Multi-Scenario Simulation of Land Use Change and Ecosystem Health Assessment in Chengdu Metropolitan Area Based on SD-PLUS-VORS Coupled Modeling

Abstract

Human activities exert considerable influence on ecosystem health, a cornerstone for fostering sustainable regional growth, largely through their effects on land use transformations. This study integrates a system dynamics (SD) model with the patch-generating land use simulation (PLUS) model and the VORS (vigor–organization–resilience–ecosystem services) model to simulate the spatiotemporal dynamics of land use/cover change (LUCC) and assess ecosystem health in the Chengdu Metropolitan Area (CMA) from 2020 to 2035. These projections were conducted under three distinct scenarios: the ecological protection scenario (EPS), the natural development scenario (NDS), and the economic development scenario (EDS). The findings indicate the following: (1) Under EPS, NDS, and EDS, both cultivated land and grassland areas decline, and construction land expands by 40.68%, 54.76%, and 75.01%, respectively. (2) Across all three scenarios, ecosystem health demonstrates improvement, and it shifts from “poor” to “moderate.” (3) Ecosystem health levels in the CMA demonstrate significant spatial heterogeneity; they exhibit “low” levels in the central city, while generally stable levels are observed throughout the surrounding region. These results offer a strong scientific foundation for cultivating sustainable land management strategies and protecting ecosystem health in the CMA.

1. Introduction

Natural ecosystems are vital for human welfare and progress, providing crucial resources and ecosystem services (ES) [1,2]. Sustaining ecosystem health holds critical significance for achieving both economically and socially sustainable development [3]. Human activities have had a profound impact on land use patterns. Land use/cover change (LUCC) exerts a significant impact on ecosystem services by modifying habitat structures, water cycles, and soil properties [4]. Traditional land use models often focus solely on urban expansion or agricultural conversion without fully considering their ecological consequences. Consequently, it is essential to adopt a comprehensive method that concurrently forecasts LUCC and assesses ecosystem health. In the face of rapid urbanization, research has focused on how to protect the health of ecosystems [5]. In 2020, China initiated its 14th Five-Year Plan for natural resource conservation and sustainable stewardship, and it highlighted the significance of forming ecological protection boundaries while designating fundamental cultivated land preservation zones [4]. Therefore, analyzing temporal and regional variations in land use and ecosystem health is essential for the region’s sustainable development, promoting the harmonious integration of human and natural systems.

In recent decades, a variety of methodologies have been developed for the evaluation of ecosystem health, including the pressure–state–response (PSR) model [6], the vigor–organization–resilience (VOR) model [7], and the vigor–organization–resilience–ecosystem services (VORS) model [8]. The VOR framework, initially proposed by Rapport (1989), has been widely applied for assessing ecosystem conditions and resilience to external disturbances [9]. The assessment of ecosystem health is contingent upon the consideration of ecosystem services; however, this was not factored into the VOR framework [10,11]. In 2021, Costanza et al. proposed the incorporation of ecosystem services (ES) into the VORS model framework and established the Ecosystem Health Index (EHI) evaluation system, which comprises vigor (ecosystem biological productivity), organization (ecosystem structural complexity), resilience (capacity to recover from disturbances), and ecosystem services (benefits ecosystems provide for human well-being) [12]. This framework provides a more objective and scientific evaluation and is currently widely used in regional ecosystem health evaluations [13,14]. Existing research generally agrees that LUCC has a significant impact on ecosystem health [15,16]. Consequently, coupled modeling approaches and scenario analysis have been widely used to predict future ecosystem health under LUCC dynamics [17,18]. Presently, multi-scenario LUCC simulations are evaluated within the context of scenario development, demand forecasting, and spatial allocation studies. When it comes to demand projections, models like the system dynamics (SD) model, Markov chains, and linear programming are commonly utilized. The SD model is particularly adept at depicting the nonlinear, systematic, complex, and dynamic attributes of land use change processes; thus, this study uses the SD model to simulate land use demand [19]. Regarding spatial simulation, models that are in wide use consist of the cellular automata (CA) [17], future land use simulation (FLUS) model [20], and the patch-generating land use simulation (PLUS) model [21]. Wang et al. conducted a comparative research in which the PLUS model was evaluated against the CA Markov model and the FLUS model, revealing that PLUS exhibited superior accuracy [22]. Therefore, this study used the PLUS model to simulate land use spatial demands to help researchers gain a deeper understanding of the changing state of ecosystems; there is increasing emphasis on using these complex land use simulation tools to predict their future ecosystem conditions [13]. For instance, Pan et al. [23] leveraged the FLUS model to explore how climate change and LUCC interact to influence the EHI under various Representative Concentration Pathway (RCP) scenarios. Li et al. [24] utilized the IBIS and VORS frameworks to evaluate EHI and examine their spatial and temporal patterns. Together, these investigations underscore both the practicality and scholarly value of assessing ecosystem health through land use analysis. This study employed the SD model to forecast land use demand, whereas the simulation of spatial land use alterations mostly utilized the PLUS model. The VORS model was employed to amalgamate many variables, including vigor, organization, resilience, and ecosystem services, to assess ecosystem health. In contrast to conventional models, the SD-PLUS-VORS coupled model offers a more thorough way of assessing ecosystem health by incorporating dynamic land demand forecasting, high-resolution land use spatial modeling, and multidimensional ecosystem health evaluation. This makes it particularly suitable for scenario-based analysis in complex urban environments, providing valuable scientific support for sustainable land use planning and ecological management.

In 2021, the Sichuan Provincial Government formally launched the “Chengdu Metropolitan Area (CMA) Development Plan”. This plan is the third metropolitan area development strategy to receive national approval, following those of the Nanjing and Fuzhou metropolitan areas. The biological environment in this region is complex and fragile. Therefore, meticulous consideration is necessary regarding the impact of ecosystem health on the city’s sustainable development. Investigations into ecosystem health assessment within the CMA are scarce, especially concerning the spatiotemporal dynamics of LUCC and their implications for ecosystem health. Previous studies have predominantly focused on a singular aspect, whether it is forecasting land use or assessing ecosystem health. This inquiry adopts a novel strategy by integrating these two approaches, offering a comprehensive assessment of the impact of land use changes on ecosystem health. The SD-PLUS-VORS coupled model framework integrates system dynamics, spatial land use simulation, and ecosystem health assessment into a unified study, facilitating a more precise and comprehensive understanding of future ecological changes. The research findings provide a solid scientific foundation for promoting sustainable land practices and ecological management in the CMA while also presenting a scalable model for other rapidly growing urban centers facing similar environmental issues. This study provides critical insights and considerable benefits for improving the ecological well-being of other metropolitan areas.

