Spatiotemporal Variation and Driving Mechanisms of Carbon Budgets in Territorial Space for Typical Lake-Intensive Regions in China: A Case Study of the Dongting Lake Region

Abstract

As sensitive human-environment systems, lake-intensive regions are critical governance areas for advancing global low-carbon development. Rapid economic growth has intensified the imbalance between economic carbon sources and ecological carbon sinks in these regions. However, methods for measuring territorial space carbon budgets tailored to “production–living–ecological” functions are underdeveloped, and the mechanisms driving carbon imbalance risks remain unclear. To address these issues, this study develops a spatial measurement model for “carbon sources-carbon sinks” in the Dongting Lake region. Using exploratory spatiotemporal data analysis, this study identifies grid-scale variation patterns in carbon budgets. Finally, using the logarithmic mean Divisia index (LMDI) decomposition model, this study examines the driving mechanisms of carbon budgets from a territorial space perspective. The results indicate the following: (1) The territorial space of the Dongting Lake region follows a pattern where “ecological spaces surround production spaces, with living spaces interspersed among water network spaces”. Between 2005 and 2020, functional transitions primarily occurred between agricultural production spaces and forest or water ecological spaces. (2) The study area’s territorial space carbon budgets increased annually, though the growth rate slowed. Construction land was the most significant carbon emission source in territorial space. Spatially, carbon budgets exhibit a radial pattern, with high values concentrated in plains near water bodies, gradually decreasing inland. Spatiotemporal differentiation followed a north–south development trend along the water system axis. High-High clusters were concentrated in municipal areas with dense water networks. In contrast, Low-Low clusters appeared in peripheral mountainous regions to the west, east, and south. (3) Land-use efficiency had the most potent inhibitory effect on carbon budgets, cumulatively reducing carbon emissions by 1.37 × 10⁸ tC. Economic development had the strongest positive effect, adding 1.31 × 10⁸ tC in carbon emissions. Therefore, the Dongting Lake region should promote intensive land use, adjust industrial structures, and develop a green ecological economy to achieve sustainable carbon source–sink management.

1. Introduction

The atmospheric CO₂ concentration has risen sharply due to intensive industrial activities, making climate warming a critical global challenge [1]. To control CO₂ emissions, China aims to peak carbon emissions by 2030 and achieve carbon neutrality by 2060 [2]. Lake-intensive regions are areas where urban development is closely linked to rivers, lakes, and other water bodies. These regions show a more potent synergy and trade-off between human carbon sources and ecosystem carbon sinks than other areas [3]. In recent years, benefiting from developed water network ecosystems and convenient waterway transportation, lake-intensive regions have experienced rapid socioeconomic growth [4]. However, this development has intensified carbon reduction pressures in these watershed regions. At the same time, production and urban expansion have eroded blue-green ecological barriers, reducing the carbon sequestration potential of forests, wetlands, and aquatic ecosystems. Therefore, the conflict between socioeconomic carbon emissions and environmental carbon sequestration in territorial space intensifies, making the balance between protection and development increasingly prominent [5].

Researchers worldwide have extensively studied land-use carbon emissions, analyzing spatiotemporal patterns and underlying drivers to address imbalanced carbon dynamics in territorial spaces [6,7,8,9]. Developing a carbon budget assessment framework for lake-intensive regions is the first step in identifying carbon imbalance risks. Key land-use carbon emissions accounting methods include model simulation, plot inventory, and emission coefficient approaches. Model simulation results are susceptible to parameter selection, making this method suitable for large-scale carbon emissions studies [10]. The plot inventory method estimates carbon emissions by measuring carbon density in vegetation and soil across land types [11]. Although effective, this method is resource-intensive, making it more suitable for small-scale analyses. The emission coefficient method estimates carbon emissions using land-use remote sensing and human activity data [12]. Due to its simplicity and data accessibility, this method is widely used. For instance, Zhang applied a carbon source–sink coupling model, integrating diverse datasets, to calculate land-use carbon budgets across different scales in the Pearl River Basin [13]. Han analyzed the carbon source–sink balance in northwest China’s Bosten Lake region using a combined top-down and bottom-up approach, identifying areas with heightened carbon emission risks from land-use changes [14]. These diverse methodologies provide a quantitative foundation for examining carbon budget dynamics in lake-dense regions from multiple perspectives.

Exploring the evolution of carbon budgets in territorial space is crucial for lake-intensive regions to achieve carbon reduction and sequestration goals. Earlier studies on carbon budget evolution primarily analyzed trends in emissions and sequestration, proposing macro-level carbon balance strategies. For instance, Dai examined land-use carbon emission trends in Jiangxi Province, China, by integrating remote sensing data with statistical yearbook information [15]. Scholars have recently employed exploratory spatial data analysis (ESTDA) and regression models to investigate the mechanisms underlying carbon budget imbalances. For instance, Li combined ESTDA with vector autoregression models to reveal significant spatial variations in carbon emissions across the Guangdong–Hong Kong–Macao Greater Bay Area [16]. Similarly, Jin applied kernel density estimation and spatial autocorrelation methods to assess the spatiotemporal characteristics of carbon emission efficiency along the Yangtze River Economic Belt [17]. These visualization and analytical techniques advancements have significantly enhanced the understanding of spatiotemporal evolution trends in carbon budgets within lake-intensive regions.

Investigating the factors influencing carbon emissions in territorial space is an essential way to formulate emission reduction policies accurately. Regional social, economic, and historical development variations lead to differing determinants of land-use carbon emissions. Scholars frequently employ spatial econometric and factor decomposition models to examine the driving mechanisms behind these emissions. For example, Yuan applied the geographically and temporally weighted regression (GTWR) method, identifying positive correlations between carbon emissions and factors such as population size, economic development, land-use intensity, and landscape configurations [18]. Zhang used Geodetector and K-means clustering to analyze the drivers and clustering patterns of county-level carbon emissions in Shaanxi Province [19]. Meng employed the logarithmic mean Divisia index (LMDI) model to assess factors influencing land-use carbon emissions across nine provinces in the Yellow River Basin over 30 years [20]. Their findings indicated that population density, economic scale, and land-use structure contributed to emissions, with dominant factors varying by stage. In conclusion, a comprehensive understanding of the factors affecting land-use carbon emissions necessitates consideration of the distinct natural and economic characteristics of each region. This approach provides targeted scientific guidance for the low-carbon management of territorial space in lake-intensive regions.

The study of land-use carbon emissions has expanded significantly in recent years. (1) In spatial scales, studies have gradually shifted focus from national and provincial scales to urban and county scales [21,22,23,24]. However, the grid scale, the smallest geospatial research unit, has rarely been explored in land-use carbon budget studies. Additionally, research hotspots have focused on core cities or urban clusters undergoing rapid urbanization, such as Shanghai and Chongqing, with limited attention to the mechanisms driving carbon budget evolution in lake-adjacent regions with contiguous development. This highlights an urgent need to integrate grid and urban scales for carbon budget studies in lake-intensive regions. (2) In differentiation patterns, prior research on regional carbon emissions and sequestration has often neglected the spatiotemporal connections and clustering dynamics of carbon budgets within lake regions and between internal grid units. (3) In driving mechanisms, lake-dense regions pose unique analytical challenges due to their distinct land-use configurations. Unlike traditional urban risks stemming from cumulative pressures of expansion, these regions demand more nuanced analyses of the factors driving their carbon dynamics.

The Dongting Lake region serves as both an active carbon source and a supplementary carbon sink amid the rapid urbanization of the Yangtze River Economic Belt. Since the early 21st century, under the coordinated development strategy of the Yangtze River Economic Belt, energy industries have expanded rapidly along rivers and lakes, leveraging locational advantages [25]. However, this rapid economic growth has intensified conflicts with the ecological carrying capacity of the Dongting Lake region, undermining the ability of lake-intensive regions to achieve carbon reduction and sequestration goals. To address these challenges, this study builds on prior research by introducing innovations in research subjects, content, and methods. Regarding research subjects, this study focuses on lake-specific cross-regional areas, accounting for spatial heterogeneity in resources, industrial structures, and economic development across cities, and locates territorial carbon budget results at the smallest grid unit. Regarding research content and methods, a “micro-meso-macro” framework is proposed to analyze the spatiotemporal dynamics and driving mechanisms of territorial carbon budgets. At the microscale, a spatial measurement model for carbon emissions and sequestration is developed to calculate grid-level carbon budget totals for the Dongting Lake region over multiple periods. At the mesoscale, spatiotemporal evolution patterns and clustering effects of carbon budgets within watershed territorial spaces are analyzed using ESTDA methods, including standard deviation ellipse (SDE) and spatial autocorrelation analysis. At the macroscale, the Kaya identity and LMDI model are applied to investigate the driving mechanisms of carbon budgets, emphasizing harmonizing production–living–ecological functions. This study aims to provide empirical data to support achieving carbon balance in territorial space.

