The impact of the Grain-for-Green Programme on carbon storage in the Upper Yangtze River Basin based on the PLUS-InVEST model

Abstract

Alterations in land use and land cover (LUCC) play a fundamental role in influencing the variability of ecosystem carbon storage. Evaluating how land use dynamics affect carbon sequestration and projecting future carbon storage scenarios are essential steps toward meeting China’s dual carbon objectives. In this study, we integrated the Patch-generating Land Use Simulation (PLUS) model with the Integrated Valuation of Ecosystem Services and Tradeoffs (InVEST) framework to investigate LUCC dynamics and their implications for carbon storage across the Upper Yangtze River Basin (UYRB) between 2000 and 2020. Furthermore, projections of regional carbon storage were made under multiple Grain-for-Green Programme (GFGP) scenarios extending to the year 2040. Our findings indicated that cultivated land (CL), forest land (FL), and grassland (GL) consistently dominated land use composition within the UYRB, collectively occupying approximately 96.45% of the total area throughout 2000–2020. During this period, construction land (CSL) steadily expanded, primarily at the expense of CL. Both CL and GL experienced substantial reductions. Spatially, carbon storage exhibited a decreasing gradient from east to west, with the Jinsha River Basin exhibiting the greatest levels. Carbon storage values over the two decades were recorded at 6.387 × 10¹⁰ t in 2000, 6.382 × 10¹⁰ t in 2005, 6.379 × 10¹⁰ t in 2010, 6.369 × 10¹⁰ t in 2015, and 6.373 × 10¹⁰ t in 2020. Despite a slight recovery between 2015 and 2020, total carbon storage fell by 0.23% (1.438 × 10⁸ t) overall. This decline was primarily driven by the conversion of high-carbon-density CL and FL into low-carbon-density CSL and GL. Future projections show distinct disparities across four policy scenarios by 2040. Under the Natural Development Scenario (NDS), rapid economic growth and land conversion are projected to result in a carbon storage loss of 1.324 × 10⁸ t. Conversely, the mild, moderate, and strong GFGPS anticipate carbon storage increases of 1.385 × 10⁸ t, 3.157 × 10⁸ t, and 5.136 × 10⁸ t, respectively. The Jialing River Basin shows the highest gains under all GFGPS. Our findings underscore the significance of the GFGP in enhancing regional carbon sequestration, primarily through encouraging afforestation of previously CL and GL and curbing the expansion of CSL. Such insights can guide land-use planning and ecological conservation strategies in the UYRB moving forward.

Introduction

Tackling climate change and its impacts is a critical focus of the United Nations Sustainable Development Goals (SDGs), which emphasize enhancing ecosystem carbon storage as a cornerstone of sustainable development [1]. Carbon storage, an essential ecosystem service, plays a pivotal role in stabilizing the global climate and mitigating greenhouse gas emissions [2]. Under this framework, ecological restoration initiatives, such as reforestation and wetland rehabilitation, are recognized as vital strategies to enhance ecosystem carbon storage [3]. By increasing vegetation cover, these efforts not only bolster carbon sequestration but also help mitigate climate change impacts [4]. However, comprehending the impacts of these projects on carbon storage and their dynamic processes remains complex, due to significant regional and methodological variations in effectiveness [5]. Therefore, it is essential to systematically assess how ecological restoration initiatives influence carbon stocks, enabling enhanced carbon sequestration and facilitating the attainment of sustainability targets [6].

Research on the effects of ecological restoration projects on carbon storage has primarily emphasized the refinement and application of methodological models [7]. Traditional approaches, such as remote sensing and GIS-based monitoring, statistical modeling, and process simulations, have significantly advanced quantitative carbon storage assessments [8]. However, these methods face challenges in accurately capturing the dynamic impacts of land use changes on carbon storage, particularly in addressing spatial heterogeneity and scenario-based simulations [9]. As a result, assessments often lack precision and fail to adequately represent the intricate characteristics of ecosystems across different regions [7, 10]. Moreover, traditional methods struggle to predict carbon storage changes under diverse future policies or climatic conditions, potentially skewing evaluations of the long-term benefits of ecological restoration and hindering data-driven policymaking for such projects [7]. To address these limitations, the InVEST model has gained traction among researchers for evaluating the impacts of ecological restoration projects on carbon storage [7, 11, 12–13]. Compared to traditional methods, the InVEST model offers high spatial resolution, the ability to simulate multiple ecosystem services, and exceptional capacity to assess the spatial distribution and dynamic changes in carbon storage resulting from restoration activities [14].

The potential of the InVEST model in carbon storage assessment has been widely acknowledged by researchers. By integrating it with land use change models such as CLUE-S, CA-Markov, and FLUS, studies have explored the effects of ecological restoration projects on carbon storage under diverse scenarios [14, 15–16]. However, these traditional models exhibit limitations in capturing the complexity and spatial heterogeneity of land use transitions [17]. To address these challenges, the PLUS model was developed to offer more precise simulations of land use changes, emphasizing spatial patterns and human activity influences [16]. Combining the PLUS and InVEST models enables a more holistic evaluation of the dynamic impacts and trends of ecological restoration projects on carbon storage under future scenarios [18]. This integrated approach enhances the spatial accuracy of carbon storage assessments and provides deeper insights into the interactions between human activities and land use. It equips policymakers with robust scientific evidence to design more effective ecological restoration policies, thereby maximizing carbon sequestration benefits and advancing sustainable regional development [19, 20]. The application of the PLUS-InVEST model is therefore indispensable, offering policymakers a comprehensive understanding of the spatial dynamics of carbon storage and the broader implications of ecological restoration initiatives.

The GFGP represents a cornerstone of ecological restoration efforts, primarily aimed at restoring forest vegetation and enhancing ecosystem services by increasing carbon storage [5, 6–7, 21]. The UYRB, a key region for GFGP implementation in China, holds critical ecological significance as both a vital water conservation zone and a strategic barrier within the national ecological security framework [22]. The ongoing, nationally coordinated execution of GFGP highlights the necessity of assessing its impacts on carbon storage to inform policy and guide practical applications [23]. This study focuses on the UYRB, employing the InVEST-PLUS model to analyze land use and carbon storage dynamics from 2000 to 2020 and to project these changes under four reforestation policy scenarios for 2040. It systematically examines the influence of GFGP on carbon storage in the UYRB, offering scientific evidence for land use optimization, ecosystem service enhancement, and sustainable development. Furthermore, this research supports the achievement of the “dual carbon” goals and fosters a harmonious coexistence between humans and nature.

Materials and methods

Study area

The UYRB (21°8′-39°20′N, 78°25′-110°11′E) is located in southwestern China (Fig. 1), spanning six provinces: Qinghai, Tibet, Sichuan, Yunnan, Chongqing, and Hubei, with a total area of approximately 9.9 × 10⁵ km² [24]. The study region exhibits a distinct topographic gradient descending from higher altitudes in the west to lower altitudes in the east, covering the first and second geomorphological terraces of China. The topography is complex, with altitudes ranging from 3000 to 5000 m [25]. The UYRB is one of China’s most crucial water sources, exhibiting diverse climatic conditions, including subarctic and subtropical monsoon climates [26]. The western part of the study area shows significant vertical variations in biodiversity and climate, with vegetation predominantly consisting of evergreen broad leaf forests and evergreen broadleaf mixed forests [27]. The central region is dominated by basins and mountainous terrains, with vegetation primarily composed of coniferous forests and evergreen broadleaf forests [28].

[See PDF for image]

Fig. 1

Study area overview

Data sources and processing

Land use data

Land use data for the UYRB, covering the years 2000–2020, were derived from the datasets provided by Yang and Huang [29] (https://zenodo.org/record/8176941) at a resolution of 1 km × 1 km. These datasets were harmonized with the “Classification of Land Use Status Quo GB-T21010-2007” standard. To efficiently capture the overall land use dynamics in the UYRB, this study categorizes land use into six primary classes: cultivated land (CL), forest land (FL), grassland (GL), water area (WA), construction land (CSL), and unused land (UL). While this simplification may overlook certain specific land types—such as wetlands, which exhibit unique carbon cycle characteristics—their relatively small extent within the study area and inclusion under the WA category minimize their influence on the overall carbon storage assessment. Moreover, this classification approach has been widely adopted in relevant literature, ensuring consistency in data processing and facilitating comparability of results [30]. This standardized classification was designed to streamline subsequent analyses and ensure consistency across all spatial and temporal comparisons.

