Regional Carbon Stock Response to Land Use Structure Change and Multi-Scenario Prediction: A Case Study of Hunan Province, China

Abstract

Modifications in land use patterns exert profound influences on the configuration, arrangement, and functioning of terrestrial ecosystems, thereby inducing fluctuations in carbon sequestration. Consequently, precise ecological decision-making and an in-depth exploration of the interplay between land use alterations and carbon storage dynamics assume paramount importance in the pursuit of optimal regional land use configurations. In this investigation, we employed the InVEST model to analyze the spatiotemporal variations in land utilization and carbon storage in Hunan Province, based on comprehensive land use data spanning the period from 2000 to 2020. Additionally, the PLUS model was utilized to project the future spatial distribution of carbon storage in Hunan Province until 2040, encompassing diverse development scenarios. The findings of our study are as follows: (1) Land use changes instantaneously impact carbon storage within the study area. From 2000 to 2020, urban construction land witnessed an expansion of 3542 km², which accounted for an increase from 1.13% to 2.78% of the total land area. Consequently, there was a decline in arable land, woodlands, and grasslands, resulting in a reduction of 3430.25 tons of carbon storage in Hunan Province. (2) The ecological protection scenario is projected to yield the most substantial increase in carbon storage, with an estimated magnitude of 7.02 × 10⁶ tons by the year 2040. According to the natural evolution scenario, the total amount of carbon storage is anticipated to remain similar to that of 2020, with a marginal increase of 2.81 × 10⁵ tons. Under the arable land protection scenario, carbon storage is predicted to decrease by 1.060 × 10⁷ tons. Conversely, the urban development scenario is expected to result in the most substantial reduction of 2.243 × 10⁷ tons of carbon storage. These findings underscore the efficacy of adopting ecological protection and natural development policies in curbing the decline in carbon storage. (3) The geographic distribution of carbon storage areas exhibits a strong correspondence with that of land use. Regions characterized by elevated carbon storage levels exhibit minimal urban construction land, an abundance of compact and contiguous ecological land, and a higher frequency of such land parcels. To enhance regional carbon storage levels and achieve sustainable development goals, future endeavors should prioritize the implementation of ecological protection and natural development policies.

1. Introduction

In recent years, the rise in temperatures caused by the greenhouse effect has had adverse effects on the living environments of regional wildlife, vegetation, and humans [1]. Rapid urbanization and industrialization have caused changes in the distribution of land uses in particular regions, leading to the disruption of the functionality and structure of terrestrial ecosystems. This has caused disruptions in carbon cycling, leading to global climate change and other environmental issues. A large amount of carbon-fixing units in terrestrial ecosystems is essential for absorbing carbon dioxide, maintaining climate stability, and preserving the balance of global carbon cycling [2,3,4]. Land use and land cover changes have a direct impact on natural ecosystems’ carbon stocks, and different land use types have varying carbon sequestration capabilities. Future regional environmental protection policies will rely heavily on the accurate evaluation of the effects of shifts in land use on the carbon storage of natural ecosystems, given that alterations in land use are expected to cause fluctuations in carbon storage. Additionally, during the 19th National Congress of the Communist Party of China, the government announced its objective to establish a modern and eco-friendly economic system, promote low-carbon and circular development, and endeavor to establish a novel model for harmonious coexistence between humanity and the natural world [5,6,7]. During the 75th session of the United Nations General Assembly, China made a pledge to achieve carbon neutrality by 2060, while also aiming to attain peak carbon emissions by 2030. Against this background, an in-depth analysis of the response relationship and mechanism between carbon stocks and land use changes will provide valuable suggestions for implementing urban low-carbon development and national spatial planning. This, in turn, will aid in achieving China’s “dual-carbon” goals of striving for peak carbon emissions before 2030 and achieving carbon neutrality before 2060 [8,9].

Currently, scholars worldwide are investigating the impacts of land use changes on carbon storage. They are employing various methodologies to accurately assess and estimate carbon stocks within ecosystems, providing quantitative information that aids in decision-making for environmental conservation and resource management. In contrast, “undesired carbon accumulation” refers to the excessive buildup of carbon resulting from human activities, which contributes to climate change and detrimentally affects ecosystems. Traditional research methods, such as field sampling, biomass estimation, and inventory methods, have certain limitations in terms of the research scale, spatial and temporal variability of the carbon storage, and visualization. To address these limitations, some scholars have adopted a coupled modeling approach to analyze regional carbon storage dynamics and predict future scenarios accurately. This approach enables the precise identification of the source–sink function of carbon in the region [10,11,12]. The InVEST model is an integrated tool developed collaboratively by Stanford University, the Nature Conservancy (TNC), and the World Wildlife Fund (WWF) to evaluate ecosystem services and trade-offs. It has been widely used in multiple countries and regions for ecosystem service assessments, and its accuracy has been validated. Researchers such as Zhu et al., Kafy et al., Piyathilake et al., and González-García et al. have utilized the InVEST model to assess carbon stocks in ecosystems [13,14,15,16]. The InVEST model offers advantages such as requiring fewer data parameters with fast processing speeds and high model accuracy, which have contributed to its widespread application. The PLUS model, a novel model created by Liang et al. at China University of Geosciences, draws upon land use simulation models such as FLUS, CLUE-S, CA-Markov, and the random forest model. The model incorporates rule mining methods to examine land expansion and cellular automaton (CA) models with an incorporated stochastic seed mechanism. This can be used to analyze the land expansion factors and predict the evolution of land use changes at the patch level [17,18,19,20]. Many scholars have achieved significant research results using the PLUS model. Li et al. [21] calculated the ecosystem service values of land use changes under three scenarios using the PLUS model and the equivalent factor method. Wu et al. [22] applied the improved patch-based land use change simulation PLUS model, in combination with the FLUS model and the InVEST model, to simulate the carbon stock changes in the Chengdu–Chongqing Economic Zone under two different development scenarios: natural development and ecological protection. While the PLUS model exhibited higher simulation accuracy, the scenarios were relatively limited and did not consider the influence of policies. However, the majority of the current research on carbon storage and ecosystem spatial–temporal changes is based on past land use changes. There is a scarcity of research focused on simulating carbon storage levels within future regions, as they may be influenced by varying development scenarios. The PLUS model uses an algorithm based on random forests to derive the likelihood of land use transitions and to simulate and forecast the carbon storage conditions under various development scenarios, including the location, quantity, fine-scale cell alterations, and landscape pattern similarities. Hence, the present study integrates the InVEST model and the PLUS model to enhance the precision of future carbon storage projections in the research region [23,24,25].