2. Materials and Methods

2.1. Study Area

The Dongting Lake region is a key geographical hub linking “nature and society”, “hydrology and humanity”, and “water and land”. Situated in the subtropical zones of Hunan and Hubei provinces, this region spans 71,279.8 km². It includes five cities: Jingzhou in Hubei, along with Yueyang, Changde, Yiyang, and Changsha in Hunan (Figure 1). As a cross-regional area, the Dongting Lake region shows significant heterogeneity in natural resource distribution and economic development across its cities and counties [26]. In recent years, expanding human activities into ecologically sensitive areas has exacerbated spatial imbalances in carbon budgets across regional cities. This study focuses on the Dongting Lake region, a representative lake-dense area in China, to examine the spatiotemporal dynamics and driving mechanisms of carbon budgets in territorial spaces, addressing these pressing challenges.

2.2. Data Source and Processing

This study primarily used multi-period remote sensing imagery and socioeconomic data. Land-cover data for the Dongting Lake region were sourced from the Resource and Environment Science Data Center of the Chinese Academy of Sciences (http://www.resdc.cn/, accessed on 12 March 2024), spanning four periods (2005, 2010, 2015, and 2020) with a 30 m spatial resolution [13]. Boundary vector data for 2020 were obtained from the National Geomatics Center of China (http://ngcc.sbsm.gov.cn, accessed on 12 March 2024). Socioeconomic data were obtained from the China City Statistical Yearbook (2005–2020), the China Urban Construction Statistical Yearbook, prefecture-level statistical yearbooks, and environmental status bulletins. For satellite image processing, Landsat TM/ETM imagery from four periods was analyzed using ENVI 15.6, applying a human–computer interactive interpretation method to extract land-use and land-cover information. Data with varying statistical criteria were standardized, and minor missing values were filled using linear interpolation or trend extrapolation.

2.3. Methodology

Figure 2 presents the methodological framework and technical approach of this study, encompassing four main aspects:

(1)

By combining multi-period remote sensing imagery with statistical yearbook data, this study classifies land-use categories within the territorial space of the Dongting Lake region and establishes a carbon source–sink indicator database.

(2)

The dynamic degree and transfer matrix models are applied to investigate the spatiotemporal evolution of the territorial space in the Dongting Lake region, identifying transition patterns across production, living, and ecological land-use categories.

(3)

Total carbon emissions and sequestration in the Dongting Lake region are estimated using the IPCC inventory’s direct and indirect estimation methods alongside the energy coefficient. The study examines carbon budget patterns at both “grid” and “city” scales, analyzing their spatiotemporal trends, spatial differentiation, and clustering effects.

(4)

The Kaya identity is utilized to decompose carbon emissions in territorial space into five influencing factors: carbon emission intensity, land-use structure, land-use efficiency, economic development, and population size. Using the additive model of the LMDI decomposition method, the cumulative contributions of each factor to carbon emissions are quantified, providing insights into the intrinsic relationships between land-use functions and carbon budgets.

2.3.1. Territorial Space Carbon Source–Sink Indicator Classification System

Land use is the foundation of territorial space development, integrating carbon sources and sinks in carbon emissions and sequestration. Carbon sources release carbon into the atmosphere through industrial energy use, urban construction, and agricultural production. Carbon sinks lower atmospheric carbon concentrations by sequestering carbon in forests, water bodies, wetlands, and grasslands. Following the functional zoning principle of production–living–ecological areas and lake-intensive regions’ carbon reduction and sequestration needs, this study develops a carbon source–carbon sink indicator classification system for the Dongting Lake region’s territorial space, as shown in

Table 1. Territorial space carbon source–sink indicator classification system.

Territorial Space Carbon Source–Sink Indicator Classification System			Corresponding Territorial Space Carbon Emission–Carbon Sequestration Activities
Territorial Space Carbon Source–Sink Categories		Land Use Type (LUCC Secondary Classification Standard)
Carbon sources: production–living space	Agricultural production space	Paddy fields, dry farmland	Carbon emissions from agricultural production processes
	Industrial-mining production space	Industrial-mining construction land	Carbon emissions from energy consumption
	Agricultural living space	Rural residential areas
	Urban living space	Urban construction land
Carbon sinks: ecological space	Forest ecological space	Forest land, shrubland, sparse forest land, other forest land	Carbon sequestration from forests
	Grassland ecological space	High-coverage grassland, medium-coverage grassland, low-coverage grassland	Carbon sequestration from grasslands
	Water ecological space	Rivers, lakes, reservoirs, ponds, tidal flats, beach land	Carbon sequestration from water bodies
	Potential ecological space	Wetlands, bare land, bare rock areas	Carbon sequestration from unused land

.

2.3.2. Territorial Space Dynamic Degree and Transfer Matrix

The dynamic degree of a single land-use type describes changes in the extent of a specific land-use type within a designated study area over a specific time period [27]. The formula is as follows:

$$F={\displaystyle \frac{\left(S_{i, t2}-S_{i, t1}\right)}{S_{i, t1}}}\times {\displaystyle \frac{1}{T}}\times 100\%$$

(1)

where $t1$ and $t2$ represent the area of the land-use type at the start and end of the study period, respectively; $S$ represents the total area; $T$ represents the duration; $S_i$ refers explicitly to the unchanged area of the i^th land-use type during the study period.

The land-use transfer matrix illustrates the evolution trends and interrelationships among different land-use categories, capturing the spatiotemporal patterns of land-use changes. This matrix includes two dimensions: rows represent land-use types from the initial period, while columns correspond to land-use types in the subsequent period. Each matrix element indicates the probability of transitioning from one land-use type to another over the study period [28]. The formula is as follows:

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

(2)

where S represents the area of production, living, and ecological land-use categories; n is the total number of land-use types; i and j denote the land-use types during the study period.

2.3.3. Total Accounting of Carbon Budgets in Territorial Space

The calculation of territorial carbon budgets depends on the difference between carbon emissions and carbon sequestration. This study uses direct and indirect estimation methods from the IPCC inventory to evaluate carbon emissions and sequestration across production–living–ecological spaces in the Dongting Lake region. Based on the IPCC inventory, the study incorporates multi-temporal and multi-source remote sensing data and statistical yearbooks to ensure accuracy and timeliness. To improve scientific rigor, carbon emission factors and sequestration coefficients are derived from authoritative literature and model data, with adaptive adjustments for the ecological characteristics of lake-dense regions. A systematic analysis of carbon source and sink characteristics across territorial space types ensures a comprehensive and accurate carbon budget calculation. The specific models for carbon emissions and sequestration are outlined below:

(1) Territorial space carbon emission estimation model

Carbon emissions in territorial space primarily focus on carbon sources in production and living spaces. This study considers farmland carbon emissions as non-construction land carbon emissions and fossil energy consumption emissions as construction land carbon emissions. Farmland carbon emissions are estimated using the direct estimation method. Based on Yu’s research [29], the carbon emission coefficient is set at 0.4970 tC/hm²·a. Fossil energy consumption emissions are estimated using the indirect estimation method, considering the consumption of the top 10 fossil fuels [30,31]. Carbon content, average lower heating values, and carbon oxidation rates are based on the IPCC [32] guidelines and the General Principles for Calculating Comprehensive Energy Consumption (

Table 2. Carbon emission coefficients for major energy types in the Dongting Lake region.