Driver data

To simulate land use changes in the study area, 15 driving factors were selected based on the actual land use conditions and previous research findings (Fig. 2; Table 1). Socioeconomic factors included GDP, population, distances to railways, distances to roads, distances to county-level settlements, and distance to rivers, comprising 6 variables. Climate and environmental factors included digital elevation model, slope, slope direction, soil organic matter content, soil pH, soil sand content, temperature, and precipitation, making up 9 variables. Climate data, such as annual average precipitation and temperature, were derived using the Anusplin interpolation method. Topographical attributes, such as slope and slope direction, were calculated using a Digital Elevation Model (DEM). Socioeconomic data, including distances to roads, railways, and county-level settlements, were extracted based on data from OpenStreetMap (OSM) and the National Bureau of Statistics and calculated by Euclidean distance tool in ArcGIS 10.7.

[See PDF for image]

Fig. 2

Main driving factors of LUCC in UYRB

Table 1. The driving factors of simulated LUCC

Data Type	Data Name	Year	Data Accuracy	Source
Land use data	Land use	2000–2020	1 km	https://zenodo.org/record/8176941
Natural factor	Precipitation	2000–2020	1 km	Resource and Environment Science and Data Center (https://www.resdc.cn/)
	Temperature	2000–2020	1 km
	DEM	2020	1 km	SRTM
	Slope direction	2020	1 km	From DEM
	Slope	2020	1 km
	Soil pH	2020	1 km	HWSD (https://lpdaac.usgs.gov/products/srtmgllv003/)
	Soil sand content	2020	1 km
	Soil organic matter content	2020	1 km
Social factor	Population	2015–2020	1 km	Resource and Environment Science and Data Center (https://www.resdc.cn/)
	GDP	2015–2020	1 km
	Distance to road	2015–2020	1 km	OpenStreetMap (https://openmaptiles.org/languages/zh/)
	Distance to railway	2015–2020	1 km
	Distance to river	2020	1 km
	Distance to county-level settlements	2015–2020	1 km

Research methods

PLUS model

The PLUS model, a tool for simulating land use changes, relies on a robust patch-generation mechanism [31]. Key components of the model include the Markov prediction module, the Land Expansion Analysis Strategy (LEAS) model, and the CA-based CARS model, which employs multiple random seed algorithms [19, 32]. By overlaying land use datasets from two distinct time periods, the model identifies regions of change and extracts statistical data to predict potential future land use states under predefined conditions [16, 18, 19, 33]. The Random Forest algorithm is utilized to analyze the interactions between land management strategies and influencing factors [18, 34]. Research indicates that various land use types demonstrate unique transformation patterns under the influence of specific factors, which directly shape their developmental potential [20, 35]. Based on these constraints, the PLUS model autonomously generates patch simulation outcomes [16, 18, 19, 36]. Previous studies have highlighted the model’s ability to dynamically integrate spatial and geographic parameters, enabling precise simulation of land use distribution patterns [20, 37].

Accuracy verification

Land use scenarios for 2020 were simulated using the PLUS model, calibrated with historical datasets from 2010 to 2015. Validation against actual observations from 2020 yielded a Kappa statistic of 0.87 and an overall classification accuracy of 0.91. These robust validation metrics confirm the model’s reliability and applicability for future LUCC projections within the region. The detailed methodological workflow is illustrated schematically in Fig. 3.

[See PDF for image]

Fig. 3

Flow chart of correlation analysis

Establishment of transition matrices under diverse scenarios

The UYRB, a central focus of China’s GFGP, has experienced significant land use transformations influenced by recent ecological restoration policies. To analyze and predict how these changes might evolve by 2040, four scenarios were developed. For the NDS, the Markov Chain module in the PLUS model was employed to simulate land use in 2040. This approach relied on historical data from 2000 to 2020 and assumed stable transition probabilities among land use categories [38–39]. Three additional scenarios based on the GFGP were also designed, each incorporating key ecological restoration measures such as reforestation and the conservation of natural forests. Under the Mild GFGPS, the probability of converting CL to FL was increased by 20%, and this conversion rate further rose to 35% in the Moderate GFGPS and 50% in the Strong GFGPS [40].

InVEST model

The InVEST model assesses variations in ecosystem services resulting from alternative land-use scenarios, providing policymakers with a robust scientific basis to balance the benefits and consequences of anthropogenic activities [41]. Its carbon storage module quantifies terrestrial ecosystems’ potential to sequester carbon by integrating carbon pools from aboveground biomass carbon (ABC), belowground biomass carbon (BBC), soil carbon (SC), and dead organic carbon (DOC). Carbon density refers to the concentration of carbon stored within a given area. Regional total carbon storage is calculated using the following equation:

1

2

In this formula, C_j denotes the carbon density of the jth land use type. C_j−above, C_j−below, C_j−soil, and C_j−dead indicate the carbon densities for ABC, BBC, SC, and DOC, respectively. S_j is the area covered by the jth land use type, and C_total represents the overall carbon storage.

To estimate carbon storage within the InVEST model, carbon density values must be assigned to each land use type. These values were derived from studies conducted by Xie et al. [42, 43, 44, 45, 46, 47–48]. By applying correction formulas developed by Chen et al. [49–50], the carbon density data for the study area were adjusted to reflect variations in precipitation and temperature.

3

4

5

C_BP and C_BT represent ABC and BBC densities adjusted according to the regional averages of precipitation and temperature, respectively, while C_SP refers to soil carbon density adjusted by average precipitation. MAT and MAP denote the annual average temperature (°C) and precipitation (mm), respectively. The ratio derived by incorporating the region-specific mean temperature (6.9 °C) and precipitation (826.0 mm), relative to national averages (7.5 °C and 575.6 mm), is applied as a correction coefficient for carbon density. The detailed formula is presented below:

6

7

In this equation, C′_BP and C″_BP represent ABC and BBC densities adjusted for regional and national mean annual precipitation, respectively. Likewise, C′_BT and C″_BT denote ABC and BBC density adjusted for regional and national mean annual temperatures. C′_SP and C″_SP denote SC density corrected based on regional and national mean precipitation. K_B and K_S are coefficients for correcting ABC, BBC and SC densities, respectively. The adjusted carbon density values are presented in Table 2.

Table 2. Carbon density for various land-use categories in UYRB (t/hm²)

Land Use Type	Density of ABC	Density of BBC	Density of SC	Density of DOC
CL	125.23	320.83	123.89	0.42
FL	168.57	460.77	140.92	1.49
GL	137.56	343.89	114.29	0.42
WA	0.00	0.00	0.00	0.42
CSL	0.00	0.00	89.15	0.00
UL	47.71	0.00	81.15	0.00

Results and analysis

LUCC patterns in the UYRB from 2000 to 2020

Changes in land use area

From 2000 to 2020, the UYRB was predominantly occupied by CL, FL, and GL, which collectively comprised 96.45% of the study area (Fig. 4; Table 3). Specifically, FL accounted for 41.18%, GL 35.22%, and CL 23.60%. Meanwhile, UL covered 2.34%, WA constituted 0.79%, and CSL represented only 0.54%. Significant changes were observed in CL, FL, and GL during this period (Table 3). CL and GL decreased by 14,615 km² and 15,480 km², respectively, with dynamic rates of -0.63% and − 0.45% (Table 3). From 2000 to 2020, CL exhibited substantial fluctuations and varying degrees of loss, with dynamic rates of -0.12%, -0.10%, -0.11%, and − 0.32% (Table 3). GL showed a similar declining trend with dynamic rates of -0.11%, -0.13%, -0.01%, and − 0.21% (Table 3). Conversely, FL displayed an expansion trend with dynamic rates of 0.11%, 0.11%, -0.02%, and 0.27% (Table 3). CSL experienced the most notable change, increasing by 4,765 km² over two decades, with a dynamic rate of 15.75% (Table 3). UL exhibited a slight yet continuous expansion, while WA underwent minor reductions from 2015 to 2020.