Hunan Province is geographically situated in the central region of China and the middle reaches of the Yangtze River, which is part of the Yangtze River Economic Belt. It was one of the first areas in central China to implement the “two-oriented society” construction approach and is also an implementation area for the “belt and road” initiative. In recent years, it has gradually become an important growth pole in the strategy of central China’s rise. However, its industrial structure, with a heavy emphasis on heavy industry, and the energy structure, with obvious high-carbon characteristics, pose significant challenges to achieving the carbon emissions reduction target by 2030 [26]. In the course of intensifying efforts to protect the ecological environment, creating an ecological barrier, promoting ecological and green integration in Hunan Province and enhancing the green competitiveness of its cities, realizing the goals of the “belt and road initiative” and the strategy of prioritizing ecology and promoting green development, and assessing the current status of the existing resources and ecosystem services in Hunan Province are particularly crucial factors. Therefore, clarifying the current status of Hunan Province’s ecosystem resources and accurately evaluating its ecosystem service functions are urgent priorities. Carbon sequestration is an essential service provided by ecosystems [27,28]. The regulation of regional climates through this process exerts a profound influence on the ecological environment of Hunan Province. In order to maintain the environmental quality and promote the health of its ecosystems, carbon sequestration serves as a critical mechanism. As early as 2010, a pilot low-carbon economy plan was introduced in the Changsha–Zhuzhou–Xiangtan urban agglomeration in Hunan Province. In April 2016, the “Hunan Province Five-Year Action Plan for Low-Carbon Development (2016–2020)” explicitly stated that “we need to establish a low-carbon development system framework, with the carbon emission control system as the cornerstone and the establishment and improvement of the carbon emission trading system as the core,” but the forms of energy conservation and carbon reduction have remained severe [29,30]. This study is based on the PLUS model and the InVEST model, aiming to explore the relationship between land use changes and ecosystem carbon stocks. By simulating the changes in land use and ecosystem carbon stocks in Hunan Province under different scenarios, we analyze the characteristics of these changes and delve into the impacts of land use changes on ecosystem carbon stocks. The research findings are expected to provide valuable guidance and support for the conservation of ecosystems and the promotion of sustainable development in Hunan Province. By gaining a deep understanding of the mechanisms through which land use changes affect ecosystem carbon stocks, we can offer scientific evidence for formulating effective land management policies and implementing sustainable development strategies. Furthermore, our research contributes to revealing the potential impacts of land use changes on carbon cycling and climate change, thereby providing important insights for ecosystem restoration and environmental protection. In conclusion, this study holds significant importance for understanding the interplay between land use and ecosystem carbon stocks and driving sustainable development in Hunan Province.

2. Materials and Methods

2.1. Study Territory Overview

Hunan Province is geographically positioned between 24°38′ and 30°08′ N and 108°47′ and 114°15′ E. It is primarily situated in the southern region of Dongting Lake and is located in the middle section of the Yangtze River. As of the end of 2022, the permanent population was 66.04 million, with 39.83 million urban residents and an urbanization rate of 60.31%. Hunan Province has a terrain that is mostly encircled by mountains in the southeast, east, and west, with the central and northern portions being low and forming a basin in the shape of a horseshoe, which opens to the north. Its total land area is approximately 21.18 × 10⁴ km² (Figure 1). The study area lies within the subtropical monsoon climate zone, with distinct seasons, abundant light and heat, plentiful precipitation, concurrent rainfall, and high temperatures, making the climate conditions relatively favorable. The annual mean temperatures in the area range between 16 °C and 18 °C, while the annual precipitation averages range from 1200 to 1700 mm. Hunan Province is predominantly characterized by terrestrial ecosystems, with a rich endowment of natural resources, such as forests, grasslands, and wetlands, among others. These ecosystems are critical to the carbon cycle, serving to mitigate the accumulation of atmospheric concentrations of carbon [31].

2.2. Data Source

Land use data from three different years, namely 2000, 2010, and 2020, were collected for the study area. The land use types were categorized into six classes, including arable land, woodland, grassland, water bodies, construction land, and unused land. The unused land class refers to areas that are neither agricultural nor built-up, primarily consisting of fallow land, saline–alkali land, wetlands, sandy areas, bare soil, bare rock, and other similar categories. This classification system has been widely accepted and applied in the field of land use research. It adopts a hierarchical approach, dividing land use into primary categories, and provides a standardized framework for the classification and categorization of different types of land use. By utilizing this classification system, one can achieve consistency and comparability in the analysis of land uses across different regions and studies, facilitating meaningful comparisons and research. Adherence to this classification system ensures the accuracy and reliability of land use classifications, enabling the integration of our research findings with the existing literature and studies. These data were obtained from satellite imagery with a resolution of 30 m × 30 m (

Table 1. Data on land use drivers.

