustering under these scenarios, and the Getis–Ord Gi* statistic identifies regions with significantly higher and lower carbon-stock changes between 2020 and 2050. The results reveal a 2.5 Tg decline in Wuhan’s carbon stock from 2000 to 2020, concentrated from the central to the outer city areas along the Yangtze River. By 2050, the ecological conservation scenario produced the highest carbon stock prediction, 77.48 Tg, while the economic development scenario produced the lowest, 76.4 Tg. High-carbon stock-change areas cluster in the north and south, contrasting with low-change area concentrations in the center. This research provides practical insights that support Wuhan’s sustainable development and carbon neutrality goals.

1. Introduction

The 2016 Paris Agreement makes it clear that global warming has become a common concern for humanity and that its main cause is the release of greenhouse gases. As the focus of research attention shifts to ways to reduce carbon emissions and increase carbon storage, land use change has been identified as a key factor influencing the carbon cycle in ecosystems. As a result of rapid urbanization and continued urban sprawl, a large amount of carbon-intensive land has been converted to urban land, which has resulted in land development contributing to 30% of total global carbon emissions [1,2]. Specifically, land-use change has a significant impact on carbon stock and sink function. Studies have shown that land-use change is the main driver of regional carbon-stock changes, e.g., Ostle et al. [3] noted that land-use change has led to a decline in soil carbon stocks by more than 95% in the UK. In addition, Lozano-García et al. [4] emphasized that different land-use patterns have a significant effect on the storage capacity of carbon and nitrogen in soils. In view of this, the IPCC [5] suggests that the carbon sink function of terrestrial ecosystems can be effectively enhanced by adjusting land-use patterns and management practices. In Tongyu County, Jilin Province, China, Tang et al. [6] successfully increased the carbon stock of terrestrial ecosystems by optimizing the land-use structure and found that this strategy was also very effective in reducing carbon emissions [7,8,9]. Therefore, we can judge that there is a strong correlation between land-use change and carbon stock. Thus, we can judge that land-use changes in the study area not only help to reduce carbon emissions, enhance carbon sequestration, and optimize land use allocation but also have important reference value for the government to formulate laws related to ecological environmental protection and land use.

Currently, different land-use change prediction models such as FLUS, PLUS, and InVEST have been widely used to analyze spatial carbon stock distribution. For example, Refs. [10,11,12] analyzed the spatial and temporal distribution of carbon stocks in the study area, based on FLUS and InVEST, while Refs. [13,14,15] and others studied and analyzed the carbon stock distribution in the study area, based on PLUS and InVEST. In addition, Zhao et al. [16] modeled the carbon stock distribution and future carbon stock-changes on the Tibetan Plateau in China using PLUS combined with the InVEST model. Wang et al. [17] projected the distribution of Chengdu-Chongqing Urban Agglomeration’s carbon stock in 2030 under different development scenarios, using these two models. The PLUS-InVEST model has been widely used not only for carbon stock modeling but also for habitat quality prediction. For example, Hu et al. [18] made a simulation evaluation of future habitat quality in Baoding, China using two models. Among the many land-use prediction models, the PLUS model can accurately simulate the spatial and temporal changes of various land types by effectively exploring the mechanisms driving land-use changes [19,20] and is, therefore, utilized in many research studies. Compared with other carbon stock prediction models such as CASA, FORCCHN, and LPJ-GUESS, the InVEST model is widely used in research because of its large amount of data, fast computation speed, and good efficiency [21,22,23,24]. Conversely, research on the carbon stock in Wuhan has also received extensive scientific attention. Liu et al. [25] simulated and evaluated the carbon dynamics of the Wuhan city circle using the CASA model and climate data. Peng et al. [26] analyzed the distribution of carbon stock in the Wuhan city circle in 2030 using the PLUS-InVEST model. Meanwhile, the correlation between land-use-related risks and carbon stocks was evaluated. Wang et al. [1] used the PLUS-InVEST model to simulate the distribution of carbon stocks in Wuhan in 2035, based on the land-use data from 1980, 1995, 2005, and 2015. Through its collation, we learned that most of the present studies, whether examining the carbon stock in Wuhan or other regions or cities, are based solely on a model that simulates the distribution of carbon stock and does not explore zoning for the high- and low-value areas, thereby showing the differences in carbon-stock changes. This gap in the literature restricts the government’s reference basis for the management of spatial zoning regarding the changes in carbon stock.

Land-use change is significantly influenced by government policies [27]. In recent years, the Chinese government’s consistent efforts to enhance sustainable urban ecological development have made a beneficial impact on the carbon cycle. Hence, the identification of areas with high and low carbon-stock changes in cities can be utilized to enhance the effectiveness of current zoning laws for carbon stock management. This has significant ramifications for the government’s future efforts to efficiently manage carbon emissions through zoning.