[See PDF for image]

Fig. 4

Spatial patterns of land-use classifications for the years 2000, 2005, 2010, 2015, and 2020

Table 3. Land use/cover area changes and single dynamic degree

	Area/km2					Single land use dynamics (%)
Land Use Type	2000	2005	2010	2015	2020	2000–2005	2005–2010	2010–2015	2015–2020	2000–2020
CL	232,111	229,308	227,073	224,618	217,509	-0.12	-0.10	-0.11	-0.32	-0.63
FL	386,482	390,661	395,038	395,633	406,134	0.11	0.11	0.02	0.27	0.51
GL	344,972	341,108	336,775	336,379	329,464	-0.11	-0.13	-0.01	-0.21	-0.45
WA	6450	7415	7989	8121	7982	1.50	0.77	0.17	-0.17	2.38
CSL	3026	3742	4925	6555	7792	2.37	3.16	3.31	1.89	15.75
UL	20,951	21,758	22,192	22,686	25,111	0.39	0.20	0.22	1.07	1.99

Between 2000 and 2020, the Jinsha River Basin experienced a slight increase in CL area, whereas the Min-Tuo River Basin, the upper Yangtze River, the Jialing River Basin, and the Wu River Basin faced varying levels of CL erosion. In contrast, FL expanded significantly by 19,653 km², with a dynamic degree of 0.51%. The Jialing River Basin exhibited the largest FL increase, contributing an additional 10,177 km², highlighting the exceptional effectiveness of GFGP in boosting forest coverage. Meanwhile, as a core region of the Yangtze River Economic Belt, the UYRB witnessed rapid economic development during this period. From 2000 to 2020, CSL increased dramatically by 4,765 km², representing a 157.52% surge and a dynamic degree of 15.75%. Furthermore, WA expanded by 1,533 km², while UL increased by 4,144 km², with dynamic degrees of 2.38% and 1.99%, respectively.

Land use transition analysis

From 2000 to 2020, the total land use transition in the UYRB reached 96,287 km², accounting for 9.69% of the total area (Fig. 5; Table 4). Specifically, CL experienced an outflow of 42,496 km², primarily transitioning into FL (31,752 km²), accounting for 74.72%. Additionally, CL transitioned into GL (4,906 km²) and CSL (4,827 km²), accounting for 11.54% and 11.36%, respectively. Notably, CSL increased by 4,766 km², with nearly all of the expansion originating from CL. FL saw an outflow of 23,314 km², primarily transitioning into CL (20,964 km²) and GL (2,219 km²), accounting for 89.92% and 9.52%, respectively. GL experienced an outflow of 25,665 km², transitioning primarily into FL (11,159 km²), CL (6,617 km²), and UL (7,177 km²), accounting for 43.48%, 25.78%, and 27.96%, respectively. Notably, the observed forest expansion in this study is primarily driven by the conversion of CL to FL. According to the land use transition matrix, an estimated 31,752 km² of CL has transitioned into FL, accounting for 74.72% of total CL loss, aligning closely with the policy objectives of the GFGP. Moreover, natural forests in the UYRB are predominantly distributed at elevations above 1,500 m [51–52], whereas 82.4% of the CL-to-FL conversion documented in this study occurred at elevations below this threshold. This spatial distinction underscores that the substantial FL expansion observed is primarily attributable to the direct effects of GFGP, with limited correlation to other policies such as the Natural Forest Protection Program and logging bans, which primarily focus on conserving existing FL rather than promoting CL afforestation. Nevertheless, these policies may have played a complementary role in safeguarding established forest ecosystems. Overall, the primary land use transitions involved CL, GL, and FL. These changes were driven by the implementation of GFGP, agricultural restructuring, and accelerated urbanization.

[See PDF for image]

Fig. 5

Land-use transition matrices for the different periods

Table 4. Transition matrix of land use types in UYRB (km²)

2005
2000	CL	FL	GL	WA	CSL	UL	Total
CL	216,374	11,515	3183	298	741	0	232,111
FL	10,248	375,555	665	3	11	0	386,482
GL	2573	3574	335,085	555	31	3154	344,972
WA	112	17	54	6181	43	43	6450
CSL	0	0	0	111	2915	0	3026
UL	1	0	2121	267	1	18,561	20,951
Total	229,308	390,661	341,108	7415	3742	21,758	993,992
2010
2005	CL	FL	GL	WA	CSL	UL	Total
CL	212,294	13,273	2071	406	1264	0	229,308
FL	11,410	378,374	865	2	10	0	390,661
GL	3286	3361	331,671	182	72	2536	341,108
WA	83	30	190	7005	42	65	7415
CSL	0	0	0	205	3537	0	3742
UL	0	0	1978	189	0	19,591	21,758
Total	227,073	395,038	336,775	7989	4925	22,192	993,992
2015
2010	CL	FL	GL	WA	CSL	UL	Total
CL	208,085	12,733	4223	433	1599	0	227,073
FL	13,992	379,934	1088	2	22	0	395,038
GL	2344	2941	328,026	237	52	3175	336,775
WA	197	21	322	7093	65	291	7989
CSL	0	0	1	108	4816	0	4925
UL	0	4	2719	248	1	19,220	22,192
Total	224,618	395,633	336,379	8121	6555	22,686	993,992
2020
2015	CL	FL	GL	WA	CSL	UL	Total
CL	204,221	16,841	2267	158	1129	2	224,618
FL	8657	386,142	817		17	0	395,633
GL	4419	3139	323,952	212	124	4533	336,379
WA	212	9	210	7402	29	259	8121
CSL	0	0	0	62	6493	0	6555
UL	0	3	2218	148	0	20,317	22,686
Total	217,509	406,134	329,464	7982	7792	25,111	993,992
2020
2000	CL	FL	GL	WA	CSL	UL	Total
CL	189,615	31,752	4906	999	4827	12	232,111
FL	20,964	363,168	2219	37	94	0	386,482
GL	6617	11,159	319,307	577	135	7177	344,972
WA	277	52	191	5445	90	395	6450
CSL	35	0	2	346	2643	0	3026
UL	1	3	2839	578	3	17,527	20,951
Total	217,509	406,134	329,464	7982	7792	25,111	993,992

During 2000–2005, significant transitions occurred among CL, FL, and GL (Fig. 6). CL transitioned into FL and GL, with areas of 11,515 km² and 3,183 km², respectively. FL transitioned into CL and GL, with areas of 10,248 km² and 665 km², respectively. GL transitioned into CL and FL, with areas of 2,573 km² and 3,574 km², respectively. Notably, CSL expanded by 716 km² during this period, with 741 km² of CL being converted into CSL. This indicates that the surge in CSL driven by rapid economic development came at the expense of CL. The period 2005–2010 continued the land use transition trends observed in the previous 5 years (Fig. 6). CL transitioned into FL and GL, with areas of 13,273 km² and 2,071 km², respectively. FL transitioned into CL and GL, with areas of 11,410 km² and 865 km², respectively. GL transitioned into CL and FL, with areas of 3,286 km² and 3,361 km², respectively. CL conversion to CSL reached 1,264 km², contributing to a total increase of 1,183 km² in CSL area. Between 2010 and 2015, 12,733 km² of CL transitioned into FL, while 13,992 km² of FL was converted into CL. CSL expanded by 1,630 km², with 1,599 km² of this increase originating from CL conversion. This indicates a deceleration in the implementation of GFGP in the study area during this period, while economic growth remained prominent. The period 2015–2020 marked a rapid development phase for GFGP, with significant increases in FL area (Fig. 6). CL transitioned into FL, with the area increasing to 16,841 km², while its conversion to GL dropped significantly. FL transitioned into CL and GL, with areas of 8,657 km² and 817 km², respectively. GL transitioned into CL and FL, with areas of 4,419 km² and 3,139 km², respectively. The conversion of CL to CSL slowed, with an area of 1,129 km² converted.

[See PDF for image]

Fig. 6

Spatial patterns of LUCC in each period

Land use projections under different GFGPS in the future

Drawing on land use data spanning 2000 to 2020 within the UYRB, the PLUS model was employed to forecast the spatial patterns of land use types in 2040 across four distinct scenarios (Fig. 7). Table 5 details the changes from 2020 to 2040. Under the NDS (Fig. 8), CL is expected to decline by 14,474 km² (6.65%), with the Jialing River Basin facing the steepest drop (5,416 km²). GL is anticipated to decrease by 12,948 km² (3.93%), largely in the Jinsha River Basin. FL is projected to expand by 17,662 km² (4.35%), with the Jinsha and Jialing River Basins accounting for most of this growth. CSL is expected to rise by 66% to 12,933 km², with significant expansion in the Min-Tuo River Basin. UL will grow by 17.34%. Under the mild GFGPS (Fig. 8), CSL expansion slows to just 2,529 km² (32.46%). The GFGP accelerates CL conversion to FL and reduces its conversion to CSL. Consequently, CL decreases by 17,316 km² (7.96%), while GL shrinks by 6,937 km² (2.11%). FL experiences significant growth, expanding by 19,259 km² (4.74%), and UL increases by 2,395 km² (9.54%). Under the moderate GFGPS (Fig. 8), CL conversion to FL intensifies, reducing CL by 25,483 km² (11.72%). GL erosion slows compared to the mild scenario, decreasing by 7,115 km² (2.56%). FL grows by 27,742 km² (6.83%). CSL contracts to 10,206 km², while UL mirrors the mild scenario. Under the strong GFGPS (Fig. 8), FL shows the most significant growth, adding 37,221 km². CL declines by 34,611 km², and GL decreases by 7,314 km². CSL falls further to 10,078 km², the lowest across scenarios. WA and UL remain consistent with the mild and moderate scenarios.