Data Type	Data Name	Year of Data	Data Accuracy/m	Data Source
Land Use Data	Land Use	2000, 2010 and 2021	30	Resource and Environmental Science and Data Center, Chinese Academy of Sciences (https://www.resdc.cn/ (accessed on 7 May 2023))
Climate and Environmental Data	Soil type	2018	30	National Earth System Science Data Center Soil Sub-Center (http://soil.geodata.cn/ (accessed on 9 May 2023))
	Average annual temperature	2020	30	National Weather Science Data Center (http://data.cma.cn/ (accessed on 9 May 2023))
	Average annual precipitation	2000–2020	30	National Weather Science Data Center (http://data.cma.cn/ (accessed on 9 May 2023))
	DEM Elevation	2020	30	Geospatial Data Cloud GDEMV3.30M resolution digital elevation data (http://gscloud.cn/home (accessed on 7 May 2023))
	Slope	2020	30	Calculated using DEM using ArcGIS to obtain
Socio-economic	GDP	2019	1000	Resource and Environmental Science and Data Center, Chinese Academy of Sciences (http://www.resdc.cn/ (accessed on 10 May 2023))
	Population	2020	1000	Resource and Environmental Science and Data Center, Chinese Academy of Sciences (http://www.resdc.cn/ (accessed on 10 May 2023))
	Distance to railroad	2020	30	Open street map (http://www.openhistoricalmap.org/ (accessed on 11 May 2023))
	Distance to highway	2020	30	Open street map (http://www.openhistoricalmap.org/ (accessed on 11 May 2023))
	Distance to primary urban roads	2020	30	Open street map (http://www.openhistoricalmap.org/ (accessed on 11 May 2023))
	Distance to urban secondary roads	2020	30	Open street map (http://www.openhistoricalmap.org/ (accessed on 11 May 2023))
	Distance to urban tertiary roads	2020	30	Open street map (http://www.openhistoricalmap.org/ (accessed on 11 May 2023))
	Distance to city hall premises	2020	30	National Geographic Information Resource Catalog Service System (http://www.webmap.cn/main.do (accessed on 11 May 2023))
	Distance to a river water body	2020	30	Calculated using DEM using ArcGIS to obtain

). Fourteen determinants of land use change were analyzed in this study, which comprised five environmental and climatic factors, as well as nine socio-economic factors (Table 1). The social-economic data reflect various aspects of social and economic development, such as the population quantity and density, economic development level, degree of urbanization, and infrastructure construction. Through a comprehensive analysis of their relationships with land use changes, a better understanding of the mechanisms and trends driving land use changes can be achieved. Additionally, these data provide guidance to researchers and decision-makers in formulating effective land management and planning strategies to achieve sustainable land use and urban development. For climatic and environmental data, the annual average temperature and annual average precipitation were obtained using the inverse distance weighting interpolation method from the National Meteorological Science Database. For socio-economic data, the distances to roads, county governments, and river bodies were calculated using the Euclidean distance analysis tool in ArcGIS (Figure 2). Calculating distances to infrastructure elements and creating Euclidean distance buffers were the activities primarily conducted to assess the spatial impact of infrastructure on land use changes. A Euclidean distance, a commonly used spatial metric, quantifies the spatial relationship and distance between different locations. Defining the influence range of the infrastructure through these buffers allows for the observation of its impact on the surrounding land use. Furthermore, when analyzing socio-economic data with varying units, standardization becomes important to ensure meaningful comparisons and analyses. Common standardization methods, such as z-score standardization and min-max scaling, transform variables to a common scale, enabling the integration and comprehensive analysis of the relative contributions of different variables to land use changes. This standardization process mitigates bias stemming from the original measurement units of the variables and enhances the reliability and comparability of the results.

2.3. Research Methodology

The research methodology in this study is depicted in Figure 3. Firstly, the PLUS model is employed to simulate multiple scenarios of land use in the year 2040 by integrating land use, natural, and socio-economic data. Subsequently, the InVEST model is used to assess and analyze the carbon stock variations in Hunan Province from 2000 to 2020 and in different scenarios for the year 2040. Furthermore, the impact of land use transitions on carbon stock changes is analyzed temporally, while the influence of factors such as the elevation, slope, and temperature on the spatial distribution of the carbon stock is explored spatially.

2.4. InVEST Carbon Module

InVEST employs a system to quantify the amount and value of ecosystem services, including watershed protection, soil and water conservation, habitat quality, and carbon storage services. The current uses of the InVEST model encompass a broad range of applications, such as ecosystem zoning, the delineation of ecological conservation redlines, and ecological restoration, as well as regional, national, and basin-level analyses of resource and environmental capacity, among others. InVEST provides a valuable tool for decision-making in protecting and managing our natural resources. The ecosystem carbon stock in the InVEST model comprises four components: aboveground biomass carbon (carbon in living vegetation on land), the belowground biomass carbon pool (carbon in live plant roots), the soil carbon pool (carbon in soil), and dead organic carbon (carbon in plant litter). The carbon storage rates of the four carbon reservoirs are calculated on a per-land-type basis using the land use classification. The carbon stock of the study area is estimated by multiplying the area of each land cover category by its corresponding carbon density and summing the results, resulting in a total carbon inventory of the region. The computation formula, as stated in [32], is expressed as follows:

$$C_i=C_{i,a}+C_{i,b}+C_{i,s}+C_{i,d}$$

(1)

$$C_t={\sum }_{i=1}^n{C}_i\times S_i$$

(2)

Equations (1) and (2) are used to calculate the total carbon stock, where $i$ symbolizes the type of land use. The variables $C_i$, $C_{i,a }$, $C_{i,b }$, $C_{i,s}$, and $C_{i,d }$ denote the carbon densities for the respective land use types, $i$.e., the cumulative carbon density, aboveground vegetation carbon density, belowground live root carbon density, soil carbon density, and vegetation litter carbon density (all measured in t·hm⁻²). The variable $C_t$ represents the cumulative carbon storage in $t$, while the variable $S_i$ represents the cumulative area of land use type $i$ in hm². The cumulative number of land use types, denoted by the variable $n$, is 6 in this study.

The carbon density data refer to [33,34,35,36,37,38], and priority is given to selecting measured or surveyed data from regions similar to or the same as the study area. For underground biotic carbon density values, the biomass conversion factor method is used, and the calculation formula is:

$$C_{i,below}=a\times b\times W_{D,i}$$

(3)

Formula (3) calculates the belowground biomass carbon density using the biomass conversion factor method. In the formula, $C_{i,below }$ represents the belowground biomass carbon density in t·hm⁻²; $i$ represents the land use type; $W_{D,i }$ represents the aboveground biomass of the $i$-th land use type in t·hm⁻²; $a$ is the conversion coefficient; and $b$ is the below-to-aboveground biomass ratio, with values of 0.2, 0.3, and 4.3 for arable land, woodland, and grassland, respectively. The formula for estimating the density of soil organic carbon can be expressed as follows:

$$Do{c}_i=SO{C}_i\times y_i\times H_i\times {10}^{-1}$$

(4)

The carbon density of the $i$ soil type in t·hm⁻² is represented by $Do{c}_{i }$ in Equation (4), while $SO{C}_i$ represents the soil organic carbon content of the $i$ soil type in g·kg⁻¹. The mean bulk density of the $i$ soil type, expressed in g·cm⁻³, is represented by $y_i$, and the soil layer thickness of the $i$ soil type, in cm, is represented by $H_{i }$. The dead carbon density is determined by adjusting values from earlier research results through a rainfall model, leading to the final carbon density values of land use types in Hunan Province (

Table 2. Carbon density values for each of the different types of land in Hunan Province.