Energy Type	Carbon Emission Factor (tC/tce·a)
Raw coal	0.5127
Washed coal	0.6292
Coke	0.7801
Natural gas	0.5390
Crude oil	0.8237
Gasoline	0.7977
Kerosene	0.8273
Diesel	0.8443
Fuel oil	0.8647
Liquefied petroleum gas	0.8458

). The formulas are as follows:

$$C_e=C_a+C_{fue}$$

(3)

$$C_a=S\times \gamma$$

(4)

$$C_{fue}=\sum _{n=1}^{10}{E}_n\times b_n\times {\beta }_n$$

(5)

where $C_e$ is the total carbon emissions (t); $C_a$ is the carbon emissions from farmland (t); $C_{fue}$ is the carbon emissions from fossil energy consumption (t); $S$ is the farmland area (hm²); $\gamma$ is the farmland carbon emission coefficient (tC/hm²·a); $E_n$ is the consumption of fossil fuel type n (t); $b_n$ is the conversion factor for fuel type n to standard coal (tce/t or tce/10³·m³); ${\beta }_n$ is the carbon emission coefficient for fuel type n (tC/tce).

(2) Territorial space carbon sequestration estimation model

Carbon sequestration in territorial space primarily focuses on carbon sinks in ecological spaces. This study employs the direct estimation method to calculate total carbon sequestration by summing the contributions of forests, grasslands, water bodies, and unused land within ecological spaces. The carbon sequestration coefficients are based on existing research and are detailed in

Table 3. Carbon sequestration coefficients of territorial space types in the Dongting Lake region.

Territorial Space Carbon Sink Type	Carbon Sequestration Coefficient (tC/hm2·a)
Forest ecological land	−0.0586
Grassland ecological land	−0.0210
Water ecological land	−0.0459
Potential ecological land	−0.0005

[29]. The formula for estimating total carbon sequestration is as follows:

$$C_s=\sum _{m=1}{e}_m=\sum _{m=1}{T}_m\times {\delta }_m$$

(6)

where $C_s$ is the total carbon sequestration of territorial space (t); $e_m$ is the carbon sequestration of land type m (t); $T_m$ is the area of land type m functioning as a carbon sink (hm²); ${\delta }_m$ is the carbon sequestration coefficient of land type m (tC/hm²).

2.3.4. Exploring Spatiotemporal Variation Patterns of Carbon Budgets

(1) Grid sampling method

The grid sampling method partitions the study area into equal-sized grids, selecting one sample point per grid. This method effectively covers the study area while ensuring a uniform distribution of sample points [33]. To visualize spatiotemporal carbon budget patterns, this study applies a grid-based approach to the Dongting Lake region. The carbon emission coefficient for construction land is 8.3102 kg(C)·m⁻²·a⁻¹. Using the grid sampling method from landscape ecology to capture spatial patterns and considering the Dongting Lake region’s size, this study employs 5 km × 5 km grids, generating 3087 sample units. Carbon sequestration, emissions, and budgets for each grid are assigned to its central point, and Kriging interpolation is used to visualize the spatial evolution of the Dongting Lake region. The ordinary Kriging interpolation formula is as follows:

$$\widehat{Z}(S_0)=\sum _{i=1}^n{\lambda }_{i}Z(S_i)$$

(7)

where $Z\left(S_i\right)$ is the measured value at location i; ${\lambda }_i$ is the unknown weight assigned to the measured value at location i; and $n$ is the number of measured values.

(2) SDE analysis method

The SDE analysis method, initially proposed by Lefever [34], has been refined over time and is now widely used to study the spatial distribution of geographic phenomena. It accurately depicts the primary directional distribution of carbon budgets and quantitatively describes spatiotemporal evolution trends. In the SDE analysis, the ellipse’s long axis represents the dominant directional distribution of carbon budgets, while the short axis reflects the degree of dispersion or clustering. The centroid coordinates of the ellipse indicate the trajectory of the carbon budget’s spatial center over time. The formula for calculating the centroid coordinates is as follows:

$$\overline{X}={\displaystyle \frac{1}{n}}\sum _{i=1}^n{x}_i,\overline{Y}={\displaystyle \frac{1}{n}}\sum _{i=1}^n{y}_i$$

(8)

where $x_i$ and $y_i$ are the center coordinates of each administrative unit in the study area; $\overline{X}$ and $\overline{Y}$ are the centroid coordinates; and $n$ is the total number of grids.

(3) Spatial autocorrelation analysis

Spatial autocorrelation is widely used to examine geographic phenomena’ spatial distribution patterns and relationships [35]. This study uses the grid sampling method to generate data on carbon sinks, carbon sources, and net carbon emissions. Global (Moran’s I) and local (Local Moran’s I) spatial autocorrelation are applied to evaluate the spatial correlation of carbon budgets within the territorial space [36]. The formulas for global and local Moran’s I indices are as follows:

$$I={\displaystyle \frac{n{\sum }_{i=1}^n{\sum }_{j=1}^n{w}_{ij}(x_i-\overline{x})(x_j-\overline{x})}{\left({\sum }_{i=1}^n{\sum }_{j=1}^n{w}_{ij}\right){\sum }_{i=1}^n{(x_i-\overline{x})}^2}}$$

(9)

$$I_L={\displaystyle \frac{n(x_i-\overline{x}){\sum }_{j=1}^n{w}_{ij}(x_j-\overline{x})}{{\sum }_{i=1}^n{(x_i-\overline{x})}^2}}$$

(10)

where $I$ is the Moran’s I index; n is the total number of grid cells; $w_{ij}$ is the spatial weight matrix; $x_i$ and $x_j$ are the carbon emissions of the ith and jth grid cells, respectively; $\overline{x}$ is the mean carbon emissions. The I value ranges from −1 to 1. An III value less than 0 indicates a negative correlation, while an $I$ value greater than 0 indicates a positive correlation. $I_L$ is the Local Moran’s I index, which classifies regions into five types: no significant clustering, High-High (H-H) clustering, High-Low (H-L) clustering, Low-High (L-H) clustering, and Low-Low (L-L) clustering.

2.3.5. Decomposition of Driving Factors of Carbon Budgets in Territorial Space

The LMDI method is a widely used analytical tool for evaluating and decomposing the effects of complex factors, particularly in energy consumption studies [30]. The model decomposes total energy consumption into contributions from selected factors, enabling the identification of primary and secondary influences [37]. This study applies the LMDI model from a territorial space perspective, expanding the Kaya identity. The selected factors include carbon emission intensity per unit of territorial space ($R_i$), land-use structure ($S_i$), land area per unit GDP ($\gamma$), per capita GDP ($g$), and population size ($P$). The formula is as follows:

$$C=\sum R_i\times S_i\times \gamma \times g\times P=\sum {\displaystyle \frac{C_i}{L_i}}\times {\displaystyle \frac{L_i}{L}}\times {\displaystyle \frac{L}{G}}\times {\displaystyle \frac{G}{P}}\times P$$

(11)

where $C$ represents the total carbon budget from land use; the formulas for $R_i$, $S_i$, $\gamma$, $g$, and $P$ are shown in

Table 4. Formula for the driving factors of carbon emissions in territorial space.

Driving Factor	Formula	Unit
Carbon emission intensity per unit of territorial space ($R_i$)	Ci/Li (carbon emissions per unit of land)	t/ha
Land-use structure ($S_i$)	Li/L (proportion of land use)	ha/ha
Land area per unit GDP ($\mathsf{\gamma }$)	L/G (land area per unit GDP)	ha/10,000 yuan
Per capita GDP ($\mathrm{g}$)	G/P (per capita GDP)	yuan/person
Population size (P)	-	10,000 people

.

The LMDI additive model is employed to decompose the contributions of various driving factors to changes in territorial space carbon budgets. The formulas are as follows:

$$\Delta C=C^T-C^0=\sum _{i=1\cdot \cdot \cdot 2}{R}_i^T\times S_i^T\times {\gamma }^T\times g^T\times P^T-\sum _{i=1\cdot \cdot \cdot 2}{R}_i^0\times S_i^0\times {\gamma }^0\times g^0\times P^0=\Delta C_{Ri}+\Delta C_{Si}+\Delta C_{\gamma }+\Delta C_g+\Delta C_P$$

(12)

$$\Delta C_{Ri}=\sum {\displaystyle \frac{C_i^T-C_i^0}{\mathit{ln}{C}_i^T-\mathit{ln}{C}_i^0}}\times \mathit{ln}{\displaystyle \frac{R_i^T}{R_i^0}}$$

(13)

$$\Delta C_{Si}=\sum {\displaystyle \frac{C_i^T-C_i^0}{\mathit{ln}{C}_i^T-\mathit{ln}{C}_i^0}}\times \mathit{ln}{\displaystyle \frac{S_i^T}{S_i^0}}$$

(14)

$$\Delta C_{\gamma }=\sum {\displaystyle \frac{C_i^T-C_i^0}{\mathit{ln}{C}_i^T-\mathit{ln}{C}_i^0}}\times \mathit{ln}{\displaystyle \frac{{\gamma }^T}{{\gamma }^0}}$$

(15)

$$\Delta C_g=\sum {\displaystyle \frac{C_i^T-C_i^0}{\mathit{ln}{C}_i^T-\mathit{ln}{C}_i^0}}\times \mathit{ln}{\displaystyle \frac{g^T}{g^0}}$$

(16)

$$\Delta C_P=\sum {\displaystyle \frac{C_i^T-C_i^0}{\mathit{ln}{C}_i^T-\mathit{ln}{C}_i^0}}\times \mathit{ln}{\displaystyle \frac{P^T}{P^0}}$$

(17)

where $\Delta C_{Ri}$, $\Delta C_{Si}$, $\Delta C_{\gamma }$, $\Delta C_g$, and $\Delta C_P$ represent the decomposition effects of the driving factors ($R_i$, $S_i$, $\gamma$, $g$, $P$) from the base period to period t, corresponding to the effects of territorial carbon emission intensity, territorial land-use structure, territorial land-use efficiency, economic development, and population size, respectively.