The rapid development of cities and the development of fossil energy are some of the factors that lead to land-use changes and, thus, contribute to the increasing concentration of greenhouse gases. At the same time, at present, China is in a stage of rapid urbanization, and the type of land use is undergoing drastic changes [28]. Among them, Wuhan, a megacity in central China and the core area of the Yangtze River Economic Belt Development Strategy, is the sixth most populous city in China, with an urbanization level of 79.41% in 2015 [1]; the built-up area of Wuhan increased 3.21 times from 2000 to 2020 [29]. At the same time, as a key city in central China along the middle and lower reaches of the Yangtze River, Wuhan is also one of the first cities in the country to set a specific quantitative target for peak carbon emissions [1]. Wuhan significantly influences ecological security and socioeconomic development in China [30]. Unfortunately, the rapid urbanization seen in Wuhan has led to the extensive expansion of construction land, which has not only significantly harmed the ecological environment [31] but also altered Wuhan’s ecological carbon storage system [32]. These changes are obviously detrimental to the sustainable development and carbon neutrality of Wuhan. Therefore, while promoting economic development, finding ways to scientifically investigate the distribution of carbon storage in Wuhan so as to promote the sustainable development of the city is a great challenge for Wuhan to face in future urban development. Wuhan’s future challenges are similar to those faced by many developing countries during the process of urbanization. Thus, examining land-use changes and estimating carbon emissions amidst rapid urbanization are crucial for the sustainable management of land resources and the long-term development of cities [33]. However, while studies on Wuhan’s carbon stock and emissions are widespread, there is a relative lack of research on the differentiation of changes in the current and projected carbon stock for zoning management purposes. Therefore, the purpose of this paper is to explore and predict the spatial and temporal trends of carbon stock distribution in Wuhan. This paper aims to uncover the distribution of future changes in the carbon stock in Wuhan and to provide theoretical support for optimizing the allocation of land resource utilization and future zonal carbon emission management policy in Wuhan.

This study predicted carbon stocks in Wuhan in 2050 under three distinct scenarios, which are based on an analysis of the land-use changes in Wuhan between 2000 and 2020, utilizing the PLUS-InVEST model. We suggest a novel approach based on this premise that uses the Getis–Ord Gi* statistical method to compare the predicted carbon stock distributions under each scenario with that in 2020, ultimately revealing the spatial distribution of regions that are expected to have significant increases or decreases in carbon stocks. The findings of this study offer a novel approach that can allow the relevant departments to concentrate their efforts on areas showing discrepancies in future carbon stock fluctuations when developing spatially based land resource management plans. This will not only be beneficial for modifying zoning management policy but also to provide a rational scientific foundation for optimizing the utilization of land resources and achieving carbon-neutral cities.

2. Materials and Methods

2.1. Study Area

Wuhan, situated at the coordinates 113°41′–115°05′ E and 29°58′–31°22′ N, is positioned in the central region of China. The region consists of 13 administrative districts. In addition, the elevation is higher in the north compared to the south. The mean annual temperature ranges from 15.8 °C to 17.5 °C, whereas the total annual precipitation varies between 1150 mm and 1450 mm. The urbanization rate of the resident population in 2023 stands at 84.56%. It serves as a primary urban center in central China and is a crucial transportation hub for movement throughout the country (Figure 1).

2.2. Data Sources and Data Processing

This paper compares the census information published by the Wuhan Bureau of Statistics (https://tjj.wuhan.gov.cn/, accessed on 10 May 2024), which shows that the population of Wuhan increased from 1990 (fourth census) to 2000 (fifth census), but the average annual growth rate decreased. From 2000 (the fifth census), to 2010 (the sixth census), and to 2020 (the seventh census), both the population and the annual growth rate show an increasing trend. Since the onset of the 21st century, Wuhan has undergone rapid urbanization and development [34]. Therefore, we believe that the year 2000 was of great significance to the urbanization development of Wuhan. Meanwhile, considering the accuracy of the census data, we set all the data to the same time as the census, i.e., 2000, 2010, and 2020.

The data used in this study are summarized in

Table 1. The data used in this study, with their special accuracies and sources (All sites were accessed on 16 May 2024).

Data Type	Factor	Spatial Accuracy	Source
Land Use Categories	Land use data	30 m	https://zenodo.org/
Natural factor	Mean annual temperature	1 km	https://www.geodata.cn/
	Mean annual precipitation	1 km	https://www.geodata.cn/
	Soil type	vector data	https://www.rserforum.com/
	Digital elevation model (DEM)	30 m	http://www.gscloud.cn/
	Slope	30 m	DEM data extraction
	Distance to open water	vector data	https://www.openstreetmap.org/
	Normalized difference vegetation index (NDVI)	1 km	http://www.nesdc.org.cn/
Social factor	Population	1 km	http://www.resdc.cn/
	Gross domestic product (GDP)	1 km	http://www.resdc.cn/
	Distance to government	vector data	https://www.openstreetmap.org/
	Distance to railway	vector data	https://www.openstreetmap.org/
	Distance to highway	vector data	https://www.openstreetmap.org/
	Distance to national highway	vector data	https://www.openstreetmap.org/

. The land-use data were derived from the China Land Cover Dataset, obtained from the Zenodo repository (https://zenodo.org/, accessed on 16 May 2024) and distributed by Wuhan University. This study uses the land-use data for Wuhan from the years 2000, 2010, and 2020, which provide further details. In accordance with China’s National Land Use Classification System (GB/T 21010-2017) and previous research [35,36,37,38], the 9 land-use classifications in the study region were re-divided into 6 categories: forested land, grassland, cropland, water area, unused land, and construction land.

This study utilized the “Wuhan Spatial Master Plan 2021–2035” and other relevant research findings to select 13 factors—comprising 7 natural influence factors, such as annual precipitation and average annual temperature, and 6 social influence factors, such as population and GDP—as drivers of land-use change in the PLUS model. After acquiring all the data, this study used ArcGIS to trim and arrange the data by aligning the row and column numbers. Finally, we spatially standardized the data into a raster dataset, maintaining a consistent 30-meter geographic resolution in the GCS_WGS_1984 spatial coordinate system.