Land Use Type	Aboveground Carbon Density	Subsurface Carbon Density	Soil Carbon Density	Dead Organic Carbon Density
Arable land	2.59	0.52	45.27	13
Woodland	30.31	8.23	98.30	1.36
Grassland	0.90	3.87	12.48	3.62
Water	0.11	0.55	1.21	0.50
Construction Land	11.29	2.26	17.97	0
Unused land	13.92	2.78	5.33	0

).

2.5. PLUS Model

The PLUS model is a high-resolution land utilization forecasting technique that employs the FLUS algorithm and can take policy-induced and guidance-induced effects into account. The PLUS model primarily comprises two modules: a transformation code processing module, which utilizes the land expansion analysis strategy (LEAS), and a cellular automata (CA) module based on a mega-type regular patch genes mechanism, referred to as the CA based on multi-type random patch seeds (CARS). The PLUS model presents a valuable approach for evaluating the impacts of numerous driving factors on alterations in land use. By employing this model, researchers are equipped with a systematic framework to comprehensively analyze the interplay between various influential elements and their respective contributions to land use transformations. This analytical tool provides a rigorous means of quantifying the effects and interactions of multiple driving factors, enabling a more nuanced understanding of the complex dynamics and processes governing land use changes. Furthermore, the PLUS model’s applicability extends beyond a single geographical region, as it facilitates the examination of driving factors and their effects in diverse contexts. Additionally, it offers the capability to generate alternative land use scenarios using methods such as Markov chains or linear regression, enhancing its utility in assessing and predicting land use dynamics in different areas [39].

2.5.1. Land Expansion Analysis Strategy (LEAS)

The LEAS is a methodology that analyzes land use data from multiple time periods and identifies the changing patterns of different categories of land utilization by considering the expansion areas of each type of changing land utilization. It can be used to define the characteristics of shifts in land utilization over a specific time period. The random forest classification (RFC) algorithm is applied to explore the relationships between the emergence of different categories of land utilization and various factors and to estimate the probabilities of development for each land use type. The mathematical expression presented in [40,41] is displayed below:

$$P_{i,k\left(X\right)}^d=\frac{\sum _{n=1}^{M}I[h_n\left(X\right)=d]}{M}$$

(5)

In Equation (5), the driving factors are represented by the input vector $X$, while the number of decision trees is denoted by $M$. The parameter $d$ can only take on binary values of either 0 or 1, with 1 indicating the possibility of converting other land use types to land use type $k$, while 0 means that such conversion is impossible. The function $h_n\left(X\right)$ predicts the land use type when the decision tree is $n$, while $I\left[h_n\left(X\right)=d\right]$ is the index function of the decision tree. Moreover, $P_{i,k\left(X\right) }^d$ represents the conditional probability of $k$-type land use growth at spatial unit $i$, given the driving factors $X$ and the conversion possibility $d$.

2.5.2. CA Model Based on Multi-Class Random Patch Seeding (CARS)

Cellular Automata with Random Segregation (CARS) is a type of computational model that replicates the geographic pattern of land use using the development probabilities of various land use categories. It is a scenario-based model for land use. The calculation of the total probability ($P_{0,i,k}^{d=1,t}$) of converting the land from other types to type $k$ is described by the following formula [40,41]:

$$P_{0,i,k}^{d=1,t}=P_{i,k}^{d=1}\times {\mathsf{\Omega }}_{i,k}^t\times D_k^t$$

(6)

Equation (6) presents a formal representation of the CARS model. Here, $P_{i,k}^{d=1}$ denotes the likelihood of developing the land use type $k$ in cell $i$, where ${\mathsf{\Omega }}_{i,k }^t$ indicates the surrounding effect of cell $i$ at time $t$ for land utilization type $k$, and $D_k^t$ represents the influence of future land use demand for land use type $k$ at time $t$. To compute the cumulative probability of conversion for land use type $k$, $P_{0,i,k}^{d=1,t}$, the formula is derived from this equation, as outlined in [40,41]:

$${\mathsf{\Omega }}_{i,k}^t=\frac{con(c_i^{t-1}=k)}{n\times n-1}\times W_k$$

(7)

$$D_k^t=\left\{\begin{array}{c}{D}_k^{t-1}(\left|G_k^{t-1}\right|\le \left|G_k^{t-2}\right|)\\ D_k^{t-1}\times \frac{G_k^{t-2}}{G_k^{t-1}}(G_k^{t-1}<G_k^{t-2}<0)\\ D_k^{t-1}\times \frac{G_k^{t-1}}{G_k^{t-2}}(0<G_k^{t-2}<G_k^{t-1})\end{array}\right.$$

(8)

The significance of Equations (7) and (8) can be understood as follows. Here, “$con$” refers to the entire number of grid cells that the $k$th land use type covered in the previous cycle over the $n\times n$ interval; “$W_k$” represents the weight assigned to various land use categories, which is typically set to 1; “$G_k^{t-1}$” and “$G_k^{t-2}$” represent, in the $t-1th$ and $t-2nd$ iterations, respectively, the variations between the present and potential demand for land use category $k$.

2.5.3. Related Parameter Settings

(1)

The LEAS model employs specific parameter settings, which are as follows. The default number of choice trees is 20 and the default sampling rate is 0.01. The maximum amount of variables tested at each node split (mTry) is set to the number of driving factors, which is 14. Additionally, the model runs with one parallel thread.