3. Results

3.1. Spatiotemporal Dynamics of Territorial Space

3.1.1. Spatiotemporal Evolution of Territorial Space

Table 5. Numerical development trends of territorial space from 2005 to 2020.

Year (km2) Period (%)	Production Space		Living Space		Ecological Space
	Agricultural Production Space	Industrial-Mining Production Space	Agricultural Living Space	Urban Living Space	Forest Ecological Space	Grassland Ecological Space	Water Ecological Space	Potential Ecological Space
2005	30266.89	177.09	1270.95	643.76	29126.38	1036.56	7873.81	848.46
2010	29527.30	389.29	1259.85	958.77	29065.38	980.12	7919.43	1143.94
2015	29312.07	687.47	1259.92	984.43	28961.47	975.65	7919.17	1143.36
2020	29320.22	1034.54	1289.27	994.39	28793.33	974.60	7711.35	1120.79
2005–2010	−0.49	23.97	−0.17	9.79	−0.04	−1.09	0.12	6.97
2010–2015	−0.15	15.32	0.00	0.54	−0.07	−0.09	0.00	−0.01
2015–2020	0.01	10.10	0.47	0.20	−0.12	−0.02	−0.52	−0.39
2005–2020	−0.21	32.28	0.10	3.63	−0.08	−0.40	−0.14	2.14

outlines the evolving trends in territorial space utilization within the Dongting Lake region. From 2005 to 2020, the most pronounced transformations occurred in production space, while living space expanded considerably, and ecological space demonstrated relatively minor variability. Detailed examination of land-use types reveals the following: (1) Production space had an average share of 42.36% over the years. Agricultural production space declined with a dynamic rate of –0.21% during the study period. Industrial-mining production space exhibited exponential growth, with a dynamic degree of 32.28%, significantly surpassing other land-use types. These findings underscore the abundant natural resources in the region, supporting agricultural output and the growth of industrial and mining sectors. (2) Living space had an average share of 3.04% over the years. It expanded by 368.96 km² during the study period. Agricultural living space exhibited minimal change. The dynamic degree of urban living space dropped from 9.79% to 0.20%, reflecting reduced construction intensity as urbanization approached saturation. (3) Ecological space, the dominant land-use category, covered 54.60% of the total area. Both forest and grassland ecological spaces showed declining trends. Forest land dynamics increased by 177.21% during the period 2015–2020 compared to 2005–2010, while grassland dynamics declined by 98.03% in the latter period relative to the earlier one. Water ecological space fluctuated, first increasing and then declining, highlighting its instability in lake-intensive regions. Potential ecological space, often designated for future construction, expanded by 32.10% over the study period, with a dynamic degree of 2.14%.

To better analyze territorial space evolution trends, this study uses maps to visualize the data from Table 5. As depicted in Figure 3, the left panel illustrates the spatiotemporal evolution characteristics of primary land-use categories. The enlarged sections on the right provide a detailed view of the distribution patterns for secondary land-use types. The results indicate that between 2005 and 2020, the Dongting Lake region displayed a “U”-shaped spatial structure, where ecological space surrounds production and living spaces. (1) Agricultural production space is mainly distributed in the lakeside plains, shaped by river network erosion, in the central and northern areas. During the study period, this category declined significantly in the lake-intensive zones of Jingzhou. In contrast, industrial-mining production space expanded continuously in suburban areas near the administrative centers of the five cities, taking advantage of urban construction land. This reflects the increasing demand for industrial-mining construction activities driven by urbanization. (2) Urban living space is clustered around the administrative centers of the five cities. Agricultural living space is dispersed in areas near rivers with favorable water and thermal conditions. Notably, urban living space in Changsha’s Xiang River waterfront area exhibits a clear pattern of radial expansion, emphasizing an intensified waterfront clustering pattern as human activities increase. (3) Forest ecological space is primarily located in the peripheral hilly regions of Yueyang, Changde, Yiyang, and Changsha, acting as a crucial ecological buffer. Water ecological space is concentrated in the core area of Dongting Lake, mainly in the western parts of Yueyang and the northern plains. The observed patterns suggest that ongoing urbanization threatens water-centered ecosystems by exacerbating water pollution, degrading ecological habitats, and increasing carbon emissions. These impacts may hinder the coordination and development of territorial space.

3.1.2. Land-Use Category Conversion in Territorial Space

Figure 4a presents the numerical characteristics of land-use category conversions in the Dongting Lake region. Over the study period, these conversions primarily involved the replacement of agricultural production space with production–living construction space (904.80 km²), the conversion of water ecological space into agricultural production space (445.02 km²), and the transition of agricultural production space into water ecological space (527.52 km²). These changes underscore the complex trade-offs between economic development, population density, and water ecological systems in lake regions. By stages as follows: (1) From 2005 to 2010, the primary land-use conversion involved agricultural production space into water ecological space, totaling 683.66 km². (2) From 2010 to 2015, the focus of land-use change shifted toward a sustained transition from agricultural production space to industrial-mining production space. (3) From 2015 to 2020, the cumulative conversion of agricultural production space into industrial-mining production space increased further, reaching 260.16 km². This trend reflects the increasing intensity of land-use category changes in the Dongting Lake region, driven by the rapid socioeconomic growth of the Yangtze River Economic Belt. During the same period, large areas of agricultural production space were converted into forest ecological space, highlighting the impact of reforestation policies in the later stages of urbanization.

Figure 4b provides a more intuitive depiction of the spatial characteristics of land-use category conversions. During the study period, these changes were concentrated in three key areas: the northeastern plains of Jingzhou, the administrative regions of the five cities, and the primary lake body in western Yueyang. By stages as follows: (1) From 2005 to 2010, agricultural production space was predominantly transformed into water ecological space in regions with extensive lake and river networks, such as the central Jianghan Plain in Jingzhou and the western areas of the eastern Hubei hills. This trend was driven by converting low-lying farmland in the Jianghan Plain into aquaculture ponds, which persisted until 2010. (2) Since 2010, accelerated urbanization has intensified land-use development in major lake basins. Urban and rural living spaces expanded outward from the urban cores of the five cities, leading to the formation of industrial-mining production chains in suburban areas. This expansion often encroached on farmland and forest land with lower economic returns. (3) From 2015 to 2020, construction land continued to expand along water bodies within the administrative boundaries of each city. Meanwhile, water ecological space near lakes in Jingzhou and Yueyang was converted into agricultural production space, highlighting the ongoing severity of “reclaiming land from lakes” in the Dongting Lake region. Overall, water ecological space in lake regions faces dual pressures from agricultural expansion and urban construction. The conflict between carbon emissions from urban development and carbon sequestration through ecological conservation remains an ongoing challenge.