2.3. Methods

2.3.1. PLUS Model

The PLUS model, developed by the China University of Geosciences, is a predictive tool for land use utilizing classical meta-cellular automata [39]. This model integrates two primary components: the land-use topological analysis strategy (LEAS) and a meta-cellular automaton (CARS), which employs multi-species random patch seeding [40,41,42]. In comparison to alternative models, PLUS demonstrates superior accuracy in forecasting future land-use changes. The LEAS module, which employs a random forest algorithm, evaluates the impact of various drivers on changes in land-use types. Concurrently, the CARS module models the autonomous generation of land patches based on developmental probabilities, incorporating both stochastic seed generation and cutoff descent mechanisms. By integrating Markov chains, the CARS module facilitates the simulation and prediction of future land-use patterns. Consequently, the PLUS model proves to be a significant asset for managing land use across diverse growth scenarios [41].

This paper begins by evaluating the accuracy of the PLUS model through a series of methodical steps. Initially, land-use data for Wuhan from 2000 and 2010, along with 13 identified drivers, are input into the land-use expansion analysis strategy (LEAS) module of the PLUS model. This step produces land-use expansion maps for the period between 2000 and 2010 and assesses the influence of the 13 drivers on the topography of various land-use types during this timeframe. Subsequently, the impacts of these drivers, together with the 2010 land use data, are incorporated into the cellular automata for randomized seed patch seeding, based on the multiple species (CARS) module of the PLUS model. A Markov chain is then employed to forecast Wuhan’s land-use status for 2020, utilizing a land transfer rule matrix. Finally, the predicted land-use data for 2020 is compared with the actual 2020 land-use data. The model’s performance metrics include a Kappa coefficient of 0.82, a FOM value of 0.47, and an overall accuracy (OA) of 0.86. The Kappa coefficient, which ranges from 0 to 1, is a key indicator of the agreement between simulated and reference images, with values above 0.8 signifying high concordance [18,43]. The Kappa coefficient of 0.82 achieved in this study indicates a high level of accuracy in the land-use simulation that was conducted using the PLUS model, suggesting its viability for predicting and simulating future land use in the study area.

2.3.2. Land-Use Transition Matrix

The land-use transfer matrix, which is two-dimensional, provides a precise representation of changes in land-use categories over different time intervals within a defined area and is essential for studying land-use changes [44]. By examining this matrix, one can discern the transformations between different land-use types across two separate time points. It has been extensively applied in academic studies [45,46]. The definition of the land-use transfer matrix is as follows:

$$S_{ij}=\left[\begin{array}{ccc}{S}_{11}& S_{12}& {\begin{array}{cc}\cdots & S\end{array}}_{1n}\\ \begin{array}{c}{S}_{21}\\ \cdots \end{array}& \begin{array}{c}{S}_{22}\\ \cdots \end{array}& \begin{array}{c}\begin{array}{cc}\cdots & S_{2n}\end{array}\\ \begin{array}{cc}\cdots & \cdots \end{array}\end{array}\\ S_{n1}& S_{n2}& \begin{array}{cc}\cdots & S_{nn}\end{array}\end{array}\right]$$

(1)

where S_ij represents the relationship between the starting land-use type i and the final land-use type j in area S, with n representing the total number of land-use categories. Thus, S₁₁ represents the area that started as land-use type 1 and did not change, S₁₂ represents the area that started as land-use type 1 and changed to land-use type 2, etc.

2.3.3. Multi-Scenario Setting

This paper establishes three development scenarios and formulates a land-use transfer rule matrix based on previous studies (

Table 2. Land-use transfer rule matrices under the three development scenarios in Wuhan are categorized as follows: A—cropland; B—forested land; C—grassland; D—water area; E—construction land; F—unused land).

	Scenario A (NDS)						Scenario B (ECS)						Scenario C (EDS)
	A	B	C	D	E	F	A	B	C	D	E	F	A	B	C	D	E	F
A	1	1	1	1	1	1	1	0	0	0	1	1	1	1	1	1	1	0
B	1	1	1	1	1	1	0	1	0	0	0	0	0	1	0	0	0	0
C	1	1	1	1	1	1	0	1	1	0	0	0	1	1	1	1	1	0
D	1	0	0	1	0	0	0	0	0	1	0	0	0	0	0	1	0	0
E	0	0	0	0	1	1	0	0	0	0	1	1	0	1	1	1	1	0
F	1	1	1	1	1	1	0	0	0	0	1	1	1	1	1	1	1	1

), relevant policies such as the Wuhan Territorial Spatial Master Plan 2021–2035, and the actual land-use changes in Wuhan from 2000 to 2020.

Scenario A (natural development scenario): The transfer matrix of land use in Wuhan from 2000 to 2020 can be analyzed through the land-use topology analysis strategy (LEAS) module in the PLUS model. The change transfer matrix of land use in Wuhan from 2000 to 2020 is taken as the basis and used without interference. The original land-use transfer probability is maintained to simulate the land-use type in Wuhan city in 2050 under the natural development scenario.

Scenario B (ecological conservation scenario): Reference is made to the Wuhan Territorial Spatial Master Plan 2021–2035 and the National Outline of Land Use Master Plan (2006–2020). In order to address the climate crisis and protect the biodiversity of Wuhan, an ecological protection scenario is set up, under which water bodies cannot be converted to other land-use types except for restricted forested land, and grassland can only be converted to forested land. Other land-use types maintain their original development probability.

Scenario C (economic development scenario): A development scenario centered on economic development is established by referring to previous studies [15,47] and the Wuhan Urban Planning Guidelines 2030. Under this scenario, economic development is taken as the core to accelerate urbanization, resulting in the rapid growth of construction land. At the same time, it restricts the conversion of construction land to arable land or unutilized land and increases the possibility of converting other types of land to construction land.