[See PDF for image]

Fig. 7

Projected spatial distribution of land-use classifications in 2040 under four different scenarios

Table 5. Projected land-use change areas and corresponding percentages over a 20-year period under four scenarios by 2040

Land Use Type	NDS		Mild GFGPS		Moderate GFGPS		Strong GFGPS
	Area change (km2)	Percentage change (%)	Area change (km2)	Percentage change (%)	Area change (km2)	Percentage change (%)	Area change (km2)	Percentage change (%)
CL	-14,474	-6.65	-17,316	-7.96	-25,483	-11.72	-34,611	-15.91
FL	17,662	4.35	19,259	4.74	27,742	6.83	37,221	9.16
GL	-12,948	-3.93	-6937	-2.11	-7115	-2.16	-7314	-2.22
WA	262	3.28	70	0.88	48	0.60	24	0.30
CSL	5143	66.00	2529	32.46	2414	30.98	2286	29.34
UL	4355	17.34	2395	9.54	2394	9.53	2394	9.53

[See PDF for image]

Fig. 8

Spatial patterns of LUCC under the four scenarios in 2040

Dynamics of carbon storage characteristics across the UYRB

Temporal variability of carbon storage in the UYRB

Between 2000 and 2020, the total carbon storage in the UYRB exhibited a consistent decline, with a minor recovery noted in 2020 (Table 6). Specifically, the recorded total carbon storage values for the years 2000, 2005, 2010, 2015,and 2020 were 6.387 × 10¹⁰ t, 6.382 × 10¹⁰ t, 6.379 × 10¹⁰ t, 6.369 × 10¹⁰ t, and 6.373 × 10¹⁰ t, respectively. From 2000 to 2005, there was a declined of 5.050 × 10⁷ t, followed by a further decrease of 3.178 × 10⁷ t by 2010. A significant drop of 9.644 × 10⁷ t occurred between 2010 and 2015. By 2020, however, a slight rebound was observed, with an increase of 3.493 × 10⁷ t in total carbon storage.

For 2040 under the NDS, the total carbon storage in the UYRB is anticipated to further decline to 6.359 × 10¹⁰ t (Table 7), reflecting a decrease of approximately 1.324 × 10⁸ t comparable to the levels observed in 2020. This decline is attributed to rapid economic growth, which is projected to lead to extensive erosion of arable and GL, culminating in significant carbon loss. Additionally, the swift expansion of urban areas is expected to exacerbate the carbon deficit. Although FL expansion will contribute to carbon sequestration, it will not suffice to offset the losses, resulting in substantial carbon storage reductions under the 2040 NDS.

Table 6. Carbon storage across individual sub-basins for the years 2000, 2005, 2010, 2015, and 2020 (10⁹ t)

Basin	2000	2005	2010	2015	2020
Jinsha River Basin	30.383	30.354	30.335	30.334	30.234
Minjiang-Tuojiang River Basin	10.759	10.705	10.679	10.674	10.681
Upper Yangtze River mainstream area	6.764	6.757	6.768	6.726	6.739
Wujiang River Basin	5.954	5.957	5.931	5.902	5.930
Jialingjiang River Basin	10.009	10.046	10.075	10.055	10.141

Table 7. Carbon storage of each basin for the four scenarios in 2040 (10⁹ t)

Basin	NDS	Mild GFGPS	Moderate GFGPS	Strong GFGPS
Jinsha River Basin	30.064	30.268	30.296	30.327
Minjiang-Tuojiang River Basin	10.58	10.644	10.663	10.684
Upper Yangtze River mainstream area	6.752	6.762	6.787	6.816
Wujiang River Basin	5.98	5.976	6.017	6.063
Jialingjiang River Basin	10.217	10.214	10.278	10.350

Under the mild GFGPS, the previously declining trend in carbon storage was reversed significantly. Instead of a sharp reduction, carbon storage increased by 1.385 × 10⁸ t, bringing the total carbon storage back to 6.386 × 10¹⁰ t—the highest level observed from 2000 to 2020 (Table 7). This remarkable rebound indicates the effectiveness of the GFGP in substantially enhancing carbon storage. The moderate GFGPS further supports this conclusion, with total carbon storage projected to rise appreciably by 2040. In this scenario, carbon storage is expected to reach 6.404 × 10¹⁰ t, marking a 3.157 × 10⁸ t increase compared to 2020 levels (Table 7). The strong GFGPS shows an even more substantial increase, with total carbon storage in the study area foretasted to climb to 6.424 × 10¹⁰ t by 2040—an increase of 5.136 × 10⁸ t relative to 2020 (Table 7). These findings underscore the considerable ecological benefits brought about by restoration initiatives such as the GFGP [53–54].

Spatial distribution patterns of carbon storage in the UYRB

The spatial patterns of carbon storage in the UYRB from 2000 to 2020, along with projections for 2040 under four distinct scenarios, are depicted in Figs. 9 and 10. A clear spatial gradient is evident, showing that eastern regions maintain higher carbon storage levels, which gradually decrease as one moves westward. From a watershed perspective, the Jinsha River Basin, Min-Tuo River Basin, and Jialing River Basin consistently exhibit substantially greater total carbon storage compared to the main Upper Yangtze River and Wu River Basin, where lower storage values predominate. Over the two-decade period, the Jinsha River Basin consistently recorded the highest annual average carbon storage, averaging 3.033 × 10¹⁰ t, attributable to its vast coverage. In contrast, the Wu River Basin, with an annual average of 5.935 × 10⁹ t, consistently displayed the lowest carbon storage, making it the region with the smallest total reserves.

[See PDF for image]

Fig. 9

Geographic patterns of carbon storage across the UYRB

[See PDF for image]

Fig. 10

Projected spatial patterns of carbon storage under four scenarios for 2040

The carbon storage distribution within the UYRB displays a clear northeast-to-southwest gradient, as illustrated in Fig. 9. High-carbon-storage zones are concentrated in the southeastern mountainous and hilly areas, where dense vegetation and significant forest cover predominate. By contrast, lower-carbon-storage areas are generally located in the northwest and central southern regions, encompassing the upper reaches of the Jinsha River Basin, the lower sections of the Min-Tuo and Jialing River Basins, and the upper segments of the Upper Yangtze River mainstream. GL and UL dominate the northwest, while CL and CSL are more prevalent in the central southern regions. These spatial patterns are closely linked to land use characteristics: FL corresponds to high-value carbon storage zones, whereas GL, CL, and CSL are associated with lower-value areas. From 2000 to 2020, carbon storage changes across the UYRB followed a distinct hierarchical trend: Jinsha River Basin > Jialing River Basin > Min-Tuo River Basin > Upper Yangtze River mainstream > Wu River Basin. Among these, the Jialing River Basin was unique in recording a positive carbon storage increase of 1.320 × 10⁸ t. In contrast, the Jinsha River Basin experienced a decline of 1.492 × 10⁸ t, the Min-Tuo River Basin decreased by 7.765 × 10⁷ t, the Upper Yangtze River mainstream dropped by 2.523 × 10⁷ t, and the Wu River Basin fell by 2.376 × 10⁷ t.

In the 2040 NDS (Fig. 10), substantial carbon storage deficits were observed in both the Jinsha River Basin and the Min-Tuo River Basin, with reductions of 1.704 × 10⁸ t and 1.004 × 10⁸ t, respectively. Conversely, increases in carbon storage were recorded in the main stream of the Upper Yangtze River, Wu River Basin, and Jialing River Basin. Among these, the Jialing River Basin achieved the most significant growth, adding 7.569 × 10⁷ t. Under the mild GFGPS (Fig. 10), notable enhancements in carbon storage were seen across all basins, though the Min-Tuo River Basin continued to show a deficit of 3.726 × 10⁷ t. Once again, the Jialing River Basin led the growth, with a 7.325 × 10⁷ t increase compared to 2020. Similar patterns were evident under the moderate and strong GFGPS, showing steady gains across all basins. Importantly, the Min-Tuo River Basin only achieved positive carbon storage growth under the strong GFGPS, highlighting the need to prioritize GFGP implementation in this basin for future ecological restoration efforts.