(2)

The parameter settings for CARS are as follows. The neighborhood range is set to the default value of 3, the thread is set to 1, the decay threshold coefficient is set to 0.5, the diffusion coefficient is set to 0.1, and the probability of the random patch seed is set to 0.0001.

(3)

In this analysis, four land use development models are established, including the natural development scenario, where land use trends continue without intervention; the urban development scenario, where restrictions are placed on the conversion of construction land to other land uses, with increased conversion from arable land, woodland, and water bodies to construction land; the arable land protection scenario, where arable land is protected and the conversion of arable land to other land uses is restricted; and the ecological protection scenario, which restricts the transformation of natural resources such as woodlands, grassland, and aquatic bodies to other land uses. Transition matrices for each scenario are provided in

Table 3. Multi-scenario transfer matrix settings.

Land Type	Natural Development Scenarios						Town Development Scenarios						Arable Land Conservation Scenarios						Ecological Conservation Scenarios
	a	b	c	d	e	f	a	b	c	d	e	f	a	b	c	d	e	f	a	b	c	d	e	f
a	1	0	1	0	1	1	1	0	0	0	1	1	1	0	0	0	0	0	1	1	1	1	1	1
b	0	1	1	0	0	1	1	1	0	0	1	1	1	1	1	0	0	1	0	1	0	0	0	0
c	1	1	1	1	1	1	1	0	1	0	1	0	1	1	1	1	1	1	0	1	1	1	0	0
d	1	0	1	1	0	1	0	1	1	1	1	0	1	0	1	1	0	1	0	0	0	1	0	0
e	0	0	0	0	1	0	0	1	1	0	1	0	0	0	0	0	1	0	0	0	0	0	1	0
f	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1

The six land use types are represented by variables a, b, c, d, e, and f, which respectively refer to arable land, woodland, grassland, water bodies, construction land, and unused land. A value of 1 indicates that the corresponding land use type can be converted, while a value of 0 indicates that it cannot be converted.

.

2.5.4. Accuracy Verification

The PLUS model was utilized to forecast the geographic pattern of land use in 2020 using land use information from 2000 and 2010. To assess the accuracy of the predicted results, actual land use information for 2020 was incorporated into the validation module of the PLUS model. The kappa and OA coefficients, which ranged from 0 to 1, were employed to evaluate the simulation accuracy, with values closer to 1 indicating higher precision. A threshold value of 0.75 was used to determine a high level of accuracy in the simulation. The resulting kappa score of 0.845 demonstrated a remarkably high level of precision in the projected land use outcomes.

3. Results

3.1. Analysis of Hunan Province’s Land Use Changes from 2000 to 2020

The land use transitions for each period were analyzed and are presented in Figure 4. It depicts the distribution of land use in Hunan Province from 2000 to 2010 and from 2010 to 2020, with woodland, arable land, grassland, water area, construction land, and unused land ranked in decreasing order based on their respective land area. As shown in Figure 4a, the period from 2000 to 2010 witnessed significant land use changes, particularly in the conversion of woodland, which accounted for a total area of 8.658 × 10³ km². Among these conversions, approximately 49.69% of the woodland was transformed into cultivated land, while the smallest conversion area occurred in unused land, with only 1.553 km². Figure 4b indicates that the conversion of woodland remained prominent during the period from 2010 to 2020, encompassing a total area of 9.489 × 10³ km². Within this period, the largest conversion rate occurred in cultivated land and grassland, accounting for roughly 46.98% and 35.42% of the total conversion area, respectively.

To comprehend the directional and structural characteristics of land use changes in Hunan Province, this study developed a land use transfer matrix (

Table 4. Land use transfer matrix for Hunan Province, 2000–2020, in km².

2000	2020
	Arable Land	Woodland	Grassland	Waters	Construction Land	Unused Land	Total Area	Transfer out
Arable land	55,900.45	5016.02	991.36	1111.33	2448.95	3.62	65,471.74	9571.29
Woodland	5222.29	110,982.44	4303.40	454.72	1004.27	8.84	121,975.98	10,993.54
Grassland	1625.85	3548.45	11,375.20	166.39	401.05	4.09	17,121.05	5745.85
Waters	1265.16	243.12	90.16	5519.98	72.18	0.53	7191.16	1671.18
Construction Land	251.91	71.78	22.29	38.55	2031.94	0	2416.48	384.54
Unused land	0.06	0.66	0.21	0	0.01	1.12	2.08	0.96
Total area	64,265.74	119,862.49	16,782.64	7290.99	5958.42	18.22	214,178.51
Transfer in	8365.29	8880.05	5407.44	1771.01	3926.48	17.1

) and a new land use distribution map (Figure 5). According to the table, woodland, arable land, and grassland are the major land use types in Hunan Province, accounting for approximately 95% of the total land area. During the time frame between 2000 and 2020, the construction land area increased by 3542 × 10³ km², and its proportion of the total land area rose from 1.13% to 2.7%. The rise in the amount of construction territory in Hunan Province was predominantly attributable to rapid urbanization and economic growth. The accelerated urban development led to a substantial enlargement of the construction land area, which in turn caused the conversion of a considerable amount of arable land to construction land. The newly added construction land was predominantly derived from arable land, and this transformation chiefly occurred in the peripheries, where the existing construction land underwent outward expansion. The arable land continued to decrease, with a decrease of 1.206 × 10³ km² during this period, and its proportion decreased from 30.57% to 30.01%. The dominant land use changes involving arable land were the conversions to woodland, construction land, and water area, representing 52.41%, 25.59%, and 11.61% of the total arable land area decrease, respectively. The conversion from arable land to woodland and water areas was primarily concentrated in the areas targeted for returning farmland to forest and wetland restoration projects, while the conversion to construction land was more prominent in economically developed regions of Hunan Province. The woodland area continued to decrease, mainly being converted to arable land, with a decrease of 2.113 × 10³ km² during this period, and its proportion decreased from 56.95% to 55.96%. The main reason for this was that the national policy of protecting forests and returning arable land was issued to adapt to the development of the times, which led to the most conversion from woodland to arable land. The grassland area decreased slightly, with a decrease of 3.384 × 10² km² during this period, and its proportion decreased from 7.99% to 7.84%. The water and unused land areas increased slightly, with areas of 7.191 × 10³ km² and 2.081 km² in 2000, respectively, increasing to 7.291 × 10³ km² and 18.220 km² in 2020, respectively.