2.3.4. Spatial Autocorrelation

Spatial autocorrelation analysis is a method that is frequently employed for evaluating the presence and magnitude of correlations between identical data types in spatially adjacent regions [48]. We can divide spatial autocorrelation into two categories: global and local. Global spatial autocorrelation is a measure of whether attributes in spatial data exhibit clustered, dispersed, or random patterns. It is usually evaluated using the global Moran’s I equation (Equation (2)) [49,50]:

$$M_I={\displaystyle \frac{{\sum }_{i=1}^n\sum _{j=1}^n{W}_{ij }({x_i-\overline{x})}^2{(x}_j-\overline{x})}{{\sum }_{i=1}^n\sum _{j-1}^n{W}_{ij}\sum _{i=1}^n{(x_i-\overline{x})}^2}}$$

(2)

in which n is the number of items, $x_i$ and $x_j$ denote the attribute values of the ith and jth elements, respectively, $\overline{x}$ represents the mean of the attribute values, and $W_{ij}$ denotes the spatial weight of the items in the ith row and jth column.

Local spatial autocorrelation is a method for identifying and studying local clustering and variations in the distribution of spatial objects. It is usually implemented using the local Moran’s I statistic (Equation (3)):

$${\mathrm{M}}_{\mathrm{i}}={\displaystyle \frac{\left({\mathrm{x}}_{\mathrm{i}}-\overline{\mathrm{x}}\right)\sum _{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{W}}_{\mathrm{i}\mathrm{j}}\left({\mathrm{x}}_{\mathrm{j}}-\overline{\mathrm{x}}\right)}{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}({{\mathrm{x}}_{\mathrm{i}}-\overline{\mathrm{x}})}^2}}.$$

(3)

The interpretation of each variable in the formula is identical to that of the global Moran’s I in Equation (2).

As a major aim of this study is to provide an understanding of the spatial arrangement of carbon stores in the region, spatial autocorrelation analysis is employed to help characterize the spatial relationships between high- and low-carbon storage land use. It is worth noting that global spatial autocorrelation can only provide a general overview of the spatial distribution of attributes. Therefore, we classified carbon stocks using local spatial autocorrelation analysis into five categories, based on the degree of concentration in adjacent areas: negligible, high–high, high–low, low–low, and low–high. The local Moran I values range from −1 to 1. A value greater than 0 indicates a positive spatial correlation, i.e., neighboring spatial units have similar carbon storage distribution properties (e.g., high–high or low–low). Conversely, a value below zero suggests a negative spatial correlation, wherein neighboring spatial units display contrasting trends in their attributes (for example, high versus low or low versus high). A value of 0 indicates no autocorrelation (negligible), meaning that the spatial distribution of those attributes is random.

2.3.5. Getis–Ord Gi* Statistic

The Getis–Ord Gi* statistic is frequently employed in local spatial autocorrelation analysis to examine the clustering of a specific feature in geographical data, which is also known as hotspot analysis [51,52]. Getis–Ord Gi* analysis examines the arrangement of spatial attributes by comparing the spatial correlations of neighboring pixels. This allows for the identification of clusters of pixels with high values, known as hot spots, as well as clusters of pixels with low values, known as cold spots [53]. The Getis–Ord Gi* statistic is more effective than Moran’s I in analyzing spatial autocorrelation because it can more clearly distinguish between spatial clusters with high and low attribute values [54]. The result of the Getis–Ord Gi* statistic is the Gi* value, which is also referred to as the z-score. The z-score’s magnitude affects the spatial grouping of high and low values [55]. The formula is computed using the following equation:

$$z\left(Gi\mathrm{*}\right)={\displaystyle \frac{\sum _{j=1}^n{w}_{i,j}{x}_j-\overline{x}{\sum }_{j=1}^n{w}_{i,j}}{\sqrt[\mathrm{s}]{\frac{\left[n\sum _{j=1}^n{w}_{i,j}^2-{\left({\sum }_{j=1}^n{w}_{i,j}\right)}^2\right]}{n-1}}}}$$

(4)

in which $x_j$ is the attribute value of element j, $w_{i,j}$ is the spatial weight between spatial elements i and j, and n is the total number of elements. When z is positive, higher z-scores indicate a stronger clustering of high-value pixels in the hot spot; conversely, when z is negative, lower z-scores indicate a stronger clustering of low-value pixels in the cold spot (i.e., low values) [56].

2.3.6. InVEST Model

The InVEST model is the result of a collaboration between The Nature Conservancy (TNC) and Stanford University. Its purpose is to provide a range of ecosystem service assessments [57]. This paper utilizes the carbon sequestration module of the InVEST model, combined with the PLUS model, to predict and simulate the distribution of carbon stocks in Wuhan. Firstly, we can calculate the carbon stock of terrestrial ecosystems in the study area by inputting the per unit area carbon density for each land-use type:

$$C_m=C_{m-above}+C_{m-below}+C_{m-soil}+C_{m-dead}$$

(5)

where $C_m$ represents the total carbon density (t/hm²) of land-use type m; $C_{m-above}$, $C_{m-below}$, $C_{m-soil}$, and $C_{m-dead}$ denote the aboveground biogenic carbon density (t/hm²), belowground biogenic carbon density (t/hm²), soil organic carbon density (t/hm²), and dead organic carbon density (t/hm²), respectively. Secondly, the carbon density is aggregated based on the land-use area data:

$$C_{m-total}=\left(C_{m-above}+C_{m-below}+C_{m-soil}+C_{m-dead}\right)\times S_m$$

(6)

where $C_{m-total}$ denotes the aggregate carbon stock (t) of a given land-use type, with $S_m$ indicating the area (hm²) occupied by the land-use type. Lastly, the areas total carbon stock in the study area is computed by summing the total stocks of each land use:

$$C_{total}=\sum _{m=1}^n{C}_m\times S_m$$

(7)

where $C_{total}$ indicates the total carbon storage in the study area in metric tonnes (t).