2. Materials and Methods

2.1. Study Area

The CMA (Figure 1) is situated in the southeastern region of Sichuan Province (102°54′–105°01′ E, 29°24′–31°42′ N), and its area encompasses approximately 33,109 square kilometers. This region encompasses 35 administrative units, blending both municipal- and county-level areas under its governance. Situated in a humid subtropical monsoon region, the CMA boasts a mild climate with clearly defined seasons. Much of the remaining land is flat, crisscrossed by an extensive river system and expansive agricultural fields. Blessed with abundant natural resources, the CMA enjoys a robust ecological foundation. Recognized as the “hub” urban zone in Sichuan Province’s Main Functional Area Plan, the health of its ecosystem not only influences local biodiversity and environmental conditions but also the long-term economic growth of the entire province.

2.2. Data Acquisition and Processing

The SD model utilizes land use data from four distinct time intervals, supplemented with pertinent information obtained from the Sichuan Statistical Yearbook, China City Statistical Yearbook, and China County Statistical Yearbook, spanning the years 2005 to 2020. The PLUS model necessitates land use raster data alongside driving factors, which include socioeconomic, natural, and accessibility-related variables. The land use raster data at a 30 × 30 m resolution for 2005 and 2020 were obtained from the CNLUCC Remote Sensing Monitoring Dataset. To ensure data consistency and comparability, all spatial datasets were processed in ArcGIS 10.8, following a specified methodology. Initially, the unification of the coordinate system was executed utilizing the Define Projection tool, standardizing all datasets to WGS_1984_UTM_Zone_48N. Subsequently, the datasets, characterized by varying spatial resolutions, were subjected to reprojection and resampling procedures. All raster data were resampled uniformly to a resolution of 30 m using the Bilinear Interpolation method. Cropping operations were subsequently performed to align with the study area’s boundaries, followed by raster resampling to ensure uniformity in row and column counts across all land use datasets. These measures eradicated possible inconsistencies in spatial coverage and guaranteed that all datasets were geometrically and thematically synchronized for further analyses. The Reclassify tool in ArcGIS 10.8 was utilized for land use classification, categorizing the data into six primary classes.

Key drivers are essential in shaping land use changes, highlighting the necessity for a solid and scientifically based framework of underlying elements to improve the accuracy of predicting land use development potential. This work utilizes prior research to identify 16 driving elements based on accessibility, spatial variation, impact, and measurability. The random forest method in the LEAS module of the PLUS model was employed to assess variable significance and derive spatial rules, therefore mitigating multicollinearity and redundancy. This strategy efficiently eliminates less pertinent factors, guaranteeing robust and dependable simulation outcomes [25]. The source of each driver is listed in

Table 1. Data information.

Category	Data Name	Data Type	Data Source
Land Use Data	2005 and 2020 Land Use Data	Raster/30 m	Chinese Academy of Sciences Resource and Environmental Science Data Center (https://www.resdc.cn, accessed on 4 January 2025).
Administrative Boundary	City Administrative Boundary	-	Resource and Environment Science Data Center (http://www.resdc.cn, accessed on 4 January 2025).
Natural Factor Drivers	Slope	Raster/30 m	Chinese Academy of Sciences Resource and Environmental Science Data Center (https://www.resdc.cn, accessed on 4 January 2025).
	DEM
	Average Annual Temperature
	Average Annual Precipitation
	Soil Type
Socioeconomic Factor Drivers	Population Density	Raster/1 km	Chinese Academy of Sciences Resource and Environmental Science Data Center (https://www.resdc.cn, accessed on 4 January 2025).
	GDP
Distance Factor Drivers	Distance from Primary Road	Raster/30 m	National Geographic Information Resources Catalog Service System (https://www.webmap.cn/main.do?method=index, Accessed on 5 January 2025)
	Distance from Secondary Road
	Distance from Tertiary Road
	Distance from Railroad
	Distance from Highway
	Distance from the County Government
	Distance from Settlement
	Distance from Water
	Distance from Industrial Area
Statistical Data	Economic Data, Demographic Data, and Annual Average Temperature and Precipitation Data for the Simulation Phase.	-	Chengdu Statistical Yearbook 2005–2020 (https://cdstats.chengdu.gov.cn/cdstjj/c155008/list.shtml, accessed on 7 January 2025)
			Deyang Statistical Yearbook 2005–2020 (https://data.dystat.cn/list/145/d/1/subsite/1.html, accessed on 7 January 2025)
			Meishan Statistical Yearbook 2005–2020 (https://www.ms.gov.cn/zfxxgk/fdzdgknr/tjxx/tjnj.htm, accessed on 7 January 2025)
			Ziyang Statistical Yearbook 2005–2020 (http://www.ziyang.gov.cn/zysrmzf/tjnj/pc/list.html, accessed on 7 January 2025)

. All spatial data were processed with ArcGIS 10.8, where the Define Projection tool was applied to standardize coordinate systems, unify spatial resolutions, and compute Euclidean distances. Moreover, projection coordinate systems and resolutions were synchronized with the land use raster data to ensure uniformity. These preprocessing steps ensured spatial uniformity across all datasets, minimizing errors caused by variations in resolution or coordinate systems and thereby enhancing the reliability of subsequent analyses.

2.3. Methods

The framework is structured around three key elements: an SD-driven simulation for land use demand, a PLUS-based approach for spatial land use simulation, and a VORS-guided assessment of ecosystem health (Figure 2). The SD model was employed to forecast land use needs in diverse development scenarios. The anticipated demands were subsequently input into the PLUS model, which utilized 16 driving factors to compute land use change probabilities through random forest-based rule mining. The spatial allocation was guided by land expansion analysis and neighborhood weighting procedures, resulting in the anticipated land use distribution for 2035. The VORS model was employed to evaluate ecosystem health, producing the EHI and analyzing the results to examine spatiotemporal changes and transitions in ecosystem health across various land use scenarios.