Influence of land use/cover dynamics on changes in carbon storage

Integrating carbon density and storage data with the land use transition matrix reveals a significant correlation between land use shifts and total carbon storage in the UYRB. Between 2000 and 2020, 9.629 × 10⁵ km² of land underwent transitions, accounting for 9.69% of the entire study area and leading to a net decrease of 1.438 × 10⁸ t in carbon storage. The most notable declines occurred in the Jinsha River Basin, Min-Tuo River Basin, Upper Yangtze mainstream, and Wu River Basin. In contrast, the Jialing River Basin saw a rise in carbon storage, mainly due to the conversion of low-carbon-density lands into FL with higher carbon density. Other basins, however, experienced reductions as high-carbon-density CL and GL were converted into low-carbon-density zones, including CSL, WA, and UL. Over the two decades, these low-carbon-density areas expanded by 1.046 × 10⁵ km², offsetting the gains from reforestation and emerging as the primary drivers of carbon storage loss. Key factors contributing to this reduction included shifts from CL to CSL, GL to UL, and FL to CL or GL.

Among the four 2040 scenarios, significant carbon storage declines were evident only under the NDS. In contrast, the mild, moderate, and strong GFGPS indicated substantial increases in total carbon storage. Under the NDS, the rapid expansion of low-carbon-density regions, such as CSL and UL, exacerbated carbon storage losses. However, the GFGP effectively mitigated this trend, promoting significant growth in high-carbon-density areas like FL and GL. As a result, total carbon storage in 2040 exhibited a progressive increase across the mild, moderate, and strong GFGPS.

Discussion

Analysis of driving forces for LUCC in the UYRB

This study selected 14 driving factors, including population, temperature, precipitation, digital elevation model, distance to rivers, GDP, slope, distance to railways, distance to county-level settlements, distance to roads, slope direction, soil organic matter, soil sand content, and soil pH, to analyze the driving forces behind LUCC in the UYRB from 2000 to 2020 (Fig. 11). For CL, the analysis showed that temperature, distance to roads, population, and distance to county-level settlements were the primary influencing factors. Warm climatic conditions facilitated the expansion of CL, particularly in mid-to-low altitude areas of the UYRB, where the favorable climate supported crop growth and the increase in CL. Road construction improved transportation and production efficiency in agricultural regions, further promoting CL reclamation. Population growth and settlement expansion increased the demand for CL, especially amidst accelerated urbanization, which often led to the conversion of CL to CSL, affecting its stability. For FL, digital elevation model, distance to rivers, and temperature were the dominant drivers. High-altitude regions, typically characterized by lower temperatures, were less suitable for extensive forest growth, while low-altitude areas were more conducive to forest expansion. River distribution influenced water availability for FL, with areas near rivers, especially in the low hilly regions along the Yangtze River and its tributaries, hosting richer forest ecosystems. Temperature also played a critical role, with warm climatic conditions favoring forest expansion, particularly under the implementation of GFGP, which significantly increased FL areas. GL changes were primarily influenced by temperature, digital elevation model, population, and slope. Warm climates supported GL growth and expansion, particularly in low-altitude areas where favorable conditions facilitated natural GL restoration and growth. Elevation affected GL distribution, with colder climates in high-altitude areas limiting growth, confining GL to mid-to-low altitudes. Population growth and overgrazing exacerbated GL degradation, especially in pastoral areas where sustainable use of GL faced significant pressure. Steeper slopes are more prone to soil erosion, leading to a reduction in GL areas. Changes in WA were mainly driven by slope and river distribution. In steeper areas, faster water flow limited the distribution of WA. The plains and low hilly regions of the Yangtze River Basin were more suitable for the formation and development of WA, with river direction and flow directly influencing their distribution. Changes in CSL were driven by settlements, digital elevation model, GDP, and railway development. Population growth and settlement expansion increased the demand for CSL, particularly during the urbanization process in the Yangtze River Basin, which significantly impacted other land use types. Low-altitude regions were more suitable for construction, particularly along river plains and low mountainous areas, which became major development zones for CSL. GDP growth drove regional economic development, with infrastructure construction and industrialization further promoting the expansion of CSL. Railway construction facilitated the expansion of transportation networks, leading to rapid development of CSL in areas along railway lines. Additionally, changes in the spatial pattern of UL were mainly related to elevation, temperature, and roads. Higher elevations are often associated with lower temperatures and unsuitability for cultivation, leading to larger areas of UL. Warmer climates promote the transformation of UL into other land use types. Improved transportation networks enhance the accessibility of UL, facilitating its development and reducing its overall area. The interplay of natural conditions and infrastructure development emerges as the primary driver of changes in UL. These findings highlight the combined influence of environmental factors and human activities on the spatial distribution and dynamic evolution of various land use types during regional development.

[See PDF for image]

Fig. 11

Contribution of driving factors

Responses of carbon storage changes to the GFGP in the UYRB

The UYRB, as a pivotal region for implementing China’s GFGP, has experienced significant land use transformations, critically affecting ecosystem carbon storage. From 2000 to 2020, forest coverage increased from 38.88 to 40.86%, contributing 6.394 × 10⁸ t of carbon storage to terrestrial ecosystems (Table 8). However, during the same period, total terrestrial carbon storage in the region declined from 6.387 × 10¹⁰ t to 6.373 × 10¹⁰ t, resulting in a net loss of 1.438 × 10⁸ t. Despite the considerable carbon input from the GFGP, the overall carbon balance remained negative. Without sustained GFGP implementation, the UYRB’s carbon storage is at risk of further decline. In addition, we derived the area-averaged carbon density for the study region by weighting the total carbon density of each land-use category by its proportional share. The average terrestrial carbon density also decreased, from 107.20 t/hm² in 2000 to 106.95 t/hm² in 2020. Phase-specific analysis reveals that the weakest GFGP implementation occurred between 2010 and 2015, with forest area increasing by just 604 km², contributing 4.661 × 10⁷ t to carbon storage. In contrast, the period between 2015 and 2020 marked the strongest implementation, with the largest carbon storage increase. During this time, 20,397 km² of CL was transformed, primarily into FL, with 16,841 km² converted to forests, resulting in a regional carbon storage increase of 3.391 × 10⁸ t.

Between 2020 and 2040, under the NDS, an estimated 17,779 km² of CL will be converted, with 11,136 km² transitioning to FL, contributing an additional 2.243 × 10⁸ t of carbon storage (Table 8). During this period, the average carbon density is anticipated to decline to its lowest value, 106.73 t/hm². In contrast, under the mild GFGPS, approximately 19,805 km² of CL is projected to undergo conversion, with 12,389 km² reforested, resulting in a carbon storage increase of 2.495 × 10⁸ t. This scenario sees a notable improvement in average carbon density, reaching 107.46 t/hm², highlighting the effectiveness of the GFGP. For the moderate GFGPS, projections indicate the conversion of 27,967 km² of CL, with 20,577 km² transformed into FL, contributing an increase of 4.140 × 10⁸ t in carbon storage. Average carbon density under this scenario rises further to 107.78 t/hm² (Table 8). Finally, under the strong GFGPS, approximately 37,090 km² of CL is expected to be converted, with 29,691 km² reforested, resulting in the highest carbon storage gain of 5.979 × 10⁸ t. The average carbon density in this scenario reaches its peak at 108.14 t/hm² (Table 8). It is important to note that while the Natural Forest Protection Program and Logging Ban Policy have been concurrently implemented in the region, these policies primarily focus on the conservation of existing forest resources rather than actively facilitating CL-to-FL conversion. The newly established forests in the study area predominantly originate from low-altitude CL, directly reflecting the impact of the GFGP. Therefore, estimating GFGP’s contribution to ecosystem carbon storage growth based on CL-to-FL transitions is a robust and reliable approach.