3.2. Spatiotemporal Variation of Carbon Budgets in Territorial Space

3.2.1. Spatiotemporal Evolution of Carbon Budgets in Territorial Space

Building on the analysis of territorial space evolution mechanisms, this section examines the spatiotemporal trends of carbon emissions, carbon sequestration, and carbon budgets. Figure 5 presents the numerical trends for the study area and individual cities. Overall, the region’s territorial space carbon budgets exhibited an increasing trend, though the growth slowed between 2015 and 2020. Detailed results show the following:

(1) Regarding land-use function categories, carbon emissions from production and living spaces increased between 2005 and 2020, with the highest growth occurring from 2005 to 2010 at an annual rate of 3.4%. Although the growth rate slowed to 0.69% between 2010 and 2015, total emissions increased by 2.78 × 10⁶ tC during this period. Carbon sequestration in ecological spaces declined annually, decreasing from 7.68 × 10⁶ tC in 2005 to 7.58 × 10⁶ tC in 2020, with an annual reduction rate of -0.19% between 2015 and 2020. The carbon budget trend followed carbon emissions, exhibiting a strong total growth rate of 19.13% between 2005 and 2010. However, from 2015 to 2020, the carbon budget growth rate slowed to 0.10%, indicating a converging trend.

(2) Regarding regional differences, Jingzhou’s territorial space carbon budget has increased. Between 2015 and 2020, the most significant changes occurred, with carbon emissions and sequestration rates of 9.22% and −1.77%, respectively. Changde’s territorial space carbon budget exhibited an overall declining yet fluctuating trend. Carbon emissions varied, with rates of 12.52%, −5.74%, and 1.01% across different periods, while carbon sequestration remained relatively stable. Yueyang contributed the largest share of carbon budgets in the Dongting Lake region, accounting for 53.43%, 42.20%, 56.99%, and 56.59% between 2005 and 2020. The carbon budgets in Yiyang and Changsha peaked in 2010 before declining from 2.23 × 10⁷ tC to 1.24 × 10⁷ tC. Notably, most cities experienced significant growth between 2005 and 2010, but growth rates approached zero after 2010. This trend reflects the initial emphasis on economic growth, which drove substantial carbon emissions through increased energy consumption. This shift aligns with the transition in development priorities since 2010 from “extensive development” to “intensive and efficient utilization”. However, the ecological carbon sequestration capacity of the Dongting Lake region remains weak relative to socioeconomic carbon emissions. The high baseline of net carbon emissions poses significant challenges for transitioning to a low-carbon and green economy.

Carbon emissions, sequestration, and budgets across the five cities were analyzed through the ArcGIS map visualization approach, categorizing the data into five levels (I–V) from minimal to maximal. Figure 6 illustrates the spatial patterns and temporal trends of carbon budgets during the study period. The results show the following:

(1) Carbon emissions from production and living spaces exhibit a radial “core–periphery” pattern, with elevated values concentrated in plains near water bodies, decreasing as distance from water sources increases. This pattern is related to clustering urban construction land and farmland along water bodies. The carbon emissions pattern was influenced by economic development, with peak values increasing year by year during the study period. Regarding individual cities, Jingzhou’s carbon sources cover a broad area, with level V regions expanding southward yearly. This trend is intricately linked to the southward relocation of Jingzhou’s economic hub across the river. In Changde, regions classified as level III and above are predominantly found in the flat eastern areas of Wuling District, Dingcheng District, and Lixian County. In Yiyang, carbon emissions are high in the northeastern Ziyang and Heshan Districts, while values are lower in the peripheral mountain areas farther from water bodies. Yueyang’s central urban area mainly falls into levels IV and V. Over the past decade, agricultural production bases in Huarong County have shifted from traditional methods to modern processing technologies, gradually increasing carbon emissions and forming large-scale carbon-source farming areas. With the most concentrated carbon sources, Changsha shows a high-value core along both sides of the Xiang River, radiating outward throughout the study period.

(2) Carbon sequestration patterns within ecological spaces demonstrate distinct spatiotemporal variations, with peripheral areas in the south, west, and east consistently showing higher values than the central regions. The spatial extent of the carbon sequestration pattern remained essentially unchanged during the study period. Regarding individual cities, Jingzhou has the smallest carbon sequestration area among the five cities, with only a tiny proportion of level III and IV regions distributed along the mountain edges. In Changde, high carbon sequestration values are found in the southwestern and northwestern areas, with a few high-value areas in Linli County. In Yueyang, extensive forest and grassland areas concentrated in the southeast exhibit high carbon sequestration. The Dongting Lake water body in the northwest also exhibits high carbon sequestration. In Yiyang, high-value areas are concentrated in southern Anhua County, where forest coverage reaches 76.17%, designating it as a key national ecological function county. In Changsha, carbon sequestration is distributed in patches across the southeastern Xiang Mountain ranges (Dawushan, Luoxiao, and the Wuling ridges) and the northwestern hills (Weishan and Huitang Mountains).

(3) The carbon budget distribution across territorial space closely follows the carbon emissions pattern, exhibiting a U-shaped distribution, with low-value areas surrounding high-value areas. At the city level, Jingzhou’s level IV carbon budget expanded rapidly southward during the study period, with low-value areas remaining minimal. In Changde, levels III and IV carbon budgets are concentrated in construction areas with dense water networks, with high values emerging between 2015 and 2020. The distribution of low-value areas remained essentially unchanged. Yueyang’s levels II to IV carbon budgets expanded rapidly between 2005 and 2010, but growth slowed from 2010 to 2020. Additionally, level III expanded rapidly near the southwestern water bodies. In Yiyang, high-value areas of level IV and above exhibited limited expansion, while level III zones spread outward from government centers. In Changsha, levels III and above of the carbon budget gradually decrease outward from the government center. The Xiang River waterfront remained the high-value core during the study period, gradually expanding into the inland suburbs.

In summary, the spatiotemporal dynamics of carbon emissions, sequestration, and budgets indicate a pronounced imbalance. The ipheral forest areas’ high ecological carbon sequestration capacity has alleviated overall carbon budget saturation. Conversely, carbon emissions from active land development in water-adjacent areas directly threaten water body stability, intensifying carbon budget pressures in the five cities’ northern lakeside plains and waterfront government centers.

3.2.2. Spatiotemporal Differentiation and Clustering Effects of Carbon Budgets in Territorial Space

To visually explore the spatiotemporal correlation effects of the carbon budget pattern in the Dongting Lake region, this section presents SDEs and centroid migration trajectories of carbon budgets, revealing differentiation patterns for the region as a whole and for individual cities (Figure 7). The left side of Figure 7 illustrates the overall spatiotemporal variation in carbon budgets: (1) For centroid migration, the carbon budget centroid is situated near the Yueyang-Yiyang boundary and follows a clockwise trajectory from west to east. From 2005 to 2010, the centroid shifted towards the southeast (11.32 km), and since 2010, it has continued to migrate south (4.69 km). (2) For ellipse shapes, the SDE of carbon budgets extends along the north–south axis. The increasing ratio of the ellipse’s major to minor axes each year further reflects the flattening of carbon budgets along the north–south direction. (3) For directional changes, the azimuth angle shows a progression of “initial contraction followed by expansion”. These results underscore the strong correlation between the north–south spatial arrangement of the Dongting Lake basin and the region’s carbon budget distribution. Moreover, the carbon budget differentiation pattern’s dependence on the water system increases annually.

Compared to the overall differentiation characteristics, the spatiotemporal differentiation in individual cities is more pronounced, as illustrated on the right side of Figure 7: (1) For centroid migration, the carbon budget centroids of the five cities predominantly shift toward internal river systems, underscoring the significant influence of the Dongting Lake region’s water ecological resources on these trajectories. (2) For ellipse shapes, the SDE flattening ratios for Yueyang and Yiyang increased by 0.035 and 0.093, respectively, reflecting a specific spatial spillover effect in carbon budgets. In contrast, the flattening ratios for Jingzhou, Yiyang, and Changsha decreased by 0.023, 0.109, and 0.241, respectively, indicating increasing concentration of carbon budgets along their primary directions. Changsha shows the highest degree of concentration, with a particularly pronounced polarization of the carbon budget pattern toward central construction land. (3) For directional changes, Yueyang exhibits the most excellent azimuth rotation (2°46′34″), driven by fluctuations in Dongting Lake’s water body area. This further highlights the instability of Yueyang’s carbon budget dynamics and its pivotal role in shaping the region’s overall carbon budget pattern. In conclusion, river system orientation and construction land distribution strongly shape the carbon budget patterns of individual cities.

This section utilizes Moran’s I index to examine the spatiotemporal clustering characteristics of territorial space carbon budgets.

Table 6. Spatial correlation of territorial space carbon budgets in the Dongting Lake region.