2.3.1. Scenario Setting

Three development scenarios were formulated after careful consideration of the temporal and spatial shifts in land use, the pivotal drivers of the CMA, regional socioeconomic trends, and future growth strategies, with reference to extant studies [26,27]. The three scenarios are the ecological protection scenario (EPS), the natural development scenario (NDS), and the economic development scenario (EDS).

• EPS (ecological protection scenario): This scenario prioritizes the containment of urban sprawl. It focuses on the creation of an ecological conservation-centered land protection system, the enforcement of stringent ecological land protection policies, and increased ecological management investment. In accordance with the Chengdu Metropolitan Circle Territorial Spatial Plan (2021–2035), stringent restrictions are imposed on the transformation of ecological land for urban development.
• NDS (natural development scenario): Based on the trend of socioeconomic shifts in the CMA between 2005 and 2020, this scenario presumes no new economic or environmental policy interventions and continues the land use transfer patterns witnessed during this period. The projected land demand, therefore, represents the organic progression of change in the CMA.
• EDS (economic development scenario): Unregulated urban expansion and a completely open population policy have spurred rapid population increases in the CMA, exacerbating competition for land resources. Therefore, land demand has transformed under extensive use.

2.3.2. Development of an SD Model for Quantitative Land Demand Forecasting

The SD model offers a robust mechanism for conducting dynamic simulations of complex systems. These experiments are designed to appraise system behavior and developments under a range of scenarios with varying parameters or strategic choices. The model’s core purpose is to aid decision-making for users [28]. The CMA land use demand system comprises four subsystems: economy, population, land, and climate (Figure 3). Economic development results in higher fixed asset investments, stimulating growth in both the agricultural and construction sectors. This growth increases the cultivated and constructed land areas. The population subsystem reflects demographic changes. Meanwhile, population expansion generates an increased appetite for agricultural products. Consequently, this demand influences alterations in land utilization. The climate subsystem takes into consideration how temperature and precipitation affect land resources such as cultivated land, forested areas, grasslands, and water. At the same time, the land subsystem reflects transitions in land use categories. The model simulation covers 2005 to 2035, with a yearly time step. The 2005–2020 period is the model calibration and historical simulation phase, while 2020–2035 is focused on projections. Through iterative optimization and continuous optimization, researchers identified the relationships among the variables. This process resulted in the final model, constructed utilizing Vensim PLE 7.3.5 software.

The SD model parameter adjustments (

Table 2. Scenario-specific parameter configurations in the SD model.

Scenario	Mean Annual GDP Growth Rate (%)	Mean Annual Population Growth Rate (‰)	Mean Annual Urbanization Rate (%)	Mean Annual Precipitation (mm)	Mean Annual Temperature (°C)
EPS	6	4.3	0.62	5.45	0.12
NDS	Keep Constant	Keep Constant	Keep Constant	Keep Constant	Keep Constant
EDS	7.5	7.2	0.84	7.14	0.45

) were grounded in three scenarios. In the NDS, the government aims for an annual GDP growth rate of 6%, while in the economic development scenario, the target is 7.5%. With the three-child policy’s implementation, population growth rates are projected at 4.3‰ under the EPS and 7.2‰ under the EDS. The Chengdu Metropolitan Circle Territorial Spatial Plan (2021–2035) projects a 75% urbanization rate for the CMA by 2035. After a thorough assessment, annual urbanization rates are estimated to increase by 0.62% and 0.84% under the EPS and HDS, respectively. Actual temperature and precipitation data from the CMA spanning 2005–2020, along with prior research, were utilized [29], and annual temperature and precipitation fluctuations are expected to intensify under both the EPS and EDS. In the NDS, these parameters are expected to maintain their original growth rates.

2.3.3. Spatiotemporal Changes in Land Use Across Multiple Scenarios Using the PLUS Model

In contrast to the CA model and other prevalent models, the PLUS model’s advantage lies in its examination of the mechanisms underlying land use change during simulation, as well as its ability to simulate the generation and evolution of random patch seeds without temporal and spatial constraints, resulting in more precise outcomes [30]. The PLUS model comprises two components: the LEAS model and the CARS model [31]. LEAS utilizes a random sampling approach, selecting 5% of land use data from regions experiencing growth during both stages of land use transformation. The random forest algorithm is then applied to break down the most crucial elements and influences shaping different land use categories, allowing for the estimation of development probabilities for each specific type [32]. This research takes into account 16 drivers, each symbolized by 16 respective features. At a default sampling rate of 0.01, roughly 1% of the pixels are earmarked for training, and 20 random forests are utilized. This method calculates development probabilities for each land use type, and it offers a comprehensive, quantitative understanding of influential factors that impact land use decisions.

The CARS model integrates a randomized seed generation process alongside a gradually diminishing threshold system. Research indicates that a spatial resolution of 30 m strikes the perfect balance for maximizing precision in land use analysis. For patch creation and expansion, coefficients are set at 0.9 and 0.1, respectively, with operations conducted within a 3 × 3 grid structure [30]. Informed by the relevant literature and the study scope [33], neighborhood weight factors under different scenario modes were determined through iterative adjustment and validation, as detailed in

Table 3. Neighborhood weights for different scenario models.

Land Use Type	EPS	NDS	EDS
Cultivated land	0.4	0.6	0.4
Forest land	1	0.7	0.3
Grassland	0.7	0.3	0.2
Water	0.8	0.3	0.6
Unused land	0.1	0.5	0.1
Construction land	0.8	1	1

. The modeling of land use transitions employs a transition matrix, in which “0” denotes the absence of conversion potential between two land use types, while “1” signifies the possibility of such conversion. After studying the land use transition matrix of the CMA from 2005 to 2020, along with insights from previous research, a series of modifications were executed. Consequently, a conversion cost restriction matrix was devised for each scenario, as illustrated in

Table 4. Land use transformation matrix (A, B, C, D, E, and F represent cultivated land, forest land, grassland, water, unused land, and construction land, respectively).