During the period between 2000 and 2020, arable land, woodland, and grassland areas were the dominant land use types in Hunan Province. The principal process of environmental transition was a vast reduction in arable land and woodlands and an ongoing rise of unused land and construction land. The primary trend of land use change in Hunan Province was the transformation of arable land into woodlands, construction land, and water bodies, and the transition of woodlands into grasslands and arable land, as illustrated in Figure 5.

3.2. Characteristics of Spatial and Temporal Variation in Carbon Stocks in Hunan Province, 2000–2020

3.2.1. Carbon Stock Change Characteristics

Using the InVEST model’s carbon module, the province of Hunan’s carbon storage rates in 2000, 2010, and 2020 were computed. The carbon storage rates were 2.132 × 10⁹, 2.129 × 10⁹, and 2.098 × 10⁹ t, respectively (

Table 5. Land use carbon stocks in Hunan Province, 2000–2020.

Land Use Type	2000		2010		2020
	Carbon Storage Volume/Million t	Percentage/%	Carbon Storage Volume/Million t	Percentage/%	Carbon Storage Volume/Million t	Percentage/%
Arable land	40,188.18	18.84	40,106.27	18.83	39,272.87	18.71
Woodland	168,584.91	79.04	168,284.33	79.02	165,046.96	78.65
Grassland	3573.45	1.68	3618.93	1.70	3488.87	1.66
Waters	170.45	0.08	148.93	0.07	171.58	0.08
Construction Land	761.69	0.36	808.03	0.38	1864.61	0.89
Unused land	0.46	0.00	0.45	0.00	3.99	0.00
Total	213,279.14	100.00	212,966.95	100.00	209,848.89	100.00

). Between 2000 and 2020, there was a reduction in carbon storage in Hunan Province, which amounted to a total decrease of 3430.25 t. The carbon storage decreased by 312.19 t from 2000 to 2010 and by 3118.06 t from 2010 to 2020. The storage of carbon was ranked according to land use categories as follows: woodland, arable land, grassland, construction land, water area, and unused land. As a key node of the national central region growth plan and the comprehensive reform pilot zone of the “two-oriented society,” Hunan Province has a developed economy, high level of urbanization, high population density, and large-scale land development and utilization, resulting in a significant reduction in carbon storage due to the urban enlargement that has occurred over large amounts of woodland and arable land, which are the land use types with higher carbon storage.

3.2.2. Characteristics of the Spatial Variation in Hunan Province’s Carbon Stocks

According to Figure 6, Hunan Province’s western and southern mountainous regions are home to the majority of its regions with elevated carbon storage, which are dominated by forests and grasslands, with large areas of arable land in some regions, and have high soil carbon density. The distribution in space of carbon storage areas remained relatively stable from 2000 to 2020, with no significant changes observed, and the regions with higher carbon storage were mainly distributed in Zhangjiajie City, the western part of Xiangxi Autonomous Prefecture, the western part of Shaoyang City, the southwestern part of Huaihua City, the southeastern part of Chenzhou City, and the central and southern parts of Yongzhou City (Figure 7). These areas are mainly forests and grasslands, with higher vegetation coverage and strong carbon storage capacity. The low-value areas are mainly water bodies such as rivers and lakes, with the lowest value being 0. In most areas, the carbon storage did not change significantly. The areas where carbon storage decreased were mainly due to the enlargement of construction land areas and an increase in water bodies, while the areas where carbon storage increased were mainly due to the addition of forests, arable lands, and grasslands and were scattered.

3.3. Analysis of Land Use Carbon Stock Projections in Hunan Province

3.3.1. Multi-Scenario Land Use Change Analysis

Based on the PLUS approach, we made predictions about the dispersion of land use types in the year 2040 under four unique development scenarios, using 2020 as the reference year, as presented in Figure 8. The natural development scenario indicated that both arable land and grassland areas would experience a decreasing trend, with reductions of 7.514 × 10⁴ km² and 2.772 × 10⁴ km², respectively. Meanwhile, woodlands, water bodies, construction land, and unused land showed an increasing trend, increasing by 2.01 × 10⁴, 6.68 × 10², 8.041 × 10⁴, and 7.83 × 10² km², respectively. The primary transformation observed was the transformation of arable land to construction land, accompanied by the conversion of grasslands to woodlands and construction land. Under the arable land protection scenario, the arable land area increased significantly by 2.767 × 10⁵ km² due to restrictions on converting arable land to other land uses. The enlargement of the construction land was constrained to a significant extent, with an increase of only 1.892 × 103 km² compared to the year 2020. Conversely, the woodland, grassland, water, and unused land areas exhibited a decreasing trend, with reductions of 1.949 × 10⁵, 2.771 × 10⁴, 5.549 × 10⁴, and 4.06 × 10² km², respectively. Under the urban development scenario, construction land showed the most significant increase, increasing by 2.867 × 10⁵ km², with all land uses except water bodies being converted to construction land. There was a decrease in the areas of arable land, woodlands, and grasslands, decreasing by 6.492 × 10⁴, 1.949 × 10⁵, and 2.772 × 10⁴ km², respectively. Water bodies and unused land areas increased by 95.94 and 7.827 × 10² km², respectively. Under the ecological protection scenario, the highest decrease was observed in arable land, decreasing by 7.514 × 10⁴ km². This scenario adopts policies for protecting water sources, returning farmland to forests and grasslands, and encircling lakes with green belts. Woodlands and water bodies increased by 8.536 × 10⁴ and 5.494 × 10³ km², respectively. However, the enlargement of construction land was uncontrolled and exhibited an increase of 1.273 × 10⁴ km².

