Integrating Linear Programming and CLUE-S Modeling for Scenario-Based Land Use Optimization Under Eco-Economic Trade-Offs in Rapidly Urbanizing Regions

Abstract

Rapid urbanization has intensified eco-economic trade-offs, necessitating integrated optimization frameworks that balance development with environmental conservation in land use planning. Traditional methods often fail to optimize both objectives simultaneously, highlighting the need for systematic approaches addressing competing demands. This study develops an integrated linear programming (LP) and CLUE-S modeling framework using Guangzhou, a rapidly urbanizing megacity in China, as a case study. The methodology combines LP quantitative optimization with CLUE-S spatial allocation under dual objectives: maximizing ecosystem service value and economic benefits across four policy scenarios: ecological protection, cultivated protection, economic development, and balanced development. Data inputs include the 2020 land-use database, 12 socio-economic and biophysical driving factors, and territorial planning constraints. Results show that the coupled framework effectively balances urban expansion with ecological protection, reducing habitat fragmentation and preserving key ecological corridors compared with business-as-usual scenarios. Accuracy assessments further confirm the robustness and reliability of the framework. The integrated LP-CLUE-S framework captures land use dynamics and spatial constraints, providing a robust tool for territorial spatial planning. This approach offers actionable insights for reconciling development pressures with environmental conservation, contributing a replicable methodology for sustainable land resource management with strong transferability potential for other rapidly urbanizing regions facing similar eco-economic challenges.

1. Introduction

Land serves as the foundational resource for human survival and development, supporting critical ecosystem services such as soil and water conservation, habitat protection, and climate regulation [1,2]. However, the accelerating pace of contemporary urbanization worldwide has precipitated dramatic land-use transformations that threaten ecological stability while fueling economic growth. This challenge is particularly pronounced in rapidly developing regions across different continents, where urban expansion has consumed significant portions of arable and forest lands, degrading ecosystem service values (ESV) and intensifying conflicts between development needs and environmental protection [3,4,5,6,7]. These regions exemplify a global challenge: how to optimize land-use allocation to balance economic prosperity with ecological sustainability—a dilemma exacerbated by conventional planning approaches that prioritize short-term gains over long-term resilience [3,8,9].

Land use optimization has evolved as a critical interdisciplinary field at the intersection of economics, human geography, and landscape ecology, with significant methodological developments occurring across different geographic contexts [10,11,12]. Early foundational work by Forman and Godron [13] established key concepts of landscape pattern optimization through their influential “centralization-decentralization” model, while Dokmeci [14] pioneered the application of linear programming (LP) to land use allocation, introducing quantitative rigor to planning processes. These theoretical advances have been applied and refined globally, with significant contributions from various regions demonstrating the universal nature of land use optimization challenges [15]. Verburg et al. [16] developed sophisticated land use modeling approaches for European contexts, emphasizing the integration of policy scenarios with spatial planning frameworks. Lawler et al. [17] demonstrated how climate change considerations could be integrated into land use optimization, providing valuable methodological insights for adaptive planning. However, these studies primarily addressed contexts with established regulatory frameworks and relatively slow urbanization rates, limiting their direct applicability to rapidly developing regions.

The development of land use simulation models, particularly the CLUE (Conversion of Land Use and its Effects) model by Veldkamp and Fresco [18], and its refined version, CLUE-S, by Verburg et al. [16], marked a major methodological breakthrough. These models combine empirical land use-driver relationships, dynamic competition modeling, and multi-scenario spatial projections, making them valuable for visualizing land use transitions and assessing planning alternatives [19,20]. Global applications have demonstrated their versatility: Zhu et al. [19] applied CLUE-S to Chinese agricultural regions, Nasiakou et al. [20] used it for Mediterranean landscape planning, while Kiziridis et al. [21] adapted it for biodiversity conservation in Greece. However, these applications also highlight a key limitation—the reliance on external demand inputs, which restricts their ability to support fully integrated planning [22,23,24,25].