	EPS						NDS						EDS
	A	B	C	D	E	F	A	B	C	D	E	F	A	B	C	D	E	F
A	1	1	1	1	1	1	1	1	1	1	1	1	1	0	0	0	1	1
B	0	1	0	0	0	0	1	1	1	1	1	1	1	1	1	0	1	1
C	0	1	1	0	0	0	1	1	1	1	1	1	0	0	1	1	0	1
D	0	1	1	1	0	0	1	1	1	1	1	1	0	0	0	1	0	0
E	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1
F	0	0	0	0	0	1	0	0	0	0	0	1	1	1	1	1	1	1

. Considering the rapid urbanization in the CMA, the EDS permits unrestricted land type conversions while constraining construction land circulation. The EPS, conversely, places partial restrictions on ecological land type conversions. These models also incorporate land policies and regulatory restrictions to identify areas with development constraints [34]. This study designates ecological barriers, including Giant Panda National Park, the Baishui River, the Longxi-Hongkou National Nature Reserve, the Minshan-Qionglai Mountains Biodiversity Reserve, the Longmen Mountains, the Longquan Mountains, the Minjiang River, and the Tuojiang River, as limiting factors (Figure 1).

This study employed the kappa and FoM coefficients to assess the reliability of the PLUS model results. The kappa coefficient quantifies the consistency between simulated and actual land classifications. An elevated kappa coefficient signifies a more precise evaluation [35,36]. The equation is as follows:

$$\mathrm{K}\mathrm{a}\mathrm{p}\mathrm{p}\mathrm{a}={\displaystyle \frac{{\mathrm{P}}_0-{\mathrm{P}}_c}{{\mathrm{P}}_p-{\mathrm{P}}_c}}$$

(1)

In the equation, P₀ expresses the proportion of accurately simulated grids relative to the overall total, P_c denotes the fraction of correctly simulated grids compared to the total in a randomized scenario, and P_p conveys the ratio of accurately simulated grids to the total in an optimal or ideal condition.

The FoM coefficient precisely measures simulation accuracy for land use changes [37]. It is computed as follows:

$$\mathrm{K}\mathrm{a}\mathrm{p}\mathrm{p}\mathrm{a}={\displaystyle \frac{\mathrm{B}}{\mathrm{A}+\mathrm{B}+\mathrm{C}+\mathrm{D}}}$$

(2)

In the equation, A refers to the region mistakenly forecasted to stay the same, B highlights the zone where land use changes were accurately predicted, C marks the area incorrectly categorized as a different land use type, and D points to the section where no real land use shift happened but was wrongly projected as a transition.

2.3.4. Evaluation of Ecosystem Health Utilizing the VORS Model

A functioning ecosystem should be self-sustaining and capable of delivering vital ES to human populations. The ecosystem health was evaluated based on four principal indicators: vigor, organization, resilience, and the provision of ES. The EHI offers an established means of measuring these indicators [2,15,38]. Human-derived ecological benefits were measured through an assessment of ES, while the structural and functional integrity of each ecosystem was evaluated by utilizing vigor, organization, and resilience. A “healthy ecosystem” is understood here to represent stable ecosystem structures, functions, and processes. These structures and processes enable the delivery of vital ES essential for human well-being and societal progress [39]. These factors were assumed to have the same impact on ecosystem health and were therefore accorded the same weight in the EHI [40,41]. The EHI is calculated below:

$$\mathrm{E}\mathrm{H}\mathrm{I}=\sqrt[4]{\mathrm{E}\mathrm{V}\times \mathrm{E}\mathrm{O}\times \mathrm{E}\mathrm{R}\times \mathrm{E}\mathrm{S}}$$

(3)

In the equation, EV denotes ecosystem vigor, EO signifies ecological organization, ER represents ecosystem resilience, and ES indicates ecosystem services.

To derive a composite index, each indicator was normalized to a scale of 0 to 1 [42,43]. The equal interval method was applied to classify the resulting EHI values for 2005 and 2020 across five designated levels: Level 1 (extremely poor, 0–0.2), Level 2 (poor, 0.2–0.4), Level 3 (moderate, 0.4–0.6), Level 4 (good, 0.6–0.8), and Level 5 (relatively good, 0.8–1) [44,45,46].

The values of EV, EO, ER, and ES can be computed in the following way:

1.

Ecosystem Vigor (EV): This concept typically denotes the metabolic efficiency or the fundamental productivity of a specific ecosystem. According to previous research, the Normalized Difference Vegetation Index (NDVI) is a reliable indicator of ecosystem vigor [38]. Therefore, the current paper utilizes NDVI to express regional ecosystem vigor.

2.

Ecosystem Organization (EO): This concept describes an ecosystem’s ability to remain stable when subjected to anthropogenic pressures [15,47]. Ecosystem organization is often defined by measures of landscape heterogeneity (LH), landscape connectivity (LC), and critical ecosystem connectivity (CEC). The SHDI [2] and modified Simpson diversity index [44] represent common metrics for measuring LH. The contagion index and landscape division index evaluate LC [48], while the aggregation index and patch cohesion index assess CEC [2]. Each component of the landscape structure index exerts a specific influence on ecosystem function, and this influence is quantifiable through weighted contributions in the assessment framework. Considering the unique and non-fungible contributions of LH, LC, and CEC to overall ecosystem health, these factors were assigned specific weights. Here, LH and LC were each given a weight of 0.35 [49]. The effect of critical ecosystem connectivity is comparatively smaller, which leads to a weight assignment of 0.3 [24,37]. The weighting scheme employed for each variable is informed by the existing literature [47,50], and the specific equation is detailed below:

EO = 0.35 × LH + 0.35 × LC + 0.35 × CEC = (0.2 × SHDI + 0.15 × MSIDI) +
(0.2 × CONTAG + 0.15 × DI) + (0.15 × AI + 0.15 × COHESION)

(4)

In the equation, LH denotes landscape heterogeneity, LC signifies landscape connectivity, CEC represents critical ecosystem connectivity, SHDI refers to the Shannon diversity index, MSIDI indicates the modified Simpson diversity index, CONTAG stands for contagion index, DI describes the landscape division index, AI conveys the aggregation index, and COHESION elucidates the patch cohesion index.

3.

Ecosystem Resilience (ER): ER denotes the capacity of an ecosystem to recuperate and preserve its original condition following a disturbance. This study measures ER utilizing the ER coefficient and the ecosystem resistance coefficient [2,51]. Based on data from previous research [40,49], weights of 0.6 and 0.4 were assigned, respectively. The equation is as follows:

ER = 0.6 × ∑P_i×R₁ + 0.4 × ∑P_i × R₂

(5)

In the equation, ER represents ecosystem resilience, while P_i indicates the percentage of land allocated to use type i. R₁ and R₂ symbolize the coefficients for ecosystem resistance and resilience, respectively, associated with each land use category.