This study draws on the National Ecological Sciences Database Centre (http://www.nesdc.org.cn/, accessed 17 May 2024) and previous studies [37,58,59] on determining carbon density values (

Table 3. Carbon density of land-use types (t/hm²).

Land-Use Type	${\mathit{C}}_{\mathit{a}\mathit{b}\mathit{o}\mathit{v}\mathit{e}}$	${\mathit{C}}_{\mathit{b}\mathit{e}\mathit{l}\mathit{o}\mathit{w}}$	${\mathit{C}}_{\mathit{s}\mathit{o}\mathit{i}\mathit{l}}$	${\mathit{C}}_{\mathit{d}\mathit{e}\mathit{a}\mathit{d}}$
Cropland	4.02	0.75	98.13	0
Forested land	22.62	18.03	126.75	0
Grassland	3.6	11.7	90.43	0
Water area	1.59	0	64.03	0
Construction land	0.83	0.08	43.71	0
Unused land	0.59	0.64	28.42	0

). Additionally, several researchers have suggested that obtaining data on dead organic carbon density is more challenging and that it represents only a small fraction of the overall carbon density [60,61]. Therefore, we decided to assign a value of 0 to dead organic carbon density in this study.

2.4. Research Framework

In general, the research framework can be categorized into four components (Figure 2):

• The PLUS model utilizes the 2020 land-use data and 13 influencing factors to forecast the land use in Wuhan in 2050. This prediction is made under three development scenarios: NDS, ECS, and EDS.
• The InVEST model predicts the distribution of carbon stocks in Wuhan under various periods and development scenarios.
• The carbon stock distribution in Wuhan in 2050 is analyzed using local spatial autocorrelation analysis under each development scenario, using the local Moran’s I statistic in the GeoDa software tool (https://geodacenter.github.io/).
• Finally, the spatial distribution of high-value aggregation locations (hot spots) and low-value aggregation areas (cold spots), in terms of the changes in carbon stocks between 2000 and 2050, was analyzed for each development scenario using the Getis–Ord Gi* statistic.

3. Results

3.1. Land-Use Changes in Wuhan

3.1.1. The Drivers of Land-Use Change

In this study, the PLUS model was employed to evaluate the relative contributions of various factors influencing land-use changes in Wuhan from 2000 to 2020 (Figure 3). This reveals several key patterns. The greatest influence on changes in arable land can be attributed to the DEM, while variations in mean annual temperature predominantly affect changes in grassland. Additionally, the mean annual precipitation value has the greatest impact on changes in forested land. The primary elements that contribute to the shifts in land use for water bodies, built-up land, and unused land are their proximity to open water, population density, and distance to highways, respectively. These high-contribution drivers include four natural factors and two societal factors among them. This study demonstrates that natural factors significantly influence land-use change. Simultaneously, the highest-contributing factor never outweighed the cumulative contributions of other land-use change drivers, which also play crucial roles in facilitating the modeling of future land-use change in Wuhan.

3.1.2. Land-Use Change from 2000 to 2020

In Wuhan, from 2000 to 2020, the total area of cultivated land dropped by 677.1 km². However, cultivated land remained the dominant land use in Wuhan, accounting for 64.85% of the total area in 2020 (

Table 4. Land-use areas and percentages (Pct) in Wuhan from 2000 to 2020.

Land-Use Type	2000		2010		2020
	Area (km2)	Pct (%)	Area (km2)	Pct (%)	Area (km2)	Pct (%)
Cropland	6242.16	72.74	5931.87	69.13	5565.06	64.85
Forested land	510.14	5.95	517.50	6.03	647.52	7.54
Grassland	3.11	0.04	3.95	0.05	1.24	0.014
Water area	1298.16	15.13	1267.00	14.76	1187.26	13.83
Construction land	526.35	6.13	860.39	10.03	1179.52	13.74
Unused land	1.05	0.01	0.26	0.003	0.37	0.004

and Figure 4 and Figure 5). Nevertheless, the extent of aquatic bodies, constituting 13.83% of the total area in 2020, was only surpassed by that of arable land and displayed a declining pattern. The unused land area also experienced a decline, but its contribution to the total area was quite modest, resulting in minimal fluctuations in its size. From 2000 to 2020, urban areas in Wuhan underwent substantial expansion, leading to a notable increase in construction land. Over this 20-year period, the construction land area grew by 653.17 km², accounting for 13.74% of the total land area in 2020. This increase was primarily attributable to the conversion of arable land into new construction sites. Thus, a significant portion of agricultural land is being transformed into new development sites. Between 2000 and 2020, approximately 88.5% of new urban land was converted from agricultural land (

Table 5. Transfer matrix for land-use types in Wuhan between 2000 and 2020 (km²).

		2020
		Cropland	Forested Land	Grassland	Water Area	Construction Land	Unused Land	Total
2000	Cropland	5267.09	194.81	1.00	179.34	599.72	0.21	6242.16
	Forested land	54.27	451.19	0.104	1.03	3.54	0.00	510.14
	Grassland	1.20	0.77	0.100	0.50	0.55	0.00	3.11
	Water area	239.88	0.65	0.038	997.55	59.89	0.15	1298.15
	Construction land	2.39	0.10	0.000	8.22	515.64	0.01	526.35
	Unused land	0.24	0.000	0.002	0.62	0.19	0.00	1.05
	Total	5565.06	647.52	1.24	1187.26	1179.52	0.37

). The area of forested land exhibited a consistent upward trajectory, experiencing a growth rate of 1.59% between 2000 and 2020; lastly, the grassland area exhibited initial growth, followed by a decline.