Year	2005	2010	2015	2020
Moran’s I	0.60	0.65	0.66	0.66
Z value	45.20	49.00	49.69	50.09
p value	0.00	0.00	0.00	0.00

displays the spatial correlation of carbon budgets in the Dongting Lake region throughout the study period. Overall, the Moran’s I indices for the four periods are 0.595, 0.645, 0.655, and 0.661. The Z-values increase annually, reaching 45.20, 49.00, 49.69, and 50.08—well above the significance threshold of 2.58. Furthermore, the p-values consistently remain below 0.01. These findings confirm a strengthening positive spatial correlation in territorial space carbon budgets in the Dongting Lake region. Figure 8 illustrates the visualized clustering effects. (1) H-H clusters declined from 7.5% to 5.5% between 2005 and 2020, with Jingzhou consistently holding the highest proportion (11.68%). These clusters are primarily located in economically active central urban areas. Carbon budgets remain high overall due to construction land’s high carbon emissions and low sequestration capacity. From 2005 to 2020, H-H cluster areas in Jingzhou and Changde exhibited a shrinking trend. In Jingzhou, spatial dependence initially decreased before rising again, aligning with fluctuations in carbon budgets. H-H cluster areas in other cities expanded, with previously scattered patches becoming more contiguous. This suggests that weak environmental protection measures have contributed to carbon source overload in these areas. (2) L-L clusters expanded from 18.8% to 20.2% between 2005 and 2015 but declined slightly to 19.8% between 2015 and 2020. L-L cluster areas are primarily located in ecologically dominant zones such as forests and water bodies, where carbon budgets remain negative due to high carbon sequestration capacity. The area fluctuations in this category reflect initial ecological land encroachment, later mitigated by reforestation policies. (3) H-L clusters remained stable at 0.1% between 2005 and 2020, with minimal proportions across all cities. These areas are sparsely distributed along the periphery of L-L cluster regions. (4) L-H clusters remained low, declining from 0.9% to 0.3% between 2005 and 2020. Like H-L clusters, these areas are located around H-H clusters and are susceptible to assimilation effects.

3.3. Driving Mechanisms of Carbon Budgets in Territorial Space

Considering the spatiotemporal variation characteristics of territorial space carbon budgets in the Dongting Lake region, this section identifies five key factors influencing carbon budgets as decomposition elements. By applying the LMDI model, the driving forces behind carbon budget changes from 2005 to 2020 are quantified. The results are shown in Figure 9.

3.3.1. Carbon Emission Intensity Effect of Territorial Space

Overall, production and living spaces contribute 95% of carbon emission intensity, highlighting the dominant role of energy consumption in human construction activities affecting carbon budgets. In contrast, the carbon sequestration capacity of ecological land has a negligible effect on carbon emission intensity, resulting in only minor fluctuations in carbon budgets. At the city level, the influence of carbon emission intensity varies considerably and fluctuates across the five cities. Yiyang’s total carbon emission intensity reaches 5.27 × 10⁶ tC, with 99% of emissions originating from production and living spaces. Over time, the intensity of carbon emission in Jingzhou, Changde, Yiyang, and Changsha initially declined before increasing. Between 2005 and 2010, carbon emissions showed a positive driving effect, but from 2010 to 2020, Changsha and Yiyang experienced negative growth, indicating suppressed carbon emissions. Notably, in Yueyang, carbon emission intensity fluctuates, with the per-unit carbon budget of production and living spaces exceeding that of other cities. Yueyang’s petrochemical industry, the city’s first trillion-yuan sector, consumes large amounts of raw coal and crude oil, resulting in high carbon emissions from production and living spaces.

3.3.2. Land Use Structure Effect of Territorial Space

Overall, the cumulative effect in the Dongting Lake region from 2005 to 2020 totaled 8.35 × 10⁵ tC. Initially, the territorial space structure had a negative impact, gradually shifting to a positive effect between 2010 and 2020. This suggests that changes in land use structure gradually became the primary driver of carbon budget increases. At the city level, the effects of land use structure in Jingzhou, Changde, and Yiyang mirrored the regional trend, transitioning from negative to positive, indicating adverse development trends. This shift suggests that as urbanization progresses and industries relocate to suburban areas, the large-scale conversion of farmland and ecological land into production and living spaces intensifies energy consumption and carbon emissions. Conversely, in Changsha and Yueyang, the positive effects have weakened, indicating a favorable development trend. These cities have progressively optimized their land use structures, transitioning toward more efficient, low-carbon land use practices.

3.3.3. Land Use Efficiency Effect of Territorial Space

Overall, the land use efficiency negatively suppressed carbon emissions, decreasing emissions from −6.47 × 10⁷ tC in 2005 to −2.63 × 10⁷ tC in 2020. The annual reduction in the production and living land required per unit of GDP indicates that economic development is driving more intensive and efficient land use. At the city level, the land use efficiency effect in all five cities showed a significant negative impact. Among them, Yueyang exhibited the most pronounced suppression effect across three time periods, reducing carbon emissions by 2.79 × 10⁷ tC, 2.13 × 10⁷ tC, and 1.88 × 10⁷ tC, respectively. In summary, against the backdrop of China’s economic transition, urbanization has entered a phase of slower growth, making land intensification increasingly challenging each year. As urbanization slows, land use efficiency weakens, and the suppression effect on carbon emissions diminishes each year.

3.3.4. Economic Development Effect

Overall, the decomposition results reveal that rising per capita GDP in the Dongting Lake region positively affects carbon budgets. This further confirms the strong correlation between economic growth and carbon emissions. During China’s rapid economic development phase (2005–2010), the economic development effect in the study area 6.14 × 10⁷ t. This indicates a notable rise in carbon emissions linked to accelerated growth. However, with the introduction of high-quality urban development strategies, per capita GDP growth slowed between 2010 and 2020. This slowdown is closely related to implementing energy efficiency and carbon reduction technologies. At the city level, economic development significantly drives carbon emissions in all five cities. In terms of temporal variation, from 2005 to 2010, economic development in Changsha, Yueyang, and Changde had a powerful positive impact on carbon emissions. Between 2010 and 2015, the impact of economic growth declined in all cities except Jingzhou. By 2015–2020, economic growth reached a saturation point for energy-intensive products, and the diminishing effect of economic development on emissions became more pronounced. This suggests that increasing awareness of carbon emission reduction is starting to play a significant role in shaping the development trajectories of these cities.

3.3.5. Population Scale Effect

Overall, the expansion of the population scale in the Dongting Lake region positively influences carbon budgets, but the effect is not significant. Between 2005 and 2020, the study area steadily declined from 3.25 to 1.16 × 10⁶ tC. At the city level, the population scale effect in Changsha significantly contributed to carbon emissions, increasing by 3.81 × 10⁶ tC. Due to regional population policies, the population scale effect in the other four cities fluctuated downward, consistent with the overall regional trend. This indicates that rapid urbanization in the early period placed significant pressure on urban ecosystems, driving increased carbon emissions from urban land use. As low-carbon urban living improves and population growth slows, the effect of population-scale factors begins to decline.

4. Discussion

4.1. Multidimensional Validation of the Robustness of the Territorial Carbon Budgets Accounting Framework in Lake-Intensive Regions

Lake-intensive regions, characterized by distinct water–land interlaced landscapes and diverse ecosystems, possess inherent carbon sequestration potential. Lakes and their surrounding wetlands, forests, and grasslands play a crucial role in carbon sequestration. These areas absorb and store substantial carbon emissions, regulate regional climate, and maintain ecological balance. Therefore, using lake-intensive regions to achieve carbon balance is scientifically justified and provides a solid theoretical basis for developing precise, region-specific low-carbon strategies. To ensure the reliability of the study’s results, this research systematically evaluates the model’s robustness through parameter sensitivity tests, spatial consistency checks, and robustness assessments of driving factor decomposition.

First, parameter sensitivity analysis shows that although the selection of carbon source and sink coefficients has minimal impact on total estimates, it is crucial for identifying spatial variation patterns [13]. To ensure reliable carbon balance results in lake-intensive regions, this study adopts the IPCC-recommended direct estimation method for carbon emissions and an indirect method based on energy coefficients, integrating remote sensing imagery and socioeconomic data for systematic carbon source and sink calculations. Additionally, comparing carbon budgets under varying carbon emission and sequestration coefficients in the Dongting Lake region confirms the consistency of carbon balance trends [38]. This reinforces the robustness of the findings. However, high-value territorial carbon budget areas are more sensitive. When the forest carbon sequestration coefficient is reduced by 20%, the L-L clustered area in edge forests shrinks by 12%. This suggests that underestimating the carbon sequestration capacity of ecological spaces may weaken their spatial adverse feedback effects.