4.

Ecosystem Services (ES): This term refers to a series of direct and indirect ecological functions provided for humans [52]. It critically measures the ecological health of a specific region. The calculation of ES is based on the equivalent factor method proposed by Xie Gaodi et al. [53]. The evaluated ecosystem services are categorized into provisioning services, regulating services, cultural services, and support services [3,54]. Provisioning services encompass the supply of vital resources, including food and raw materials, fundamental to human subsistence. These services involve food production and the provision of biological materials to meet various human needs. Regulating services encompass ecological processes that maintain environmental balance and promote sustainability. These strategies encompass climate regulation and hydrological regulation, all intended to augment ecological resilience. Cultural services enhance human well-being and general quality of life. They represent the intangible advantages conferred to human society. Supporting services are essential ecological processes that uphold other ecosystem functions, such as soil development and preservation, alongside biodiversity conservation. These processes are crucial for sustaining ecosystem stability. Standard equivalent values corresponding to the actual production conditions of the study region were established based on the classification of ecosystem services (ES). A spatial overlay analysis was then conducted to integrate the area of each land use type. The spatial distribution of ES values within the research region was visualized accordingly, and the formula is as follows:

ES = ∑(A_i + VC_i)

(6)

ECi = ∑D_j × EV

(7)

In the equation, A_i depicts the expanse of land use type i. VC_i indicates the ecosystem service coefficient tied to land use type i. D_j refers to the value of the service function for every land use type. EV denotes the economic value of the equivalent factor, i denotes the land use type, and j conveys the service function of each land use type.

Calculations based on the per-unit value of ES in Chinese ecosystems exhibit that the average biomass factor for cropland in the EHI is 1.28. In 2005, the average grain price was 1.73 yuan/kg, grain output was 5226.95 kg/hm², and the average equivalent value for ES was 1292.93 yuan/hm². By 2020, the average grain price reached 2.74 yuan/kg, and the grain output was 5062.33 kg/hm². The average ES equivalent value was 1985.73 yuan/hm².

3. Results

3.1. Assessment of the Accuracy of the SD-PLUS Coupled Model

The SD model’s accuracy was confirmed through a comparison of its 2020 land demand projections with the real-world data in

Table 5. Precision of the land demand system dynamics simulation model.

Unit (km2)	Cultivated Land	Forest Land	Grassland	Water	Unused Land	Construction Land
2020 actual value	23,978.41	6544.38	181.63	321.71	8.84	2074.54
2020 simulation value	23,748.59	6687.17	179.49	335.13	8.42	2150.71
Relative error (%)	−0.96	+2.18	−1.18	+4.17	−4.75	+3.67

. The findings reveal minimal deviation between simulated and actual data, with a relative error within ±5%, thus indicating the model’s high simulation precision and its efficacy in mirroring land use pattern changes.

The land use distribution for 2020 was projected based on land demand forecasts derived from the SD model (Table 2). The examination of the confusion matrix produced a kappa coefficient of 0.81 and a FoM coefficient of 0.4, thereby confirming the reliability of the simulation results [55].

3.2. Land Use Scenarios for 2035 Predicted Using the SD-PLUS Coupled Model

The validated SD-PLUS coupled model forecasted land use alterations in the CMA up to 2035 across three distinct scenarios.

Table 6. The anticipated demand for each category of land use across various situations.

Scenario	Cultivated Land	Forest Land	Grassland	Water	Unused Land	Construction Land
Actual in 2020	23,978.41	6544.38	181.63	321.71	8.84	2074.54
EPS	22,775.61	6880.01	170.93	353.28	11.24	2918.43
NDS	22,907.43	6448.37	153.39	380.16	9.69	3210.45
EDS	22,168.04	6801.93	127.38	368.61	12.98	3630.57

showcases the predicted land requirements, and Figure 4 depicts the spatial arrangement of land uses that stemmed from the simulation.

Figure 4 and Figure 5 illustrate how, by 2035, land use demand in the CMA will be significantly affected by the different scenarios. The EPS, with its ecological protection policies and urban greening projects, effectively limits ecological land reduction. Cultivated land and grassland experience only modest declines of 5.02% and 5.89%, respectively, and grassland demonstrates the smallest decrease. The stringent policy of converting cultivated land to forest land results in a 335.64 km² expansion of forest area. Restrictions on construction land expansion lead to measured growth of the urban center, and this minimizes increases in built-up areas. In the NDS, reduced restrictions make population expansion and economic progress key factors in land use transformation. Following the existing urban expansion strategy, construction land encroaches upon some ecological land, and both forest land and grassland decline by 1.47% and 15.55%, respectively. This rapid expansion of construction land, a 54.76% increase in area, is fueled by urban population growth. Compared to the EPS, the central urban area expands more noticeably, but it is still regulated by policy. Land use changes are primarily concentrated in peripheral areas. Under the EDS, a clear economic policy orientation leads to a considerable expansion of construction land, and this increase is 1556.04 km², marking the greatest increase among the three scenarios. Fueled by swift economic expansion and rapid urban development, vast areas of cultivated land and natural habitats are repurposed for industrial and commercial use. This conversion results in a 7.55% reduction in cultivated land and a 29.87% reduction in grassland. Forest land demonstrates a 3.94% increase, and this increase is largely due to the extensive urban greening and ecological restoration efforts integrated into urban development projects. This scenario presents the most significant expansion of construction land. Urban expansion is evident, and pressure on the ecosystem intensifies. While forest land increases thanks to large-scale greening and ecological restoration projects, the overall ecological environment experiences significant effects, placing critical strain on the regional ecosystem. An analysis of land use changes in three scenarios indicates that the expansion patterns are typical. The implementation of distinct policy frameworks gives rise to varied scenarios concerning the spatial distribution of land use. These patterns offer valuable guidance for future land use planning and ecosystem health protection.