3.1.3. Land Use in 2050 under Three Development Scenarios

Scenario A (NDS): Under this scenario, agricultural land decreases by 10.5%, grassland decreases by 0.003%, and unused land decreases by 0.0008% between 2020 and 2050 in Wuhan (

Table 6. Land use areas and percentages (Pct) in Wuhan in 2050 under different development scenarios.

Land Use Type	2050 NDS		2050 ECS		2050 EDS
	Area (km2)	Pct (%)	Area (km2)	Pct (%)	Area (km2)	Pct (%)
Cropland	4659.73	54.3	4659.73	54.3	4748.08	55.3
Forested land	682.49	7.9	748.74	8.7	647.61	7.54
Grassland	0.90	0.01	0.91	0.01	1.10	0.01
Water area	1309.17	15.2	1355.15	15.7	1187.77	13.8
Construction land	1928.38	22.47	1816.15	21.1	1996.11	23.2
Unused land	0.30	0.003	0.29	0.003	0.30	0.003

and Figure 6), with small fluctuations. Conversely, forested land increases by 0.4%, water bodies increase by 1.4%, and construction land increases by 8.7%. Thus, cultivated land decreases the most, losing 905.33 km², while built-up land increases the most, gaining 748.86 km². This suggests that under natural development conditions, cultivated land continues to be the predominant land use in Wuhan in the year 2050. However, the urban land area increases rapidly.

Scenario B (ECS): Between 2020 and 2050, this scenario produces a predicted 905.33 km² reduction in cultivated land area (Table 6 and Figure 6), while grassland and unused land decrease by 0.33 km² and 0.08 km², respectively. Conversely, built-up land increases by 636.63 km², and water bodies and forested land increase by 167.89 km² and 101.22 km², respectively. The conversion of agricultural land is the main contributor to the expansion of construction land, water bodies, and forested land, with construction land accounting for 70% of the total shift from cropland.

Scenario C (EDS): Under this scenario, the built-up area increases by 816.59 km² between 2020 and 2050 (Table 6 and Figure 6). In addition, the area of forested land increases by 0.09 km², and the area of water bodies increases by 0.51 km². The remaining land-use types show a decreasing trend, with the area of cropland decreasing by 816.98 km², the area of grassland decreasing by 0.14 km², and the area of unused land decreasing by 0.07 km². In terms of land-use conversion, the majority of urban development takes place on agricultural land, and almost all of the lost agricultural land is converted into urban areas, while conversions between other land types are insignificant.

Upon analyzing the land-use changes across the different development scenarios, we discovered that the EDS scenario produces the most significant expansion in urban land area. This finding aligns with the expected influence of economic development on construction land, as supported by previous studies [62,63]. Nevertheless, the reductions in the cultivated land area are nearly identical under both the ECS and NDS scenarios, although there are variations in the land transfer parameters. However, in contrast to the NDS scenario, the ECS scenario shows a drop in the share of cultivated land being converted to construction land, while the share being converted to water bodies and forested land increases. Additionally, the carbon stock in 2050 under the ECS scenario is considerably higher than that under the other two scenarios. This verifies the importance of implementing ecological protection policies when promoting the future sustainable development of Wuhan.

3.2. Carbon-Stock Changes in Wuhan

3.2.1. Carbon-Stock Changes from 2000 to 2020

We used the InVEST model in this study to predict and simulate the distribution of carbon stocks in Wuhan from 2000 to 2020. The total carbon stock in Wuhan was 83.67 Tg in 2000, 81.88 Tg in 2010, and 81.17 Tg in 2020 (

Table 7. Carbon reserves in each land-use type in 2000, 2010, and 2020 and as predicted under three development scenarios in 2050.

Land-Use Type	2000	2010	2020	2050 NDS	2050 ECS	2050 EDS
Cropland	64.23	61.03	57.26	47.95	47.95	48.85
Forested land	8.53	8.66	10.83	11.42	12.52	10.84
Grassland	0.03	0.04	0.01	0.01	0.01	0.01
Water area	8.51	8.31	7.80	8.58	8.89	7.79
Construction land	2.34	3.83	5.26	8.60	8.10	8.90
Unused land	0.03	0.007	0.01	0.01	0.01	0.008
Total	83.67	81.88	81.17	76.57	77.48	76.40

). These values demonstrate a consistent downward trend. During the period from 2000 to 2010, there was a decline of 1.79 Tg, and from 2010 to 2020, there was a 0.71 Tg decrease. The decrease between 2000 and 2010 was comparatively bigger. This is because Wuhan saw significant urban growth and the expansion of its built-up land areas during this period, driven by economic and demographic development. During the 2000–2020 period, the distribution of carbon stocks revealed that cropland held the highest proportion. However, the carbon stocks of cropland experienced a declining trend over these two decades. The main reason for this decline may be the rapid urbanization of Wuhan, which resulted in a large amount of agricultural land being turned into urban areas. The spatial distribution of carbon-stock changes in Wuhan between 2000 and 2020 reveals a consistent radial reduction in carbon stocks, expanding from the Yangtze River in the city’s center toward the city’s outskirts (Figure 7), which finding aligns with the ongoing urban growth of Wuhan.