Drawing on the previous analysis, it can be concluded that of the four development scenarios, the only factor that substantially restricts the growth of construction land is the preservation of agricultural land, while the growth of construction land in the other three scenarios still maintains the previous trend and speed. In the ecological protection and natural development scenarios, the decreases in arable land is comparable. However, the enlargement of construction land is greater in the natural development scenario. Meanwhile, the ecological protection scenario shows an increase in areas with high ecological value, such as woodlands and water bodies. Thus, incorporating ecological protection measures into the scenario of protecting cultivated land can effectively regulate the enlargement of construction land and promote the conservation of the local ecological environment.

3.3.2. Multi-Scenario Carbon Stock Change Analysis

Using the PLUS model, land use changes in Hunan Province in 2040 were predicted under four scenarios, and carbon stocks under each scenario were calculated using the InVEST model and analyzed for changes (Figure 9). In comparison with the carbon stockpiles in 2020, the urban development and cultivated land protection scenarios reduced the total carbon stocks. In contrast, the ecological protection and natural development scenarios led to an increase in the total carbon stocks (

Table 6. Land use structure and carbon stock projections in Hunan Province in 2040.

Different Scenarios	Arable Land		Woodland		Grassland		Waters		Construction Land		Unused Land
	Area/ km2	Carbon Stock/ Million t	Area/ km2	Carbon Stock/ Million t	Area/ km2	Carbon Stock/ Million t	Area/ km2	Carbon Stock/ Million t	Area/ km2	Carbon Stock/ Million t	Area/ km2	Carbon Stock/ Million t
2020	63,983.18	39,272.87	119,426.16	165,046.96	16,717.17	3488.87	7239.86	171.58	5915.64	1864.61	18.11	3.99
A	63,330.01	38,874.02	117,463.13	162,355.05	16,437.97	3431.06	7239.74	171.61	8782.41	2768.27	25.93	5.71
B	66,745.83	40,970.66	117,463.13	162,355.05	16,437.97	3431.06	6683.79	158.43	5934.44	1870.59	14.04	3.09
C	63,227.88	38,811.34	120,265.90	166,228.49	16,437.97	3431.06	7293.72	172.89	6042.82	1904.75	10.91	2.40
D	63,227.88	38,811.34	119,622.32	165,339.07	16,437.97	3431.06	7245.46	171.74	6719.63	2118.08	25.93	5.71

). The ecological protection scenario yielded the greatest increase in carbon stock, amounting to 7.02 × 10⁶ t, owing to the implementation of effective measures to limit reductions in ecological land areas, which slowed down the urban expansion and greatly preserved and increased the woodland areas. Under the scenario for protecting cultivated land, the carbon stock decreased the least, by only 1.060 × 10⁷ t, primarily because the woodland area decreased. The urban development scenario had the most significant decrease in carbon stocks, by 2.243 × 10⁷ t, primarily because of reductions in cultivated land, grassland, and water areas. Under the natural development scenario, the total carbon stocks were not significantly different from those in 2020, and the total carbon stocks increased by 2.81 × 10⁵ t, mainly due to an increase in woodland areas compared to 2020. In conclusion, the decline in carbon stocks was considerably restrained under the ecological protection scenario, to a lesser extent under the natural development situation, and to the greatest extent under the urban development situation. Therefore, by 2040, some environmental safeguards based on natural development should be adopted to effectively limit the growth of construction land and halt the decline in carbon stocks.

In Table 6, A is the urban development scenario in 2040, B is the arable land conservation scenario in 2040, C is the ecological conservation scenario in 2040, and D is the natural development scenario in 2040.

4. Discussion

4.1. The Relationship between Carbon Stocks and Land Use Changes in Hunan Province

Land use change is recognized as a key driver of carbon stock variation in global terrestrial ecosystems, including Hunan Province. The government of Hunan Province has implemented various ecological conservation measures to protect the environment. Measures such as the 1985 “Implementation Regulations for the Management of Forests and Wildlife Nature Reserves in Hunan Province” aim to sustainably protect and utilize forest resources, promoting the restoration and maintenance of forest ecosystems and increasing carbon stocks. Additionally, ecological restoration projects, such as the “Conversion of Cropland to Forests and Grasslands Program” and wetland restoration projects, have been carried out from 2020 to 2023. These projects restore ecological functionality, enhance vegetation coverage, and increase carbon stocks. Urban greening regulations implemented since 2020 aim to expand green spaces, improve urban air quality, and increase urban carbon stocks. The implementation of these policies and measures positively influences the relationship between land use changes and carbon stocks in Hunan Province. They contribute to reducing carbon emissions, increasing carbon stocks, improving the ecological environment, and achieving sustainable development goals. However, the study findings show a declining trend in carbon stocks in Hunan Province from 2000 to 2020, reflecting changes in ecosystem services provided by the province.

In conclusion, the relationship between land use changes and carbon stocks in Hunan Province is a complex issue requiring the comprehensive consideration of natural factors, human activities, and management policies. Further research can deepen our understanding and provide a scientific basis for developing effective land use policies and carbon emission reduction measures. Moreover, this research offers valuable insights for other regions and countries in terms of land use management and carbon stock protection.

4.2. Impact Analysis of Carbon Stock Estimations in Hunan Province

The soil carbon storage in Hunan Province is subject to dynamic changes influenced by both climate change and human activities, due to the complexity and diversity of its natural environment. Using modeling simulation methods can help to better estimate carbon storage at large scales and provide scientific references to optimize the ecological spatial patterns under the “double carbon” goal [42]. The precision of the InVEST model’s carbon module estimates of ecosystem carbon storage is highly dependent on the accuracy of the carbon density data and the dependability of the land use data. Therefore, this study found that the carbon storage difference between each period was about 0.35 × 10⁶ t carbon by comparing 1 km and 30 m land use data, indicating that a more accurate assessment of ecosystem carbon storage requires more precise data to reduce errors. Regarding the carbon density data, the previous studies often applied correction formulas to estimate the carbon density in the research area based on national or the adjacent regions’ data. However, in this research, the carbon density statistics were obtained predominantly from previous researchers’ measurements of Hunan Province ecosystems, ensuring its reliability [43,44].