3.3. Analysis of Ecosystem Spatial Pattern

3.3.1. Ecosystem Vigor

Figure 6 presents the shifts in ecosystem vigor projected for 2035 relative to 2020 across the EPS, NDS, and EDS, with respective mean values of 0.1972, 0.2281, 0.2170, and 0.2125. The most significant boost in vigor occurs under the EPS, with the NDS and EDS trailing behind in that order. When it comes to spatial distribution, vigor levels display a clear west-to-east decline, with the western regions showing stronger values compared to the eastern areas. The regions with the highest vigor levels are situated in the western section of Chengdu, which is characterized by lush vegetation. In comparison, the lowest values occur in construction land across various urban centers. By 2035, EV is projected to increase significantly in the southeastern Yanjiang District of Ziyang City, alongside Lezhi County and Zhongjiang County of Deyang City. This increase suggests improvements in vegetation cover and broader ecosystem health, especially under the EPS. Government policies and initiatives advocate for environmental conservation and the reforestation of cultivated land. This notably increases vegetation cover and enhances the ecological environment, especially in densely populated urban construction zones.

In addition, the western CMA (Dayi County, Chongzhou City, and Dujiangyan City) is projected to experience a certain decline in vigor from 2020 to 2035. This decline is primarily attributed to grassland and forest land being converted into construction land in this area. Rapid urbanization intensifies ecological pressures, and the limited metabolic capacity of urban ecosystems impacts EV. In the cities of the western region, the overall EV value decreases due to urban expansion after ecological restoration measures are taken.

3.3.2. Ecosystem Organization

The distribution of ecosystem organizations is uneven and fragmented, significantly influenced by human activities and urbanization. As presented in Figure 7, the organization value for the research region in 2020 is 0.4254, while the corresponding values for the EPS, NDS, and EDS are 0.4852, 0.4919, and 0.4881, respectively, establishing an improvement gradient of NDS > EDS > EPS. The largest gain in organization value under the NDS likely arises from the regulated urban growth through current policies, which encourages more efficient expansion of construction land. This suggests that strict ecological protection or widespread urban development may be detrimental, and that ecological organization cannot be understood through a single factor. In general, the ecological organization across all three 2035 scenarios remains comparatively low, displaying a scattered spatial distribution pattern. The highest organization values are concentrated in forest land areas situated between cities, while the lowest values appear in the urban centers of Chengdu, Deyang, Meishan, and Ziyang. The cause is that the fragmented land use practices in metropolitan regions diminish ecological stability and connection. The western region of the research area, encompassing the Longmen Mountains, shows a disrupted and weakened ecosystem structure. Excessive tourism development contributes to this degradation. The accompanying urban expansion and infrastructure construction both lead to localized ecosystem fragmentation. Therefore, the ecological organization in this mountainous region deviates from an ideal configuration, and it instead demonstrates characteristics of heterogeneity and fragmentation, thereby reducing ecological functions to varying degrees.

3.3.3. Ecosystem Resilience

Figure 8 depicts the changes in ER across the CMA between 2020 and 2035 under the three scenarios. The average resilience value demonstrates a radial decline from the center. In 2020, the resilience value is 0.4952. Under the EPS, NDS, and EDS scenarios, the values are 0.4910, 0.4841, and 0.4788, respectively. By 2035, ER in Chengdu City’s urban core (Jinjiang District, Qingyang District, Jinniu District, Wuhou District, and Chenghua District) will have experienced the greatest decline, and this is especially evident under the EDS scenario. This decrease is attributable to rapid urbanization, resulting in significant encroachment upon natural ecological areas, which leads to a corresponding loss of ER. The EPS scenario is represented by a relative scarcity of grassland, while an abundance of forest land contributes to greater ER and resistance. This slight increase in resilience allows for the ecosystem to maintain or regain stability amid greater external pressures from economic development. From 2020 to 2035, ER in Zhongjiang County (Deyang City) and Yanjiang District (Ziyang City), situated in the eastern portion of the study area, is projected to increase considerably. This primarily agricultural region experiences less urbanization, with land use dominated by cultivated land; therefore, ecological protection efforts prove effective. Policies that promote conversion from cultivated land to forest land and vegetation restoration significantly improve the self-rejuvenating potential of the local ecosystem. These improvements lead to benefits in soil health and water conservation, finally contributing to a more resilient ecosystem.

3.3.4. Ecosystem Services

Figure 9 depicts ecosystem services, employing the LUCC matrix to measure and chart the provision, need, and equilibrium of these services for 2020 and 2035. The value of ecosystem supply under the EPS, NDS, and EDS in 2020 and 2035 is 0.4376, 0.4501, 0.4387, and 0.4346, respectively. The values of ES are elevated in the northwestern part of the CMA, whereas the middle and eastern sections have lower values, indicating a geographic gradient distribution. By 2035, the most substantial increases in ecosystem service supply are anticipated in Deyang City and Ziyang, whereas the urban center of Chengdu is projected to undergo the most considerable decline due to its elevated population density and swift urbanization. Conversely, under the EPS, the eastern region demonstrates the most significant improvement in ecological service levels, particularly in Deyang and Ziyang. This enhancement is closely linked to decelerated urban expansion and effective ecological restoration initiatives. Efforts to convert cultivated land into forest land and restore grasslands show significant success in this regard. The high ES supply value in the Western Longmen Mountains region remains owing to the prevalence of ecosystems, including forest land and grassland. This situation is further supported by policies such as the designation of ecological protection areas.

3.4. Prediction of Changes in the EHI

As presented in Figure 10, the CMA’s EHI levels for 2035 are displayed across various scenarios. The mean EHI values were 0.3858 in 2020, projected to be 0.4286, 0.4044, and 0.4097 under the EPS, NDS, and EDS, respectively, in 2035. Across all three scenarios, ecosystem health is projected to improve from a poor to a moderate condition, and the spatial distribution in each scenario remains consistent. In 2020, Level 1 EHI areas were primarily concentrated in the central urban zone, covering 10,488.78 km² (36.68% of the total research area). Levels 2 and 3 were widely distributed throughout the non-central urban areas, reflecting anthropogenic pressures on ecosystem health. In contrast, Level 4 and Level 5 EHI areas—primarily national scenic spots and nature reserves—were concentrated in Deyang City (Zhongjiang County), Meishan City (Hongya County), and western Chengdu City (Dayi County, Chongzhou City, and Dujiangyan City), covering 8093.82 km² (24.44%). These areas exhibited higher EHI values, suggesting that while the CMA maintains significant vegetation cover, there remain substantial gaps in systematic ecosystem protection. By 2035, the CMA experiences significant shifts in both EHI values and their spatial distribution across the three scenarios. Level 1 and Level 2 areas exhibit notable reductions, with the most significant decline under the EPS, where Level 1 shrinks by 5124.93 km² and Level 2 reduces by 3659.68 km². Particularly in Deyang City and Ziyang City, low EHI areas contract significantly, while Level 3 and Level 4 zones expand, reaching 19,565.51 km², 19,066.38 km², and 19,703.84 km² under the EPS, NDS, and EDS, respectively. The transition of areas from Level 1 and Level 2 to higher EHI categories suggests that land conservation policies and afforestation efforts are major factors in enhancing ecosystem health.