More recently, ecosystem service valuation has emerged as a critical component of land use decision-making globally, emphasizing the interdependence of ecological and economic systems [26]. Advances in valuation methodologies [27] and policy-driven sustainability metrics [28] have reinforced its importance. Notable applications such as Wang et al.’s [29] integration of ESV into LP models and Liang et al.’s [8] ecological-oriented investment optimization frameworks represent important steps forward. However, these approaches often focus on singular ecological benefits while neglecting regional socio-economic dynamics, limiting their applicability across different development stages and geographic settings [30]. This highlights the need for more synthetic approaches that can simultaneously evaluate economic and ecological outcomes across land use types, enabling truly balanced optimization of competing land use demands [28,31,32]. While scenario-based approaches are widely used, few studies systematically integrate multiple policy priorities within a single analytical framework [33]. Additionally, existing ESV applications often rely on generalized coefficients that may not reflect local ecological and economic conditions, particularly in rapidly changing urban environments [28].

Building on these gaps, this study aims to address several key scientific questions: How can land use optimization frameworks simultaneously account for competing policy priorities such as ecological protection, food security, and economic development in rapidly urbanizing regions? How can the trade-offs between urban expansion and ecosystem service provision be quantitatively assessed and spatially represented? And to what extent do region-specific ecological and socio-economic factors influence optimal land use allocation strategies? By explicitly posing these questions, this research seeks to advance the theoretical understanding of integrated land use planning under complex urbanization pressures and to provide a foundation for evidence-based policy interventions.

Most existing studies apply either optimization models or spatial simulation models separately, failing to capture the full complexity of land use systems [12]. The integration of linear programming with spatial simulation models like CLUE-S represents a significant methodological advancement that addresses these limitations. While LP models excel at quantitative optimization by maximizing objective functions under specified constraints, they lack spatial explicitness and cannot capture geographic processes driving land use change. Conversely, CLUE-S models provide sophisticated spatial allocation capabilities but depend on externally provided demand quantities, limiting their optimization potential. This integration creates a synergistic framework where LP determines optimal land use quantities based on policy objectives, while CLUE-S translates these demands into spatially explicit patterns considering local driving factors and constraints. Recent studies have demonstrated the effectiveness of this integrated approach compared to alternative methods, showing superior performance over standalone optimization models in capturing both quantitative trade-offs and spatial heterogeneity [34]. Unlike machine learning approaches that function as “black boxes” with limited policy interpretability [35], or agent-based models that lack mathematical rigor for systematic evaluation [36], the LP-CLUE-S integration provides an optimal balance between analytical transparency, spatial realism, and policy relevance for rapidly urbanizing regions.

Using Guangzhou as a representative case study of rapid urbanization challenges common to many developing megacities, this research develops and validates a comprehensive framework that bridges quantitative optimization with spatial simulation. Since 2000, Guangzhou, a megacity of Guangdong–Hong Kong–Macao Greater Bay Area (GBA), has experienced extensive conversion of agricultural and forest land to construction land, reflecting pressures faced by many global megacities. The findings in this study offer actionable policy recommendations for balancing urban expansion with environmental protection, representing a significant step toward operationalizing sustainable development principles in urban planning. The proposed methodology is adaptable to other rapidly urbanizing cities facing similar sustainability challenges, aligns with the UN Sustainable Development Goals, and bridges global methodological advances with local contextual requirements—thereby contributing to both theoretical understanding and practical application of land use optimization in contemporary urban development contexts.

2. Materials and Methods

2.1. Study Areas

Guangzhou (112°57′–114°3′ E, 22°26′–23°26′ N), located in southern China at the heart of the GBA, represents a typical rapidly urbanizing megacity. As China’s fourth-largest city by GDP with a population of 18.68 million (2020) and density of 2512 persons/km², Guangzhou exemplifies the demographic and economic pressures common to rapidly expanding urban centers worldwide. The municipality spans 7434.4 km² across 11 administrative districts, including Liwan, Haizhu, Tianhe, Baiyun, Huangpu, Panyu, Huadu, Nansha, Conghua, Zengcheng, and Yuexiu (Figure 1). The city features diverse topography that transitions from mountainous terrain in the northeast (peaking at Tiantangding, 1210 m) to coastal plains in the south. The city lies at the confluence of the Xijiang, Beijiang, and Dongjiang Rivers, forming a dense and well-developed water system.