Table 8. Effect of GFGP on carbon storage in terrestrial ecosystem in UYRB

	2000–2005	2005–2010	2010–2015	2015–2020	2000–2020	2020–2040 NDS	2020-2040Mild GFGPS	2020-2040Moderate GFGPS	2020-2040Strong GFGPS
Area of GFG/km2	11,515	13,273	12,733	16,841	31,752	11,136	12,389	20,557	29,691
Carbon storage contributed by GFG/108t	2.319	2.673	2.564	3.391	6.394	2.243	2.495	4.14	5.979

Limitations and future directions

This study integrates the PLUS and InVEST models to analyze LUCC and carbon storage trends in the UYRB from 2000 to 2020 and to predict dynamics under four scenarios for 2040. The InVEST model’s capability to assess ecosystem services, combined with the spatial simulation and predictive strengths of the PLUS model, provides a comprehensive framework to evaluate the impacts of different policies on land use and carbon storage. This integrated approach offers critical insights, enabling decision-makers to better understand variations in ecosystem services across scenarios while establishing a scientific basis for ecological policy and regional land management [19, 37]. Furthermore, this coupling addresses limitations inherent in single-model predictions for future land use changes [32, 33–34, 55]. However, implementing this model combination in the UYRB is not without challenges. These include the reliance on literature-derived carbon density parameters, which may compromise accuracy due to insufficient field measurements [43], and the uncertainty introduced by the study area’s vast size, diverse land use patterns, and complex terrain [56]. Additionally, balancing computational efficiency with spatial resolution may limit the detection of subtle local changes [57]. Furthermore, while this study quantitatively attributes forest expansion primarily to the GFGP by analyzing CL-to-FL conversion data, the lack of spatially explicit datasets for other ecological policies, such as the Natural Forest Protection Program and Logging Bans Policy may limit the ability to fully disentangle their independent and synergistic effects. Finally, the land use data employed in this study has a resolution of 1 km × 1 km. While this resolution may limit the ability to capture small-scale land use changes, such as localized wetlands or small construction areas, it remains appropriate for assessing large-scale land use dynamics and carbon storage variations across the study region. Given the regional scale of this analysis, the dataset effectively captures the overall trends and major transformation patterns, ensuring a robust representation of land use changes. To overcome these challenges, future studies should focus on improving data accuracy through field-based carbon density measurements, refining model inputs to better represent intricate landscapes, and utilizing high-performance computing and multi-scale analyses to enhance efficiency [36–23, 37]. Furthermore, future research should incorporate more detailed policy implementation data and field investigations to better quantify the relative contributions of different policies to forest expansion and carbon storage changes, thereby minimizing the potential overestimation of the GFGP effects. Incorporating higher-resolution data would further refine the characterization of localized land use changes. Lastly, policymakers must also account for the influence of human activities on land carbon sequestration, emphasizing the evaluation of GFGP and other ecological restoration initiatives while formulating detailed land spatial planning strategies. Examples include protecting ecologically sensitive areas and promoting sustainable forestry or ecological agriculture in suitable regions [58]. Aligning policy assessments with natural factors will facilitate more effective ecological conservation and land use planning in the UYRB.

Conclusion

(1) Over the two decades from 2000 to 2020, CL, FL, and GL together accounted for 96.45% of the UYRB’s total area. Meanwhile, CSL steadily increased by 4,766 km², primarily through the conversion of CL. This period saw average carbon density decline to its lowest recorded value, 106.73 t/hm². Land use shifts primarily involved transitions among CL, FL, and GL. FL expanded substantially, increasing by 19,652 km², while CL and GL decreased by 14,602 km² and 15,508 km², respectively. Total carbon storage exhibited a continuous decline, with a slight recovery between 2015 and 2020. Over the 20 years, total carbon storage fell by 0.23% (1.438 × 10⁸ t), driven mainly by the conversion of CL to CSL and the shift of FL into CL or GL. Spatially, carbon storage was highest in the eastern regions and gradually decreased moving westward. The Jinsha River, Min-Tuo River, and Jialing River basins recorded the highest levels of carbon storage, while the Yangtze main stem and Wujiang River basins had the lowest levels.

(2) Among the four 2040 scenarios, only the NDS indicated a decline in total carbon storage, with an estimated loss of 1.324 × 10⁸ t. The mild, moderate, and strong GFGPS showed gains of 1.385 × 10⁸ t, 3.157 × 10⁸ t, and 5.136 × 10⁸ t, respectively. In the NDS, economic growth drove significant land conversion, causing notable carbon losses, which forest expansion partially mitigated but failed to offset entirely. The GFGPS benefited from enhanced forest protection, accelerated conversion of CL and GL to FL, and CSL restrictions, collectively boosting carbon storage. The Jialing River Basin experienced the largest gains, with increases of 7.325 × 10⁷ t, 1.369 × 10⁸ t, and 2.090 × 10⁸ t under mild, moderate, and strong GFGPS, respectively, attributed to converting low-carbon-density CL into high-carbon-density FL.

In conclusion, the UYRB, as an ecological barrier under the “dual carbon” goals, significantly contributes to national ecological security and global climate objectives. The GFGP not only increases forest coverage and optimizes land use patterns but also enhances ecosystem carbon sequestration, fostering steady growth in terrestrial carbon storage. This study offers robust scientific support for the GFGP’s continued implementation in the UYRB.

Author contributions

Conceptualization, M.P.; J.L. and Y.Y.; methodology, M.P.; Y.D.; Z.J.; H.C.; and Y.H.; software, M.P.; H.C.; Z.J.; G.L.; and Y.H.; validation, M.P., Y.Y.; H.W.; T.X.; J.L.; and G.L.; formal analysis, M.P.; H.C.; J.L. and Y.Y.; investigation, M.P., Y.Y.; Y.D.; H.W.; T.X.; J.L.; and Z.J. resources, M.P.; Y.D.; H.C.; H.W.; and Y.H.; data curation, M.P.; Y.D.; T.X.; H.C.; G.L.; and Y.H.; writing—original draft preparation, M.P., Y.Y.; J.L.; and Z.J. writing—review and editing, M.P., Y.Y.; H.W.; T.X.; J.L.; H.L.; and G.L.; visualization, M.P.; H.C.; H.L.; and Y.H.; supervision, J.L.; project administration, M.P.; funding acquisition, M.P. All authors have read and agreed to the published version of the manuscript.

Funding

This research was supported by State Key Laboratory of Oil and Gas Reservoir Geology and Exploitation (No. CDUT-PLC2024014), Opening Fund of Sichuan Key Provincial Research Base of Intelligent Tourism (No. ZHYR24-01), Open Fund of sichuan Oil and Gas Development Research Center (No. 2024SY005), Tuojiang River Basin High-quality Development Research Center (TJGZL2025-08) and Key Laboratory of Philosophy and Social Sciences in Sichuan Province — Key Laboratory for Intelligent Management and Ecological Decision Optimization of Baijiu in the Upper Reaches of the Yangtze River (No. zdsys-02).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

The authors declare no competing interests.

Competing interests

The authors declare no competing interests.

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

1. Zhou, JJ; Zhao, Y; Huang, P; Zhao, X; Feng, W; Li, Q; Dou, J; Shi, W; Wei, W; Zhu, GF; Liu, CF. Impacts of ecological restoration projects on the ecosystem carbon storage of inland river basin in arid area. China Ecol Indic; 2020; 118, 106803. [DOI: https://dx.doi.org/10.1016/j.ecolind.2020.106803]

2. Rahman, M; Zimmer, M; Ahmed, I; Donato, D; Kanzaki, M; Xu, M. Co-benefits of protecting mangroves for biodiversity conservation and carbon storage. Nat Commun; 2021; 12, 24207.1:CAS:528:DC%2BB3MXhsFKnsbrO [DOI: https://dx.doi.org/10.1038/s41467-021-24207-4]

3. He, Y; Kou, W; Chen, Y; Lai, H; Zhao, K. Returning CL to GL as a potential method for increasing carbon storage in dry-hot Valley areas. Sustainability; 2024; 16, 4150. [DOI: https://dx.doi.org/10.3390/su16104150]

4. Wang, J; Han, G; Wang, Z; Yun, J; Li, Z; Lv, S; Qin, J. Plant-soil carbon storage in dynamic succession of ecological restoration in National GL natural park. Sustainability; 2023; 15, 15837.1:CAS:528:DC%2BB3sXisFeqt7bJ [DOI: https://dx.doi.org/10.3390/su152215837]

5. Huang, B; Lu, F; Wang, X; Wu, X; Zheng, H; Su, Y; Yuan, Y; Ouyang, Z. The impact of ecological restoration on ecosystem services change modulated by drought and rising CO2. Glob Chang Biol; 2023; 29, pp. 5304-20.1:CAS:528:DC%2BB3sXhtlKrtLnK [DOI: https://dx.doi.org/10.1111/gcb.16825]

6. Li, H; Advancing,. Carbon peak and carbon neutrality in china: A comprehensive review of current global research on carbon capture, utilization, and storage technology and its implications. ACS Omega; 2023; 8, 45 pp. 42086-101.1:CAS:528:DC%2BB3sXit1Gltb%2FE [DOI: https://dx.doi.org/10.1021/acsomega.3c06422]