The ecosystem health trends across the three scenarios indicate that the EPS fosters the greatest improvement in EHI, while the EDS exhibits the weakest enhancement. In the EPS, ecosystem health benefits from forest land expansion and restricted urban sprawl, demonstrating the effectiveness of conservation policies. Conversely, in the EDS, rapid urbanization and reduced ecological restoration efforts contribute to a marginal improvement in EHI, as extensive land conversion fragments ecosystems and weakens ecological resilience. The NDS represents a balanced approach, where moderate urbanization and ecological conservation measures lead to stabilized ecosystem health improvements.

To further illustrate the spatial shifts in EHI, differences between the 2035 ecosystem health distribution maps and the 2020 baseline were analyzed in ArcGIS 10.8. Changes were categorized into Significantly Improved, Slightly Improved, Unchanged, Slightly Deteriorated, and Significantly Deteriorated (Figure 11). The EDS enhances ecological health relative to 2020. Declines in EHI are observed mainly within the urban centers of Wenjiang District, Shuangliu County, Qingbaijiang District, Anyue County, Guanghan City, and Hongya County in Meishan City and Ziyang City. The EHI index remains relatively steady in these highly urbanized areas; nevertheless, the adjacent regions experience considerable degradation due to the transformation of extensive cultivated land and grassland into developed land. Under the NDS, the marginal improvement in EHI suggests that ecosystem health remains relatively stable under current urban growth policies, though localized declines persist. However, the most pronounced increase in EHI is observed under the EPS, particularly in Jingyang District (Deyang City), Dongpo District (Meishan City), and Yanjiang District (Ziyang City), where forest expansion and ecological conservation measures drive significant gains. This finding suggests that high-quality development strategies integrating ecological protections can promote regional sustainability, balancing economic growth and environmental conservation. Overall, the results demonstrate that, between 2020 and 2035, ecosystem health in the CMA improves across all scenarios, though the magnitude of improvement varies depending on policy interventions. The disparity between the northwestern region, characterized by a greater EHI, and the middle and eastern region, which exhibits a lower EHI, is progressively diminishing, underscoring the critical significance of land use planning and ecosystem health preservation in fostering regional ecological stability and sustainable development.

4. Discussion

4.1. Spatiotemporal Analysis of Ecosystem Health Under Different Land Use Scenarios

The findings underscore the intricate link between land use alterations and ecosystem health, stressing the necessity of policy-driven interventions in shaping ecological circumstances. Within the EPS framework, enhancements in EHI are most evident, suggesting that afforestation policies and ecological restoration initiatives can effectively alleviate the detrimental impacts of urbanization. The government’s Grain for Green Program and Natural Forest Protection Program have been instrumental in rehabilitating degraded landscapes and enhancing forest coverage, hence stabilizing biological processes. Moreover, ecological redlines, which prohibit construction in ecologically vulnerable regions, guarantee that urban development does not intrude onto vital habitats. These focused conservation efforts not only augment biodiversity but also mitigate soil erosion and bolster ecosystem stability. These results align with contemporary research on conservation-oriented land use planning that enhances regional ecological resilience. The outcomes obtained from the NDS indicate that the implementation of improved ecological protection measures can significantly bolster ecosystem health and that proactive environmental policies are very beneficial in securing the long-term survival of ecosystems.

The NDS indicates a moderate rise in EHI, implying that a balanced strategy for urbanization and ecological preservation can yield stable ecosystem health results. In this context, urban growth persists; however, current rules and regulations mitigate severe ecological damage while curbing unchecked urban sprawl, thereby fostering sustainable urban development to some degree. The research findings indicate that with suitable regulatory oversight, urban expansion does not inevitably result in significant degradation of the ecological environment. Nonetheless, whereas the NDS mitigates the adverse effects relative to the EDS, it fails to attain the significant ecological advantages evident under the EPS. This indicates that policy enforcement and afforestation initiatives are crucial for ecological recovery. Furthermore, whereas regional land use regulations strive to balance economic development and environmental preservation, the lack of stringent ecological restoration methods results in slower enhancements in ecosystem health compared to those seen under the EPS.

On the flip side, the EDS illustrates a grim scenario of environmental stress, where rapid economic expansion and extensive urbanization disrupt ecosystems and diminish essential ecological processes. The expansion of construction land has resulted in a substantial decrease in cultivated land and grassland, consequently impairing ecosystem health. In this context, inadequate implementation of ecological protection measures, coupled with land-intensive economic expansion methods, exacerbates landscape degradation. The findings indicate that the lack of land use rules in urban expansion zones intensifies habitat fragmentation, hence diminishing landscape connectedness and heightening ecological fragility. These findings align with previous research that identifies land fragmentation as a primary factor in long-term ecological decline, further underscoring that policy frameworks significantly influence the mitigation of environmental damage.

A comparative examination of the three forecasted scenarios for 2035 reveals a correlation between land use change and ecosystem health. Although land use regulation provides essential ecological protection, it is inadequate by itself. Land use change in each scenario is influenced not only by economic, demographic, urbanization, and similar forces but also by the ecological protection measures themselves. Consequently, attaining substantial enhancements in ecosystem health requires comprehensive actions, encompassing stringent ecological restoration and soil conservation initiatives. It has been demonstrated that the negative impacts on ecosystem health can be mitigated by strategically coordinating land use planning with the establishment of ecological protection areas. This comprehensive strategy has the potential to promote economic growth while simultaneously enhancing the health of regional ecosystems.

4.2. Policy Recommendations

To mitigate the tension between land utilization and ecosystem health preservation, the government has released several policy documents, including the National Territorial Space Planning Outline, the Overall Plan for Ecological Civilization System Reform, and the Land Management Law. These policies emphasize the rational utilization of land, the appropriate management of the relationship between economic development and ecosystem preservation, the proactive promotion of the transition from extensive to intensive land use, the enhancement of land utilization, and the maintenance of a dynamic equilibrium between ecosystem health and the judicious use of land.