Guangzhou’s natural geographical advantages have supported its transformation from a traditional agricultural society into a major industrial and urban center, particularly following China’s reform and opening-up policies. The city initially developed labor-intensive industries through regional integration with Hong Kong, forming a “front shop-back factory” model. However, reliance on similar industrial structures and excessive land and environmental exploitation led to inefficient competition and rising pollution levels. Over the past two decades, Guangzhou has actively embraced industrial restructuring, prioritizing high-tech sectors and absorbing industrial transfers from Hong Kong. Between 2000 and 2020, the city’s gross domestic product (GDP) grew from 237.59 billion yuan to 2501.91 billion yuan, reflecting an average annual growth rate of 16.58%. This economic expansion has been accompanied by construction land increasing by over 60%, primarily at the expense of agricultural and forest areas. Despite this rapid development, Guangzhou continues to face challenges such as environmental degradation, unsustainable land and resource use, imbalanced land-use structures, and growing tensions between human activities and land capacity, issues that mirror sustainability dilemmas confronting many rapidly urbanizing megacities globally.

2.2. Data

2.2.1. Land Use Classification

The land use classification for this study was derived from 2020 Landsat TM imagery (30 m resolution) of Guangzhou, processed according to China’s national land use classification standard (GB/T 21010-2017) [37]. Publicly available datasets such as GlobeLand30 and GLC-FCS30 were not directly adopted, as their classification systems differ from the national standard and may introduce local-scale inconsistencies. Instead, we applied a tailored approach combining supervised classification and visual interpretation, which allowed better adaptation to Guangzhou’s heterogeneous landscape.

The remote sensing data processing involved three main steps: (1) atmospheric and geometric correction using ENVI software, version 5.6; (2) hybrid classification combining supervised classification with visual interpretation to ensure accuracy; and (3) categorization into six primary land use types: agricultural land, forest land, grassland, water body, construction land, and unutilized land. To verify accuracy, we conducted an assessment using 500 randomly sampled validation points from high-resolution Google Earth imagery and field survey records, achieving an overall accuracy of 87.6% and a Kappa coefficient of 0.84.

To facilitate spatial analysis in the CLUE-S model, the processed 30 m resolution data were resampled into a 200 m × 200 m grid using the WGS1984 coordinate system through a majority-rule approach. When multiple land use types occurred within a grid cell, the type with the largest area proportion was assigned as the dominant category. This aggregation ensured consistency in land use representation while minimizing spatial information loss.

2.2.2. Driving Factors

Based on data availability, feasibility, and stability considerations, we selected 12 key spatial driving factors that comprehensively capture the socio-economic and biophysical determinants of land use change. These included external socio-economic factors such as GDP, population density, and active cropland; and internal environmental factors including distance from roads, distance from inland rivers, distance from lakes, elevation, slope, undulation, soil organic matter, normalized difference vegetation index (NDVI), and soil texture. A brief description of these factors is presented in

Table 1. Summary of the 12 spatial driving factors used in this study.