7. Liang, Y; Hashimoto, S; Liu, L. Integrated assessment of land-use/land-cover dynamics on carbon storage services in the loess plateau of China from 1995 to 2050. Ecol Indic; 2021; 120, 106939.1:CAS:528:DC%2BB3cXhvFGqs77L [DOI: https://dx.doi.org/10.1016/j.ecolind.2020.106939]

8. Liang, Y; Liu, L; Huang, J. Integrating the SD-CLUE-S and invest models into assessment of Oasis carbon storage in Northwestern China. PLoS ONE; 2017; 12, e0172494.1:CAS:528:DC%2BC2sXhsFyjtrbP [DOI: https://dx.doi.org/10.1371/journal.pone.0172494]

9. Zhang, Y; Liao, X; Sun, D. A coupled InVEST-PLUS model for the Spatiotemporal evolution of ecosystem carbon storage and multi-scenario prediction analysis. Land; 2024; 13, 509. [DOI: https://dx.doi.org/10.3390/land13040509]

10. Zhao, M; He, Z; Du, J; Chen, L; Lin, P; Fang, SB. Assessing the effects of ecological engineering on carbon storage by linking the CA-Markov and invest models. Ecol Indic; 2019; 96, pp. 1027-52. [DOI: https://dx.doi.org/10.1016/j.ecolind.2018.10.052]

11. Li K, Cao J, Adamowski J, Biswas A, Zhou JJ, Liu Y, Zhang Y, Liu CF, Dong X, Qin Y. Assessing the effects of ecological engineering on Spatiotemporal dynamics of carbon storage from 2000 to 2016 in the loess plateau area using the invest model: A case study in Huining county, China. Environ Dev. 2021;100641. https://doi.org/10.1016/j.envdev.2021.100641.

12. Li, P; Chen, J; Li, Y; Wu, W. Using the InVEST-PLUS model to predict and analyze the pattern of ecosystem carbon storage in Liaoning province, China. Remote Sens; 2023; 15, 16 19. [DOI: https://dx.doi.org/10.3390/rs15164050]

13. Li, YH; Yao, S; Jiang, H; Wang, H; Ran, Q; Gao, X; Ding, X; Ge, D. Spatial-Temporal evolution and prediction of carbon storage: an integrated framework based on the MOP–PLUS–InVEST model and an applied case study in hangzhou, East China. Land; 2022; 11, 2213. [DOI: https://dx.doi.org/10.3390/land11122213]

14. He, C; Zhang, D; Huang, Q; Zhao, Y. Assessing the potential impacts of urban expansion on regional carbon storage by linking the LUSD-urban and invest models. Environ Model Softw; 2016; 75, pp. 44-58. [DOI: https://dx.doi.org/10.1016/j.envsoft.2015.09.015]

15. Li, Y; Yao, S; Jiang, H; Wang, H; Ran, Q; Gao, X; Ding, X; Ge, D. Spatial-temporal evolution and prediction of carbon storage: an integrated framework based on the MOP–PLUS–InVEST model and an applied case study in hangzhou, East China. Land; 2022; 11, 2213. [DOI: https://dx.doi.org/10.3390/land11122213]

16. Yu, Z; Zhao, MS; Gao, Y; Wang, T; Zhao, ZD; Wang, SH. Spatio-temporal evolution and prediction of carbon storage in Huaibei City based on InVEST-PLUS model. Environ Sci; 2024; 45, pp. 3270-83.1:CAS:528:DC%2BB2MXhslGmt7bO [DOI: https://dx.doi.org/10.13227/j.hjkx.202306110]

17. Ding, Y; Wang, LZ; Gui, F; Zhao, S; Zhu, W. Ecosystem carbon storage in Hangzhou Bay area based on invest and PLUS models. Environ Sci; 2023; 44, pp. 3343-52. [DOI: https://dx.doi.org/10.13227/j.hjkx.202204080]

18. Wang, C; Li, T; Guo, X; Xia, L; Lu, C; Wang, C. Plus-InVEST study of the Chengdu-Chongqing urban agglomeration’s land-use change and carbon storage. Land; 2022; 11, 1617. [DOI: https://dx.doi.org/10.3390/land11101617]

19. Wang, Q; Zhang, W; Xia, J; Ou, D; Tian, Z; Gao, X. Multi-scenario simulation of land-use/land-cover changes and carbon storage prediction coupled with the SD-PLUS-InVEST model: A case study of the Tuojiang river basin, China. Land; 2024; 13, 1518.1:CAS:528:DC%2BB2cXhsVWmsb3E [DOI: https://dx.doi.org/10.3390/land13091518]

20. Jiao, Y; Wang, Y; Tu, C; Hou, X; Lyu, C; Fan, X; Xia, L. Spatiotemporal evolution and future of carbon storage in resource-based Chinese province: A case study from Shanxi using PLUS–InVEST model prediction. Sustainability; 2024; 16, 4461.1:CAS:528:DC%2BB2cXhsVejtbjL [DOI: https://dx.doi.org/10.3390/su16114461]

21. Kong, R; Zhang, ZX; Zhang, F; Tian, J; Chang, J; Jiang, S; Zhu, B; Chen, X. Increasing carbon storage in subtropical forests over the Yangtze river basin and its relations to the major ecological projects. Sci Total Environ; 2019; 709, 136163.1:CAS:528:DC%2BB3cXhtlKhsQ%3D%3D [DOI: https://dx.doi.org/10.1016/j.scitotenv.2019.136163]

22. Wang, J; Delang, CO; Hou, G; Gao, L; Lu, X. Net primary production increases in the Yangtze river basin within the latest two decades. Glob Ecol Conserv; 2021; 26, e01497. [DOI: https://dx.doi.org/10.1016/j.gecco.2021.e01497]

23. Jiang, Y; Ouyang, B; Yan, Z. The response of carbon storage to multi-objective land use/cover Spatial optimization and vulnerability assessment. Sustainability; 2024; 16, 2235.1:CAS:528:DC%2BB2cXnvFahur0%3D [DOI: https://dx.doi.org/10.3390/su16062235]

24. Zhang, Y; Wang, Y; Fu, B; Dixit, A; Chaudhary, S; Wang, S. Impact of Climatic factors on vegetation dynamics in the upper Yangtze river basin in China. J Mt Sci; 2020; 17, pp. 1235-50. [DOI: https://dx.doi.org/10.1007/s11629-019-5649-7]

25. Liu, F; Gao, H; Pan, B; Li, Z; Su, H. Quantitative analysis of planation surfaces of the upper Yangtze river in the Sichuan-Yunnan region, Southwest China. Front Earth Sci; 2018; 13, pp. 55-74.1:CAS:528:DC%2BC1cXhtFemsL7K [DOI: https://dx.doi.org/10.1007/s11707-018-0707-y]

26. Yuan Z, Wang Y, Xu J, Wu Z. Effects of Climatic factors on the net primary productivity in the source region of Yangtze river. China Sci Rep. 2021;11. Article 2651. https://doi.org/10.1038/s41598-020-80494-9.

27. Qu, S; Wang, L; Lin, A; Zhu, H; Yuan, M. What drives the vegetation restoration in Yangtze river basin, china: climate change or anthropogenic factors? Ecol. Indic; 2018; 90, pp. 438-50. [DOI: https://dx.doi.org/10.1016/J.ECOLIND.2018.03.029]

28. Deng, W. Influence of regional climate change on vegetation cover in the middle and lower Yangtze river basin. Res Environ Sci; 2014; 27, pp. 1032-42. [DOI: https://dx.doi.org/10.13198/j.issn.1001-6929.2014.09.12]

29. Yang, J; Huang, X. The 30 m annual land cover datasets and its dynamics in China from 1985 to 2022. Earth Syst Sci Data; 2023; 13, pp. 3907-25. [DOI: https://dx.doi.org/10.5281/zenodo.8176941]

30. Wang, ZH; Shi, XG; Dou, ST; Cheng, MM; Miao, LL. The 30 m land cover dataset for capturing land cover changes induced by ecological restoration from 1990 to 2022 on the Chinese Loess 2025; 10.1038/s41597-025-04575-y

31. Xu, L; Liu, X; Tong, D; Liu, Z; Yin, L; Zheng, W. Forecasting urban land use change based on cellular automata and the PLUS model. Land; 2022; 11, 652. [DOI: https://dx.doi.org/10.3390/land11050652]