3.2.2. Carbon Stocks in 2050 under Three Development Scenarios

This study utilized the InVEST model to characterize the distributions of carbon stocks in Wuhan in 2050, as predicted based on three unique development scenarios. The findings indicate that carbon stocks in 2050 for all three scenarios are below the level observed in 2020. The carbon stock is 77.48 Tg under Scenario B (ECS), 76.57 Tg under Scenario A (NDS), and 76.4 Tg under Scenario C (EDS) (Table 7). Thus, the ECS scenario has the greatest carbon storage capacity. This is because the environmental protection measures in place effectively limit the expansion of construction land. It also suggests that implementing effective ecological protection measures will assist Wuhan in achieving its long-term goal of sustainable development. Analysis of the projected spatial distributions of carbon stocks in Wuhan reveals that the spatial patterns anticipated for 2050 align with the trends observed between 2000 and 2020: low-carbon-stock areas exhibit a declining trend as one moves away from the city’s center on the Yangtze River toward the surrounding areas (Figure 8).

3.3. Spatial Autocorrelation

3.3.1. Local Spatial Autocorrelation, Based on Moran’s I Statistic

The study area underwent a resampling process to achieve a resolution of 1 km. Subsequently, spatial autocorrelation analysis was performed using the GeoDa software to examine the local aggregation of carbon stock distributions in Wuhan for the year 2050 under three different development scenarios. In 2050, the calculated Moran’s I values for scenario A (NDS), scenario B (ECS), and scenario C (EDS) are predicted to be 0.659, 0.664, and 0.656, respectively (Figure 9). These findings suggest that the distribution of carbon reserves in Wuhan exhibits a non-random pattern and demonstrates a positive spatial association. Moreover, the scatter plot indicates that most of the sampling points are grouped in the first quadrant (high–high) and the third quadrant (low–low). This suggests that regions in Wuhan with higher carbon stocks are typically surrounded by other areas with higher carbon stocks. In contrast, locations with low carbon reserves are typically found near other areas with low values.

In order to deeply understand the spatial distribution of carbon-stock aggregation under the three development scenarios in 2050, we carried out local indicators of spatial association (LISA) analyses of carbon stocks using ArcGIS. In Scenario A (NDS), the low–low aggregation area is primarily located in the central part of Wuhan, while the high–high aggregation area is primarily located in the northern and southern parts of Wuhan (Figure 10). The spatial distributions of high- and low-carbon stock aggregation areas in Scenario B (ECS) and Scenario C (EDS) are very similar to those in Scenario A (NDS). It can be hypothesized that by 2050, Wuhan’s carbon stock will be mainly concentrated in the northern and southern regions, which are mostly dominated by cropland and forests. Conversely, the central region, mostly composed of built-up urban land, will exhibit the primary areas with lower carbon stocks. These assessments have the potential to improve Wuhan’s future carbon stock management, promoting better and more sustainable city development.

3.3.2. Getis–Ord Gi* Statistic

In this study, the differences in the distributions of carbon stocks in Wuhan between 2020 and 2050, as predicted under three development scenarios, were statistically analyzed with the Getis–Ord Gi* statistic using ArcGIS, revealing the spatial distribution of high- and low-carbon stock-change regions. The results of the study show that the spatial arrangement of areas with and without significant changes in carbon stocks between 2020 and 2050 remains consistent across the three development scenarios (Figure 11). Specifically, the regions with notable fluctuations in carbon storage are mainly located in the southern and northern parts of Wuhan, while the areas with the least variation in carbon storage are mainly concentrated in the central part. Compared with the other scenarios, Scenario B (ECS) has carbon stock-change hotspots, i.e., there are fewer regions with high carbon stock-change differences. Conversely, Scenario C (EDS) produced the largest differences; in other words, it has the largest number of areas with a high carbon stock-change difference.

4. Discussion

4.1. Impact of Different Development Scenarios on Carbon Stocks in the Future in Wuhan

Wuhan plays a crucial role in China’s endeavor to construct an eco-friendly and resource-conserving society. Therefore, it is critical for Wuhan’s future to improve land resource utilization, foster a low-carbon economy, and protect the environment. An analysis of carbon stock distributions in Wuhan in 2050 across three development scenarios revealed that Wuhan has the highest carbon stocks in 2050 under the ECS (ecological conservation scenario), which surpassed the other two scenarios. These findings support the notion that Wuhan’s ecosystems would experience improvements if effective policies, such as ecological protection, were implemented today. As a result, Wuhan’s carbon stocks would increase, allowing for more optimal utilization of the city’s carbon sequestration potential.

Through an examination of land-use changes and carbon stock analysis, it becomes evident that agricultural land in Wuhan holds significant importance. Agricultural land, possessing the highest total carbon stock, serves as the primary source of land for urban development. This highlights the importance of cultivated land for Wuhan’s urban development and as the basis for its sustainable development goals. Additionally, agricultural land serves as a means of ensuring food security and, thus, is a crucial factor in maintaining social stability and promoting sustainable development [64]. As Wuhan experiences urban expansion, there is a growing need for construction land, which, in turn, leads to a significant decrease in arable land. These findings align with the research of other experts, who have also identified urban expansion as a significant contributor to the loss of arable land [65,66,67]. While the loss of arable land is an unavoidable consequence of urbanization, it is illogical to only prioritize urban expansion, disregarding the ecological importance of arable land in terms of carbon storage, especially when considering the progress made thus far regarding sustainable development. Henceforth, Wuhan should establish an ecological protection policy that aligns with its specific urban development circumstances. This policy should prioritize the ecological significance of arable land in terms of carbon storage and enhance the efficiency of converting arable land. By doing so, Wuhan can enhance its environment ecologically, optimize land and resource utilization, and ultimately achieve sustainable development objectives.