Category	Variable	Description	Unit	Data Source
Socio-economic	GDP	Gross domestic product per capita	CNY	Statistical Yearbook of Guangzhou
	Population density	Population per unit area	persons/km2	Statistical Yearbook of Guangzhou
	Active cropland	Percentage of cultivated land area	%	Statistical Yearbook of Guangzhou
Terrain	Elevation	Altitude above sea level	m	DEM, Geospatial Data Cloud
	Slope	Degree of inclination	%	DEM, Geospatial Data Cloud
	Undulation	Local terrain variation	m	DEM, Geospatial Data Cloud
Space length	Distance from roads	Euclidean distance to nearest road	km	OpenStreetMap/BigMap
	Distance from inland rivers	Euclidean distance to nearest river	km	OpenStreetMap/BigMap
	Distance from lakes	Euclidean distance to nearest lake	km	OpenStreetMap/BigMap
Soil	Soil organic matter	Organic matter content in soil	%	Harmonized World Soil Database
	Soil texture	Classification of soils	Categorical	Harmonized World Soil Database
Vegetation	NDVI	Normalized Difference Vegetation Index	Dimensionless	MODIS Vegetation Index Products

, while their spatial distribution patterns (multi-year averages) are shown in the Supplementary Materials (Figure S1).

The 30 m resolution digital evaluation model (DEM) data, obtained from the Geospatial Data Cloud platform, served as the foundation for deriving terrain parameters (e.g., slope, elevation, and undulation). Infrastructure and hydrological data (e.g., roads and rivers) were extracted from OpenStreetMap and BigMap databases, with subsequent calculation of river network density and proximity metrics using spatial analysis tools in ArcGIS 10.8. Soil properties including organic matter content and texture composition were obtained from the Harmonized World Soil Database (HWSD). Vegetation indices were derived from Moderate Resolution Imaging Spectroradiometer (MODIS) Vegetation Index Products to ensure high temporal resolution and global consistency. Socio-economic indicators such as population density, per capita GDP, and active cropland were compiled from the Statistical Yearbook of Guangzhou. All datasets underwent standardized processing, including spatial interpolation (spline method for socio-economic data) and conversion to 200 × 200 m raster format (ASCII) to ensure model compatibility.

2.3. Methods

To address the dual challenges of quantitative land demand and spatial allocation, this study integrates an LP model with the CLUE-S model to simulate land use optimization for Guangzhou in 2035 under the four policy scenarios (EPS, CPS, EDS, and BDS). The LP model was used to depict the direction of land use shifts and calculate future land use demands based on ecological and economic objectives, while the CLUE-S model was utilized to allocate the land use demands to spatial distribution under each scenario. A methodological framework diagram is shown in Figure S2.

2.3.1. Linear Programming (LP) Model

The LP model was applied to forecast land use demand in Guangzhou for 2035, using 2020 as the baseline year. Two objective functions were defined: maximizing ESV and maximizing economic benefit. Land use constraints and targets were defined for each scenario based on the Master Plan of Land Use in Guangzhou (2006–2020), the Master Plan of Guangzhou Territorial Space (2021–2035), and the forest projection and utilization plan of Guangdong Province (2021–2035).

• Ecosystem service value (ESV)

The quantitative assessment of ESV was first proposed by Costanza et al. [38]. Building upon this framework, Xie et al. [39] refined the method to better align with the characteristics of China’s ecosystems and developed a value equivalent factor table specifically tailored for ecosystem service valuation in China. In this study, we adopted this modified approach and adjusted it to account for the regional characteristics of Guangzhou. Based on statistical data such as grain crop yield and sown area in Guangzhou in 2020, and using Equation (1), the economic value of ecosystem services per unit area ($E_a$, CNY/ha) was estimated to be 1146.27 CNY/ha [40].

$$E_a={\displaystyle \frac{1}{7}}{\sum }_{i=1}^n{\displaystyle \frac{p_i{q}_i}{M}}$$

(1)

where $p_i$ is the national average price of the grain crop i (CNY/t), $q_i$ is the yield of the ith grain crop (t), and M is the total area of grain crops (ha).

Referring to China’s equivalent value coefficient table for ecosystem services and applying Equation (2), we further calculated the per-unit-area ecosystem service value (Table S1). Ultimately, the coefficient of ecological benefits for different land use types in Guangzhou was determined.

$$ESV=\sum (A_i\times {VC}_i)$$

(2)

where $A_i$ represents the area (ha) of land use type i in the study region, and ${VC}_i$ denotes the ecosystem service value per unit area of land use type i.