32. Liang, X; Guan, Q; Clarke, K; Liu, S; Wang, B; Yao, Y. Understanding the drivers of sustainable land expansion using a patch-generating land use simulation (PLUS) model: A case study in wuhan, China. Comput Environ Urban Syst; 2021; 85, 101569. [DOI: https://dx.doi.org/10.1016/j.compenvurbsys.2020.101569]

33. Zhou, L; Dang, X; Sun, Q; Wang, S. Multi-scenario simulation of urban land change in Shanghai by random forest and CA-Markov model. Sustain Cities Soc; 2020; 55, 102045. [DOI: https://dx.doi.org/10.1016/j.scs.2020.102045]

34. Xing, W; Qian, Y; Guan, X; Yang, T; Wu, H. A novel cellular automata model integrated with deep learning for dynamic spatio-temporal land use change simulation. Comput Geosci; 2020; 137, 104430. [DOI: https://dx.doi.org/10.1016/j.cageo.2020.104430]

35. Yu, X; Xiao, J; Huang, K; Li, Y; Lin, Y; Qi, G; Ren, P. Simulation of land use based on multiple models in the Western Sichuan plateau. Remote Sens; 2023; 15, 2142. [DOI: https://dx.doi.org/10.3390/rs15082142]

36. Zhu, K; Cheng, Y; Zang, W; Zhou, Q; El Archi, Y; Mousazadeh, H; Kabil, M; Csobán, K; Dávid, L. Multi-scenario simulation of Land-Use change in Hubei province, China based on the Markov-FLUS model. Land; 2023; 12, 744. [DOI: https://dx.doi.org/10.3390/land12040744]

37. Yulianto, F; Maulana, T; Khomarudin, MR. Analysis of the dynamics of land use change and its prediction based on the integration of remotely sensed data and CA-Markov model, in the upstream citarum watershed, West java, Indonesia. Int J Digit Earth; 2018; 12, pp. 1151-76. [DOI: https://dx.doi.org/10.1080/17538947.2018.1497098]

38. Lü, D; Fu, J; Mao, D; Li, W; Zhou, W; Liu, Z; Zhao, W; Sun, G; Chen, H; Wang, X; Li, X. An effective accuracy assessment indicator for credible land use change modelling. Ecol Indic; 2020; 114, 106552. [DOI: https://dx.doi.org/10.1016/j.ecolind.2020.106552]

39. Liang, X; Guan, Q; Liu, Q; Clarke, KC; Liu, S; Wang, B; Yao, Y. Understanding the drivers of sustainable land expansion using a Patch-Generating land use simulation (PLUS) model: A case study in wuhan, China. Comput Environ Urban Syst; 2021; 85, 101569.

40. Deng, Y; Yao, S; Hou, M; Zhang, T; Lu, Y; Gong, Z; Wang, Y. The impact of the grain for green project on ecosystem carbon storage services: A case study of Zichang County in the loess plateau hilly and gully region. J Nat Resour; 2020; 35, pp. 826-44.

41. Guo Z, Sharma H, Jadav MB, Zhang W, Spatially explicit ecosystem service assessment using invest model. mapping above ground carbon storage and sequestration in thoria watershed, India: A. 2022 10th Int. Conf. Agro-Geoinformatics. 2022; 1–5. https://doi.org/10.1109/Agro-Geoinformatics55649.2022.9858976

42. Xie, X; Sun, B; Zhou, H; Li, Z. Soil organic carbon storage and their influencing factors under different vegetation types in China. Acta Pedol Sin; 2004; 41, pp. 687-99.

43. Li, M; Luo, H; Qin, Z; Tong, Y. Spatial-temporal simulation of carbon storage based on land use in Yangtze river delta under SSP-RCP scenarios. Land; 2023; 12, 399. [DOI: https://dx.doi.org/10.3390/land12020399]

44. Zhu, C; Zhao, S; Zhou, D. Estimation of organic carbon storage in Built-Up areas of China from 1997 to 2006. Chin J Appl Ecol; 2012; 23, pp. 1195-202.1:CAS:528:DC%2BC38XhtVahur3K

45. Wang CY. Simulation study on spatial and temporal changes of land space and carbon storage of chengdu-chongqing urban agglomeration based on PLUS-InVEST. Master’s Thesis, Chongqing Three Gorges University, Chongqing, China, 2023.

46. Fang, JY; Guo, ZD; Piao, SL; Chen, AP. Estimation of terrestrial vegetation carbon sinks in China from 1981 to 2000. Sci China Ser D Earth Sci; 2007; 50, pp. 804-12. [DOI: https://dx.doi.org/10.1007/s11430-007-0030-0]

47. Xu, Q; Rui, WY; He, H; Zhang, XP; Wang, JK; Zhang, HL. Characteristics and regional differences of soil organic carbon density in china’s farmland under different utilization modes. Sci Agric Sin; 2006; 39, pp. 2505-10. [DOI: https://dx.doi.org/10.3864/j.issn.0578-1752.2006.12.010]

48. Wan, X; Ji, GX; Chen, WQ; Zhang, YL; Huang, JC; Guo, YL; Chen, YN. Simulation and prediction of Spatiotemporal evolution of ecosystem carbon storage in the Yangtze river basin based on PLUS-InVEST model. J Agric Resour Environ; 2023; [DOI: https://dx.doi.org/10.13254/j.jare.2022.0841]

49. Chen, G; Yang, Y; Xie, J; Du, Z; Zhang, J. The allocation of underground carbon in china’s forests. Acta Ecol Sin; 2007; 27, pp. 5148-57. (In Chinese)

50. Alam, SA; Starr, M; Clark, BJF. Tree biomass and soil organic carbon densities across the Sudanese woodland savannah: A regional carbon sequestration study. J Arid Environ; 2013; 89, pp. 67-76. [DOI: https://dx.doi.org/10.1016/j.jaridenv.2012.10.002]

51. Zhu, WZ; Wang, YK; Fan, JR; Sheng, X; Wei, Z. Types and distribution of priority conservation ecosystems in the upper reaches of the Yangtze river. J Mt Sci; 2011; 29, 5 pp. 520-8.1:CAS:528:DC%2BC38XnsVOisr4%3D [DOI: https://dx.doi.org/10.1029/2011JD016494]

52. Dong, X; Yuan, Q; Kou, Y; Li, S; Ren, P. Distribution and ecological network construction of National natural protected areas in the upper reaches of Yangtze river. Sustainability; 2023; 15, 1012. [DOI: https://dx.doi.org/10.3390/su15021012]

53. Xiang, W; Xu, L; Lei, P; Ouyang, S; Deng, X; Chen, L; Zeng, Y; Hu, Y; Zhao, Z; Wu, H; Zeng, L; Xiao, W. Rotation age extension synergistically increases ecosystem carbon storage and timber production of Chinese Fir plantations in Southern China. J Environ Manag; 2022; 317, 115426.1:CAS:528:DC%2BB38XhvFyitLzL [DOI: https://dx.doi.org/10.1016/j.jenvman.2022.115426]

54. Tian, H; Zhu, J; Jian, Z; Ou, Q; He, X. The carbon neutral potential of forests in the Yangtze river economic belt of China. Forests; 2022; 13, 721. [DOI: https://dx.doi.org/10.3390/f13050721]

55. Zhao, SH; Zhou, DM; Wang, DM; Chen, JK; Gao, YJ; Zhang, J; Jiang, J. Ecosystem carbon storage assessment and multi-scenario prediction in the Weihe river basin based on PLUS-InVEST model. J Appl Ecol; 2024; 35, pp. 2044-54.1:CAS:528:DC%2BB2MXhslCntb%2FI [DOI: https://dx.doi.org/10.13287/j.1001-9332.202408.023]

56. Yang, X; Qian, B; Ji, G; Chen, W; Huang, J; Guo, Y; Chen, Y. Characteristics of Spatial and Temporal changes in carbon storage in the middle and upper reaches of the Huaihe river basin and future multi-scenario simulation prediction. Environ Sci; 2024; 45, pp. 5970-82. [DOI: https://dx.doi.org/10.13227/j.hjkx.202311034]

57. Zhang, Y; Ding, H; Xu, J; Zheng, B. A simulation-based prediction of land use change impacts on carbon storage from a regional imbalance perspective: A case study of Hunan province, China. Land; 2024; 13, 1721. [DOI: https://dx.doi.org/10.3390/land13101721]

58. Zhou, W; Wang, J; Han, Y; Que, H; Yang, L; Wang, R. Scenario simulation of the relationship between land-use changes and ecosystem carbon storage: A case study in Dongting lake basin, China. Int J Environ Res Public Health; 2023; 20, 4835. [DOI: https://dx.doi.org/10.3390/ijerph20064835]