4.2. Policy Recommendations for Future Carbon Stock Zoning Management in Wuhan

This article examines the disparities in carbon stock distributions between 2020 and 2050 in Wuhan under three development scenarios. The results show that, under all scenarios, the southern and northern parts of Wuhan are the areas with large changes in carbon stock, and, when combined with the various land-use types, it is evident that the southern and northern parts of Wuhan are currently dominated by cropland and forested land. This is likely due to changes in the original land-use type or the destruction of the original ecosystem. Therefore, in these areas, management measures aimed at maintaining the original ecosystem, such as restricting some of the land-use conversion activity, should be taken to help Wuhan reach its carbon neutrality target. In the case of the central city of Wuhan, which has not changed much, measures such as connecting the existing ecological network of the city, improving the construction of urban parks and green spaces, and restricting urban sprawl should be taken. In light of the differences in carbon-stock changes over the study period, the development of scientific zoning management is essential to promote sustainable development and carbon neutrality in Wuhan. We believe that not only Wuhan but also other cities can develop targeted and scientific policies by analyzing the differences in carbon stocks and their relationship with land use. Although studying the distribution of carbon stocks in the city as a whole can help to promote sustainable urban development, there are differences in carbon stocks and land use in different regions within the city. Therefore, zoning management to address these regional differences can improve the efficiency of sustainable development and help more cities achieve balanced economic and ecological development. At the same time, we compare urban policy studies in other countries and regions and find that multi-scenario land use modeling and policy planning play an important role in urban expansion [68,69,70], which is consistent with our view. Therefore, we conclude that land-use change simulations provide an important reference for sustainable urban development and policymakers in both Wuhan and other regions.

4.3. Limitations and Future Prospects

This paper uses the PLUS-InVEST model to simulate and analyze land-use changes and carbon-stock changes in Wuhan. When combined with local spatial autocorrelation analysis, it can offer favorable theoretical support for the future management of carbon emission zoning in Wuhan and the sustainable development of land resources. However, there are certain limitations to the study. First of all, this paper predicts future land-use changes using the PLUS model, but the prediction does not fully consider the influence of future land-use planning and ecological protection policies, so the prediction results are uncertain [71,72]. Furthermore, the InVEST model, while superior to other models in terms of accuracy and computational efficiency, primarily predicts the carbon stock based on the type of land use and the carbon density of each land-use type, while neglecting the impact of other factors on the carbon stock. These influencing factors encompass human activities, the cycle of vegetation growth, and the environment surrounding vegetation growth, among others. These influencing factors have strong uncertainties and individual differences. It is a challenging process to accurately grasp the changing patterns of these influencing factors and the differences between individuals, as well as to quantify and analyze them. Therefore, we think that the latest land-use policy factors should be fully considered in future research on land-use change, along with the needs of the new policies and the appropriate adjustments to the land-use prediction model. In this manner, we can discern the distinctions between the land-use change prediction model under the new policy and the previous model, thereby determining the rationality of the new policy. Simultaneously, exploring new research avenues could enhance the analysis’s precision. In terms of measuring and counting carbon stocks, our comparison of studies for other regions [73,74] shows that when carbon stock projections are made for small-scale areas such as forests, ecological reserves, etc., most of the studies combine field sampling with other methods. This is because field sampling, although challenging, significantly improves the accuracy of the study. However, in studies targeting urban carbon stocks, few previous studies have used field-sampling methods. This may be due to the complexity of urban environments and the high cost of in situ sampling, and researchers have preferred to rely on models to make predictions. Therefore, we believe that when the city as a whole or in a large-scale region is studied as an object site, there are certain difficulties in grasping the variability of each different individual in the region, both economically and in terms of manpower. On the contrary, for the study of localized or small areas of the city, a simulation using the carbon stock model, combined with fieldwork, is more conducive to accurately grasping the distribution of carbon stock in the local area, so as to provide a more scientific and reasonable theoretical basis for the efficient and sustainable development of the city in the future.

5. Conclusions

This study employed the PLUS-InVEST model to simulate and evaluate changes in carbon stock distribution in Wuhan from 2000 to 2020, and to forecast these changes for 2050 under three different development scenarios. We used local spatial autocorrelation to analyze the future distribution of carbon stock clusters in Wuhan. Additionally, the differences in carbon stock between different regions were also visualized and analyzed using the Getis–Ord Gi* statistical method. The research results provided four main findings:

• Between 2000 and 2020, the predominant land-use type in Wuhan was agricultural land. However, as urbanization progressed and built-up land gradually increased, most new urban areas were derived from cropland. The carbon stock declined by 2.5 Tg between 2000 and 2020 due to changes in land use, and low-carbon storage land areas are now concentrated at the city center on the Yangtze River, radiating out into the surrounding areas.
• Our carbon stock analysis for the three development scenarios in 2050 indicate that the ECS (ecological conservation scenario) yields the highest projected future carbon stock, maintaining 77.48 Tg. This suggests that implementing ecological conservation policies today can effectively support Wuhan in achieving sustainable development goals and carbon neutrality in the future.
• The spatial distributions of carbon stocks in Wuhan under all three development scenarios in 2050 were positively autocorrelated, and the regions with significant carbon-stock accumulation were primarily situated in the southern and northern parts of Wuhan, areas characterized by forests and cultivated land. Conversely, the areas with minimal carbon-stock accumulation were predominantly found in the central part of Wuhan, consisting mainly of built-up land. This suggests that urban areas have an important influence on carbon stocks.
• Analyzing the differences in carbon-stock changes between 2020 and 2050 under each development scenario, we found that the ECS has the least number of high-carbon stock-change areas, while the EDS (economic development scenario) has the most. The south and north primarily host the high-carbon stock-change areas, while the center hosts the low-carbon areas. To foster Wuhan’s sustainable growth, we need to develop regionalized strategies for managing these carbon stock-change variations and execute scientific and effective ecological and environmental safeguards to enhance Wuhan’s carbon reserves.