2.

Coefficient of economic benefit

In this study, we obtained the economic benefit coefficient of each region and calculated the economic benefit coefficient of each land use type by obtaining the output value of the first, second, and third industries recorded in the Guangzhou Statistical Yearbook 2020 and dividing that coefficient by each land use area (

Table 2. Coefficients of ecological and economic benefits of land use in Guangzhou.

Coefficient	Agricultural Land	Forest Land	Grassland	Water Bodies	Construction Land	Unutilized Land
Ecological benefit coefficient (CNY/ha)	4458.98	26,306.86	5811.58	143,982.75	0.00	229.25
Economic benefit coefficient (CNY/ha)	81,947.90	3749.31	314,761.16	160,534.65	10,229,151.41	0.00

).

Then the objective functions for ecological and economic benefits were defined as:

$$f_{ecological}\left(x\right)=max{\sum }_{j=1}^n{c}_j{x}_j$$

(3)

$$f_{economic}\left(x\right)=max{\sum }_{j=1}^n{d}_j{x}_j$$

(4)

where $x_j$ represents the area of land use type j; $c_j$ and $d_j$ are the coefficients of ecological and economic benefit, respectively.

3.

Scenario setting and constraints

Four policy-oriented scenarios were established to reflect distinct development priorities in land use planning. The ecological protection scenario (EPS) aimed to maximize ESV by prioritizing forest and water conservation, with constraints ensuring minimum area thresholds for these land use types. The cultivated protection scenario (CPS) emphasized safeguarding agricultural land to ensure food security, with strong restrictions on the conversion of farmland. The economic development scenario (EDS) focused on promoting urban expansion to stimulate economic growth, allowing higher allocations of construction land within policy-defined boundaries. The balanced development scenario (BDS) sought to achieve a compromise among ecological, agricultural, and economic objectives, applying moderate constraints across all land use categories.

The constraints and target values for each scenario were established by integrating three sources of information:

(1)

Historical land use patterns in Guangzhou, which provided realistic baselines for forest, farmland, and water distribution;

(2)

Territorial spatial planning documents and national policies, such as the “red line” for cultivated land protection and ecological security requirements, which set mandatory lower bounds;

(3)

Development trend analysis, ensuring that GDP and ESV growth targets align with average or projected growth rates rather than unattainable ideals.

Based on these sources, the BDS was used as a reference scenario, with ecological and economic targets reflecting average growth trends, while the other scenarios were adjusted to highlight specific policy orientations. The detailed eco-economic targets and land use thresholds for all scenarios are summarized in

Table 3. Four simulation scenarios with constraint conditions under two optimization objectives.

Optimization Objective	Scenarios	Eco-Economic Benefits Constraint	Land Use Area Constraint
Economic objective	EPS	ESV increased by 3%	Forest land not less than 43%, water bodies not less than 7%, and grassland and unutilized land not less than current levels
	CPS	ESV not less than current levels, GDP increased by 15%	Water bodies, forest land, and grassland not less than current levels
	EDS	ESV not less than current levels, GDP increased by 20%	Water bodies, forest land, and grassland not less than current levels
	BDS	ESV not less than current levels	Forest land not less than 43%, water bodies not less than 7%, and grassland and unutilized land not less than current levels
Ecological objective	EPS	GDP increased by 5%	Forest land not less than 43%, water bodies not less than 7%, and grassland and unutilized land not less than current levels
	CPS	GDP increased by 8%	Water bodies, forest land, and grassland not less than current levels
	EDS	GDP increased by 10%	Water bodies, forest land, and grassland not less than current levels
	BDS	GDP increased by 8%	Forest land not less than 43%, water bodies not less than 7%, and grassland and unutilized land not less than current levels

, which directly links constraints to optimization objectives.