4.3. Existing Studies

In Hunan Province, there is a dearth of research examining the link between shifts in land use and the storage of carbon. The existing literature mainly focuses on forest carbon storage, such as the study by Liu et al. [45] on carbon retention, volume, and geographic distribution in forests in Hunan Province; and the study by Li et al. [46] on carbon storage and distribution in broad-leaved forest ecosystems. Liao et al. [47] analyzed the spatiotemporal differences and influencing factors of carbon outputs from land uses in Hunan Province from 2000 to 2018 but did not make predictions about future land uses and carbon emissions. Furthermore, carbon emissions and carbon storage have fundamentally different connotations, and the study area selected by the author is quite different from previous studies.

The InVEST model is used to calculate carbon sequestration, while the PLUS model simulates prospective future land use. To forecast the future retention of carbon and evaluate the influence of future zoning changes on future carbon storage, several development scenarios have been formulated. This study provides guidance for future land use planning in rapidly developing regions [48].

4.4. The Research Results’ Implications for Future Planning

By analyzing the land use change trends and the predicted carbon storage in 2040, the research reveals that the future carbon storage in Hunan Province is significantly affected by ecological protection policies and urban protection policies. The results of the carbon storage calculations over the years show that the greatest carbon reserve in the land use areas of Hunan Province is in woodland areas, followed by cultivated land. This highlights the critical importance of protecting forest carbon reservoirs and soil carbon reservoirs associated with cultivated land in order to maintain and enhance carbon storage in the province. It is important to note that the conversion of arable land to building land under the “ecological conservation scenario” is subject to stringent regulations and conditions. The current national spatial planning, particularly the implementation of the “three zones and three lines” approach, places significant emphasis on protecting lasting fundamental farmland and ecological resources. The designation of redlines for these areas aims to prevent the arbitrary conversion of arable land to non-agricultural uses and ensure the long-term preservation of vital resources [49,50]. However, further research is required to examine the specific delineation methodologies of the “three zones and three lines” and their impacts on localized carbon reservoirs. A comprehensive understanding of the interrelationships among land use policies, ecological conservation, and carbon storage is crucial for enhancing future planning strategies. This involves evaluating the sustainable conversion of arable land to building land within the designated zones and lines, while safeguarding carbon stores and the overall ecosystem’s health [51,52]. Therefore, while China’s conservation policy acknowledges the significance of ecological protection and sustainable land use, continuous research and evaluation are necessary to ensure that the conversion of arable land to building land under the “ecological conservation scenario” aligns effectively with the goals of preserving carbon stores and achieving long-term environmental sustainability.

4.5. Relevant Limitations of the Study

This research simulates the future carbon storage pattern by coupling the InVEST and PLUS models, which to some extent explains the trend of carbon storage variations. However, the simulation results have certain uncertainties due to the lack of careful consideration of the impacts of climate, socioeconomic, land planning, and ecological conservation redlines in future land use change simulations [53,54]. In addition, it should be noted that the carbon density data used in this analysis did not account for the interannual variability in carbon densities within distinct land categories; therefore, the variations in carbon storage were predominantly due to the conversion between land use classes. Therefore, although this study cannot comprehensively reflect the impacts of all factors on ecosystem carbon storage, it can reveal the trend of carbon storage changes and provide a scientific reference for sustainable development and ecological province construction in Hunan Province [55,56,57].

5. Conclusions

This study utilized the PLUS and InVEST algorithms to assess Hunan Province’s carbon storage from 2000 to 2020, and subsequently conducted simulations and analyses of land use and carbon storage scenarios for the year 2040. The following conclusions were drawn:

(1)

During the period from 2000 to 2020, arable land, woodland, and grassland areas were the dominant land use types in Hunan Province, comprising nearly 95% of the total land area. The main process of land use change was the significant decrease in arable land and woodland areas, while unused land and construction land continued to increase. During the study period, the predominant trends of land use change were the conversion of arable land to woodlands, the construction of land and water bodies, and the conversion of woodlands to grasslands and arable land;

(2)

During the period from 2000 to 2020, the carbon storage in Hunan Province experienced a downward trend, with a decrease of 312.19 t between 2000 and 2010, followed by a further decrease of 3118.06 t from 2010 to 2020. Woodland areas had the highest carbon storage rate, followed by arable land, grassland, construction land, aquatic, and unused land areas. The carbon accumulation in Hunan Province decreased primarily due to the conversion of woodland and agricultural land areas into construction land;

(3)

In the 2040 natural development scenario, arable land and grassland areas are projected to decrease, while woodland, aquatic, construction land, and unused land areas will continue to expand. In the arable land protection scenario, the conversion of arable land to other land uses was limited; the expansion of construction land was significantly controlled; and woodland, construction land, and grassland areas were converted to arable land, resulting in an increase in arable land. In the ecological protection situation, ecological territories including woodlands and lakes and rivers were safeguarded and their conversion to construction land was limited, resulting in an increase in ecological land and some control over construction land growth. In the urban development scenario, the growth of building sites and urban development was the primary driver, resulting in continuous losses of woodlands, arable land, and grasslands;

(4)

By 2040, the projected carbon storage rates showed different trends in the various scenarios. The total carbon storage decreased as a consequence of the urban development and arable land protection scenarios, while the ecological protection and natural development scenarios showed an increase in total carbon storage. The ecological protection scenario exhibited the most substantial increase in carbon storage, amounting to 7.02 × 10⁶ t, by safeguarding ecological land areas such as woodlands and water bodies and mitigating urban expansion pressures. The arable land protection scenario resulted in the smallest decrease in carbon storage, with a decrease of only 1.060 × 10⁷ t, indicating that protecting arable land can have a certain effect on controlling the decline in carbon storage. The urban development scenario resulted in the largest decrease in carbon storage, with a decrease of 2.243 × 10⁷ t, as the arable land, woodland, and grassland areas all decreased significantly. Due to the increase in woodland areas, the overall quantity of carbon kept in the natural development scenario was not substantially different from that in 2020, with a total increase of 2.81 × 10⁵ t. To maintain sustainable development, a combination of natural growth and ecological protection should be considered. Consequently, ecological and environmental preservation measures must be incorporated into the natural development plan to prevent the transformation of ecological land into building land, thereby controlling the expansion of construction land and increasing the regional carbon storage.