Second, this study employs a multi-scale and multi-resolution cross-validation approach to eliminate potential biases in selecting spatiotemporal exploratory analysis methods. The spatial autocorrelation results of carbon budgets on a 5 km × 5 km grid show strong consistency with Moran’s I index at the urban scale (p < 0.01). Moreover, SDE analysis of carbon budgets reveals that high-value areas significantly overlap with significant water systems, consistent across scales. After adjusting the grid resolution to 3 km × 3 km, the spatial coverage variation of each cluster remains below 5%, indicating that the core conclusions exhibit low sensitivity to spatial unit scale. These findings further reinforce the robustness of spatiotemporal variation patterns in carbon budgets. Furthermore, spatial clustering and SDE analysis results align with trends reported in the relevant literature [39], reinforcing the reliability of this study’s conclusions.

Finally, the LMDI model remained stable in ranking driving factor contributions across scenarios. Even when excluding contentious parameters (e.g., potential ecological carbon sink coefficients), the economic development effect (ΔCg) and land use efficiency effect (ΔCγ) retained their dominance as primary positive and negative contributors. This finding parallels Meng’s Yellow River Basin decomposition results, underscoring the model’s robustness in isolating critical driving mechanisms [20].

In summary, this study’s “carbon source–carbon sink” spatiotemporal accounting framework demonstrates strong robustness in facing the threefold challenges of parameter uncertainty, spatial heterogeneity, and driver complexity. While localized deviations in carbon accounting parameters may affect absolute value accuracy, the model reliably captures key patterns, including the annual growth trend of regional carbon budgets, high-emission clusters near aquatic systems and urban centers, and the dominance of economic development and land use efficiency as primary drivers. These findings provide a reliable foundation for developing carbon balance policies in lake-dense regions.

4.2. Spatiotemporal Variation Patterns of Territorial Space Carbon Budgets in Lake-Intensive Regions

Lake-intensive regions are pivotal for advancing regional green and low-carbon transitions. Clarifying the mechanisms of regional carbon budget imbalances is the first step in formulating carbon balance regulation strategies. Previous studies have highlighted diverse carbon budget trends across regions in China [40,41]. Unlike prior studies, this study introduces a novel territorial space carbon source–sink measurement system integrating production–living–ecological functions. Building on this, the study employs ESTDA models to examine the variation and clustering effects of territorial space carbon budgets. This study fills gaps in research subject selection and accounting methodologies by uncovering the spatiotemporal evolution of carbon budgets in lake-intensive regions.

The spatiotemporal evolution of carbon budgets reveals an annual increase in the territorial space carbon budget within the Dongting Lake region, though the growth rate has slowed from 19.13% to 0.10%. Despite this deceleration, the region remains short of achieving its “carbon neutrality” target. The recent plateau in carbon budget growth may be attributed to policies such as converting farmland to forests and lakes. As a key grain production base, the Dongting Lake region’s agricultural production space has experienced significant spatial shifts in recent years. Converting farmland to ecological land has partially alleviated the region’s carbon deficit. Spatially, variations in the carbon budget arise from differences in resource endowments and economic development among cities [42]. High-carbon budget areas are typically plains with well-developed water networks and higher economic development. Carbon sources from production and living spaces are primarily on the lake’s outskirts in Yueyang and along both banks of the Xiangjiang River in Changsha. Carbon sinks in ecological spaces are found in forested areas, vegetation cover, and the main lake body, including western Changde, southern Yiyang, and western and eastern Yueyang. These high carbon source–sink areas shape the Dongting Lake region’s carbon budget pattern and are critical for achieving China’s future carbon balance goals. This finding aligns with An’s research [43].

Using ESTDA, this study uncovers the spatiotemporal variation and clustering effects of carbon budgets in the Dongting Lake region. SDE analysis highlights how the orientation of water systems and the layout of construction land in the Dongting Lake basin influence carbon budget patterns at both the regional and city levels. The water-adjacent ecological advantages of the Dongting Lake region spur economic activities, leading to a spatial divergence in carbon budgets with water bodies as the central axis, radiating outward toward forested areas. For example, Changsha’s carbon budget center gradually shifts toward the central Xiangjiang River. The declining SDE flattening ratio reflects the convergence of carbon budgets from the periphery to the riverbanks. Jingzhou shows a similar trend to Changsha, with the carbon budget pattern gradually converging toward the Songzi River. At the spatiotemporal clustering level, H-H clusters concentrate in the government-seat areas of each city. L-L clusters occur in peripheral forest areas. H-L and L-H clusters are scattered on the outskirts of L-L and H-H areas, respectively, and are easily assimilated by them. Notably, H-H clusters show distinctly different development trends across government-seat areas. This suggests that although carbon budgets cluster clearly in government-seat areas, development trends are shaped by local economic structures, policies, and regional cooperation. Therefore, lake regions should adopt location-specific, differentiated, and precise carbon reduction and sequestration management strategies [44,45].

4.3. Analysis of Factors Influencing Territorial Space Carbon Budgets in Lake-Intensive Regions

Analyzing how territorial space elements affect carbon budgets in lake-intensive regions informs targeted strategies to mitigate carbon imbalances. In the Dongting Lake region, socioeconomic development is closely tied to carbon budget changes. Using the Kaya identity and the LMDI method, this study finds that regional GDP growth, population density, industrial restructuring, and energy consumption shifts significantly influence carbon emissions. In recent years, industrial restructuring and clean energy adoption have slowed carbon emission growth. Urbanization has reshaped urban–rural spatial patterns, further affecting carbon budget dynamics. Regional policy optimization has reduced carbon emissions, including high-carbon industry transitions, green infrastructure development, and energy structure improvements. Regarding the various factors influencing territorial carbon budgets in the Dongting Lake region, economic development and land-use efficiency are the primary driving forces. This finding aligns with existing research and highlights its applicability to lake-intensive regions [20,46].

Specifically, economic development in the Dongting Lake region has cumulatively contributed approximately 1.31 × 10⁸ tC of carbon emissions, making it the most significant positive contributing factor. Existing literature similarly identifies economic development as a key driver of increased carbon emissions [47]. For example, Zheng demonstrated that per capita GDP is the most influential contributor to China’s carbon emissions growth [48]. Li also highlighted that economic development contributes the most to China’s carbon emissions [49]. Economic development typically brings an influx of capital, labor, and resources, driving production activities and living demands and substantially increasing territorial space carbon emissions [50]. Therefore, in advancing a low-carbon circular economy in lake-intensive regions, it is crucial to prioritize preserving carbon sink potential and implementing stringent emission reduction measures, particularly in lakeside areas.

Furthermore, enhancing land use efficiency in territorial space is essential for mitigating carbon imbalances. This study reveals that improving energy and resource use efficiency can significantly reduce carbon emissions in the Dongting Lake region, especially given the rapid expansion of construction land near water bodies. Research by Wu indicates that GDP per unit of land use area suppresses the total carbon emissions of China’s Yangtze River Delta urban agglomeration [51]. Liu reported that land-use intensity per unit of GDP negatively affects carbon emissions from land use in Zhejiang Province’s counties [22]. These findings support this study’s conclusions, highlighting that improving land use efficiency in territorial space effectively curbs carbon emissions.

In conclusion, strengthening land use management, optimizing territorial space layout, and limiting construction land expansion near water bodies are critical strategies for balancing carbon sources and sinks in lake-intensive regions. Rational spatial planning and improved land use efficiency can promote economic development in lake-intensive regions while sustaining the ecological environment and achieving harmony between economic growth and environmental sustainability.