This study’s findings indicate that the EPS exerts the most substantial influence on enhancing ecosystem health. To mimic the beneficial effects of the EPS, the CMA must enforce more stringent land use regulations, particularly in ecologically significant regions like the Longmen and Longquan Mountains, where ecosystem services are vital for biodiversity preservation and climate management. Moreover, enhancing afforestation initiatives and prioritizing the transformation of degraded land in buffer zones into forest land can augment ecosystem resilience. The NDS indicates a moderate enhancement in ecosystem health, primarily due to controlled urban development and regulated land use rules. Nonetheless, the situation illustrates that regions with low EHI values, especially the swiftly urbanizing peri-urban zones surrounding Chengdu, Deyang, and Ziyang, continue to be susceptible to land degradation and habitat loss. To alleviate these concerns, the CMA should implement diversified ecological management measures. In regions with low EHI ratings, stringent land conversion regulations must be enforced to avert additional environmental deterioration. In regions with moderate EHI values, it is advisable to advocate for investments in green infrastructure, such as the creation of urban green belts, ecological buffers, and nature-based stormwater management systems. Areas with a high EHI should be classified as severe protection zones, and development activities must adhere to rigorous environmental impact assessments to avert irreparable ecological harm. The EDS simulation findings indicate a significant transformation of arable land into building land, with urban growth resulting in ecological fragmentation. To mitigate anticipated adverse effects, the CMA must emphasize integrated urban development and enhance land use efficiency. Urban growth should prioritize vertical and high-density development over outward expansion, hence minimizing the need for new construction land. Policies ought to promote urban revitalization and land reutilization.

This study presents three recommendations for governmental decision-making to attain a more equitable balance between land utilization and ecosystem vitality. Firstly, a strict ecological protection policy should be formulated, precise territorial space planning should be implemented, ecological red lines should be strictly adhered to, and comprehensive land remediation and restoration measures should be taken. Secondly, this study calls for the strengthening of comprehensive ecological management in areas with low EHI values in the CMA. This should include the strict regulation of land use changes, with priority given to increasing regional vegetation coverage. In areas exhibiting moderate EHI values, the formulation of judicious land use plans is strongly advocated. In areas with high EHI values, while infrastructure development is permitted, strict protection measures must be implemented to prevent ecological damage. Thirdly, the CMA must effectively regulate the total amount of construction land and maximize land use efficiency. The expansion of urban construction land has a direct impact on the distribution of areas with low EHI values, which in turn affects the overall ecosystem health. It is therefore recommended that economic growth be promoted by encouraging regional redevelopment and investment in high-tech industries, with a view to reducing the reliance on new construction land.

4.3. Limitations and Future Prospects

While land use demand projections are crucial for scenario planning, the PLUS model relies on Markov chains, which may introduce bias. The land use simulations for 2035 are based on parameters calibrated using empirical data from 2005 and 2020. Future research should consider temporal variations in model parameters to improve accuracy. Ecosystems are complicated and affected by both natural and anthropogenic elements. While the VORS framework provides a comprehensive assessment of ecosystem health, it does not establish absolute ecological standards. Although factor weights in this study were based on CMA-specific conditions and previous studies, they involve subjective judgments, which may not fully capture all ecosystem dynamics. Future research should explore more objective weighting approaches to enhance the reliability of indicator assessments. Additionally, the SD-PLUS-VORS model predicts EHI at a specific time, but it does not fully capture the dynamic processes driving ecosystem health. Future studies should incorporate landscape pattern analysis and process-based ecological models to develop a more holistic framework for ecosystem health assessment. Despite these limitations, this paper offers an insightful exploration into ecosystem health dynamics, supports land use policy formulation, and contributes to sustainable regional development in the CMA.

5. Conclusions and Recommendations

The present study takes the CMA as its research object and employs the SD-PLUS-VORS coupled model to simulate the spatiotemporal changes in land use and ecosystem health in 2035. The primary research outcomes are as follows:

(1)

Simulated land use patterns for the CMA in 2035 exhibit significant differences across the three scenarios. In the context of the EDS, the total construction land area is projected to increase to 3630.57 km², representing a 75.01% growth. The majority of the new construction sites will be situated on the periphery of the city center. Conversely, under the EPS, the expansion of construction land is constrained by numerous legislative limitations, resulting in a negligible increase, whereas the area of forest land has markedly expanded due to afforestation and conservation initiatives. Under the NDS, maintaining the current urban development policy, land use achieves a balance between ecological protection and economic development.

(2)

The EHI distribution in the CMA shows spatial heterogeneity, with lower values in urban core areas and overall relative stability. Projections for 2035 indicate the following ranking of the CMA EHI across the three scenarios: EPS (0.4286) > EDS (0.4097) > NDS (0.4044). Level 4 and Level 5 health zones are predominantly located in the forested areas of the western Longmen Mountains and the central and western Longquan Mountains, whereas Level 3 health zones are mainly found in the peripheral municipalities. Level 1 and Level 2 health zones are predominantly situated in the construction land surrounding the primary urban areas of the four cities within the research area. These areas are highly susceptible to the effects of land use transformations. Under the EDS, the widespread expansion and uncontrolled growth of construction land leads to a slight increase in EHI, with the smallest increase. In the NDS, the increase in the EHI is balanced. Conversely, in the EPS, the implementation of stringent ecological protection measures could lead to a substantial enhancement in the regional EHI, underscoring the pivotal role of ecological protection policies in attaining regional sustainability.

(3)

The spatial distribution of land use is closely related to ecosystem health. The EHI in the three scenarios exceeds the 2020 level, with the EPS demonstrating the most significant increase and the NDS exhibiting the least increase. These findings underscore the pivotal role of ecological protection policies in mitigating the adverse impacts of urban expansion and ensuring long-term sustainability. This study proposes a scalable modeling approach that can be used to enhance scenario-based land use simulation and ecosystem health assessment, thereby informing long-term sustainable land management strategies. The model’s scalability and the insights it provides are noteworthy, as they can be applied to other rapidly urbanizing regions and offer decision-makers valuable tools to balance economic development and ecological sustainability.