4.4. Policy Interventions and Planning Pathways for Carbon Balance in Lake-Intensive Regions

Policy interventions play a crucial role in regulating carbon budgets in lake-intensive regions. In recent years, the Dongting Lake region has mitigated the growth rate of carbon emissions by promoting green economic development, advancing industrial transformation and upgrading, and optimizing energy structure policies. For instance, the returning farmland to forest policy has significantly enhanced the carbon sequestration capacity of forest ecological spaces in the Dongting Lake region. Meanwhile, low-carbon industrial transformation policies have effectively reduced carbon emission intensity in the secondary sector. However, policy effectiveness often varies across regions. For example, economically developed areas such as Changsha and Yueyang have experienced a significant slowdown in carbon budget growth due to early industrial restructuring. In contrast, Jingzhou, an agriculture-dominated region, still faces a prominent carbon deficit due to delayed policy implementation. This disparity reflects the profound impact of different socioeconomic development levels on carbon budget patterns. Therefore, policy design should emphasize regional coordination and establish cross-regional ecological compensation mechanisms to balance carbon emissions from economic development with carbon sequestration in ecological conservation [52,53]. In this context, this study formulates a carbon balance planning pathway for the Dongting Lake region’s production–living–ecological functional framework, focusing on three key priorities: economic development, farmland protection, and ecological security, specifically as follows:

(1) For carbon reduction and sequestration in agricultural production space, water and soil resource advantages in water–land interlaced zones should be maximized to modernize irrigation areas along the lake. Based on this, the agricultural structure should shift toward a low-carbon industry model with enhanced ecological benefits. The strict protection of permanent bare farmland must be maintained. In contrast, initiatives such as returning farmland to lakes, afforestation, and reforestation should be promoted to develop waterfront agricultural systems that align with the carrying capacity of watershed resources and the environment. For example, in Jingzhou, a waterfront farmland-dominated region, efforts should be intensified to return farmland to lakes and adjust the agricultural structure, emphasizing lakeside green agriculture, eco-friendly circular farming systems, and enhanced agricultural carbon sequestration. As a key agricultural production area, Changde should strengthen farmland ecological protection, promote green agricultural technologies, advance smart agriculture, and maximize agricultural carbon sequestration benefits.

(2) For carbon reduction in urban–rural living space, functional land use for industrial, residential, and public service facilities should be adjusted based on the watershed region’s hydrogeographical characteristics to guide cities toward low-emission urban spatial development. Unnecessary urban development near riverbanks should be restricted, and high-carbon industries should transition to low-carbon or zero-carbon models alongside energy system innovations, promoting hydropower, hybrid wind–solar energy, and other green sources. Regarding spatial layout, high-pollution, energy-intensive industries should be relocated from waterfront areas to enhance green ecological functions along the water. As the region’s industrial and technological innovation hub, Changsha should accelerate the transformation of energy-intensive industries, expand green energy adoption, promote green building and smart city initiatives, enhance low-carbon public transportation, and foster compact, low-carbon urban-rural development. As a key nodal city in the Dongting Lake region, Yiyang should optimize its urban spatial layout, limit unnecessary expansion, facilitate traditional industry transformation, and encourage green infrastructure to align urban growth with carbon reduction goals.

(3) For carbon sequestration in blue-green ecological space, the watershed’s spatial ecological configuration should follow a “surface→corridor→node” structure, prioritizing key carbon sequestration areas such as forests, water bodies, and wetlands. Forest ecological green belts and water system corridors should be interconnected, along with developing wetland and lakefront parks as carbon sequestration hubs. Ecological restoration projects, including water replenishment and aquaculture conversion to wetlands, should be implemented to sustain the carbon sequestration function of watershed cities and offset losses from degraded lake ecosystems. As a critical water ecological barrier in the Dongting Lake region, Yueyang should increase investments in wetland restoration, strengthen wetland resource protection, develop an efficient ecological corridor system to enhance carbon sequestration and establish a regional ecological compensation mechanism to reinforce wetland conservation policies. Changde and Yiyang should prioritize forest restoration, improve the connectivity of ecological corridors, enhance regional carbon sequestration, and maximize the sequestration benefits of forest ecosystems.

In conclusion, although ongoing urbanization may threaten aquatic ecosystems, sustainable urbanization has become a mainstream trend in global urban management. An increasing number of studies and practices are exploring eco-friendly development strategies to foster the harmonious coexistence of urban production–living spaces and aquatic ecosystems. For example, promoting green infrastructure, optimizing land use patterns, and advancing low-carbon technologies can enhance ecosystem carbon sequestration capacity while ensuring economic development. Scientific territorial planning and ecological restoration projects can achieve an integrated balance between urban development and ecological conservation, particularly in lake-intensive regions. This approach not only helps mitigate carbon emission pressures but also provides crucial support for the sustainable development of lake ecosystems.

4.5. Limitations and Future Prospects

Looking ahead, further research is needed to accurately identify high-value carbon budget areas in lake-intensive regions and guide carbon reduction and sequestration strategies:

(1) Although the carbon balance model in this study is robust in the Dongting Lake region, future changes in urban planning or socioeconomic demands may alter the balance between carbon emissions and sequestration. For example, if industrial transformation policies in the Yangtze River Economic Belt encourage a shift from traditional manufacturing to clean energy industries, structural changes in GDP would significantly lower carbon emission intensity per unit of output. This transition may also increase water demand, particularly for hydropower, affecting the carbon sequestration capacity of aquatic ecosystems. To address these changes, future research should dynamically update carbon emission coefficients, adjust model parameters, and incorporate the latest socioeconomic and environmental data. By further exploring the interaction between socioeconomic policies and carbon emissions, the model’s responsiveness to real-world changes can be ensured. For instance, we plan to integrate scenario analysis into the model and adjust carbon source–sink parameters dynamically using policy-driven factors (e.g., carbon taxes and ecological compensation) to enhance the flexibility of carbon balance projections. We will also explore integrating dynamic carbon sequestration models to track carbon emission fluctuations driven by regional policies, climate change, and industrial restructuring. Additionally, we recommend establishing a dynamic monitoring system to continuously track changes in socioeconomic activities and greenhouse gas emissions in the Yangtze River Basin. This will ensure forward-looking research findings.

(2) This study does not comprehensively account for carbon-to-methane conversion across different land-use types. Methane, a potent greenhouse gas, is primarily emitted from rice cultivation in agricultural production spaces, coal mining in industrial-mining production spaces, wastewater treatment in urban living spaces, and lake sediments in water ecological spaces. Under anaerobic conditions, these land-use types may produce methane, influencing the overall carbon budget pattern. Due to data and model limitations, this study has not yet systematically analyzed methane emissions. Future research should investigate carbon-to-methane conversion mechanisms across land-use types, assess their dual effects on the carbon budget, and improve the scientific rigor of carbon balance evaluations.

(3) Wetlands in lake-intensive regions are widely recognized as key carbon sequestration areas. Although wetlands have strong carbon sequestration capacity, their stability is highly sensitive to hydrological changes, land use alterations, and other factors, leading to fluctuations in their sequestration function. Future research should assess both the benefits and risks of wetland carbon sequestration, refine management strategies, and ensure the long-term sustainability of sequestration benefits.

(4) Future studies on carbon budget driving mechanisms could incorporate additional factors related to economic development, urbanization, and carbon budgets to provide broader insights and more comprehensive findings. However, this does not affect the feasibility of the current research framework and findings. These limitations do not undermine this study’s conclusions but instead highlight opportunities to improve the model’s applicability in complex socioeconomic contexts.

5. Conclusions

This study introduces a comprehensive framework for analyzing “carbon sources–carbon sinks” in territorial space, emphasizing spatiotemporal evolution trends and the driving mechanisms of carbon budgets. Using the Dongting Lake region as a case study, the findings reveal that living spaces in the Dongting Lake region expanded significantly around city government centers, encroaching on high-quality water bodies and farmland. Land-use transitions are significantly correlated with spatiotemporal carbon budget evolution. The Dongting Lake region’s total carbon budget increased annually, although the growth rate slowed. A “core–periphery” spatial development pattern along water bodies became more pronounced. Carbon budget differentiation followed a north–south development trend, with increasing reliance on the water system. The clustering effect of the carbon budget exhibited significant spatiotemporal heterogeneity, with high-carbon source areas concentrated around Dongting Lake water bodies and urban centers. High-carbon sink areas were concentrated in the western, eastern, and southern regions, although clustering intensity declined over time. Various factors influenced carbon emissions in the Dongting Lake region’s territorial space. Between 2005 and 2020, land-use efficiency had the most potent inhibitory effect on the carbon budget (−1.37 × 10⁸ tC), whereas economic development exerted the most significant positive impact (1.31 × 10⁸ tC). To advance low-carbon development in similar lake-intensive regions, we recommend revising territorial space planning, optimizing construction land use, and restricting excessive waterfront development. Enhancing ecological protection is essential, particularly for carbon sinks such as forests and wetlands. Optimizing economic structures to promote sustainable agricultural practices is essential for fostering functional synergy in lake-intensive regions.