Guangzhou’s rapid urbanization over the past two decades has caused extensive conversion of agricultural land to urban construction land, intensifying food security challenges as the remaining agricultural land struggles to support the city’s population. To address these pressures, critical land use constraints were set in line with planning and policy requirements. Specifically, at least 1287.99 km² of agricultural land was preserved to secure the minimum cultivated land “red line”; over 418 km² of water bodies were maintained for ecological protection; urban construction land was capped at 1772 km² to control sprawl; and a forest baseline of 2914.81 km² was retained, consistent with 2020 conditions. The total optimized area across all scenarios was constrained to Guangzhou’s administrative size of 7072.36 km², ensuring internal consistency in land use allocation.

4.

Objective function solving

The optimization model (Equations (3) and (4)) was implemented using the linprog function in MATLAB 7.01 to solve the multi-objective linear programming problem. Taking the EPS scenario with economic objective as an example, we introduced a constraint requiring the total ESV to exceed the baseline levels by at least 3%, thereby guaranteeing a harmonious balance between ecological conservation and economic development. The LP model was formulated as follows:

• Objective function (maximize economic benefits)

$$f_{economic}\left(x\right)=max(\mathrm{81,947.90}{x}_1+3749.31x_2+\mathrm{314,761.16}{x}_3+\mathrm{160,534.65}{x}_4+\mathrm{10,229,151.41}{x}_5+0x_6)$$

• Subject to ecological constraint (ensure ESV exceeds threshold)

$$f_{ecological}\left(x\right)=\mathrm{m}\mathrm{a}\mathrm{x}(4458.98x_1+\mathrm{26,306.86}{x}_2+5811.58x_3+\mathrm{143,982.75}{x}_4+0x_5+229.25x_6)>\mathrm{161,995,111.20}$$

• Land use constraints

$$x_1\ge 1287.99$$

$$x_2\ge 2914.81$$

$$x_4\ge 418.00$$

$$x_5\le 1772.00 (\mathrm{m}\mathrm{a}\mathrm{x} \mathrm{a}\mathrm{l}\mathrm{l}\mathrm{o}\mathrm{w}\mathrm{a}\mathrm{b}\mathrm{l}\mathrm{e} \mathrm{b}\mathrm{a}\mathrm{s}\mathrm{e}\mathrm{d} \mathrm{o}\mathrm{n} 2002)$$

$${\sum }_{j=1}^6{x}_j=7072.36$$

where $x_1$ to $x_6$ represent the land use areas (km²) allocated to agricultural land, forest land, grassland, water area, construction land, and unutilized land, respectively.

2.3.2. CLUE-S Model

The CLUE-S model was used to simulate the spatial allocation of land use types by combining ecological and socio-economic driving factors [41,42]. The spatial distributions of the land-use types are quantified by using a binomial logit model with the percentages of the types as the dependent variables and 12 socio-economic and biophysical driving factors as the independent variables. The probabilities of the conversion of land use type were defined by using the following logistic model:

$$log\left\{{\displaystyle \frac{p_i}{1-p_i}}\right\}={\beta }_{0,i}+{\beta }_{1,i}{X}_{1,i}+{\beta }_{2,i}{X}_{2,i}+\cdots +{\beta }_{n,i}{X}_{n,i}$$

(5)

where $p_i$ is the probability of the land use type i occurring in a grid cell; X_i are the driving factors; and ${\beta }_{n,i}$ represents the logistic regression coefficient of this driving factor. The quantitative relationships between driving factors and land use patterns were analyzed using IBM SPSS Statistics for Windows, version 22.0 (Table S2 and Figure S3). The reliability of these correlations was assessed through Receiver Operating Characteristic (ROC) curves, with all values exceeding the 0.75 threshold. This confirms that the dataset meets regression requirements, and the selected driving factors are statistically valid for regional modeling applications.

The specific conversion settings of the six land use types affect the temporal dynamics of the simulation, which are composed of two parameters: the transition matrix and the conversion elasticity (ELAS). The value for the first parameter, transition matrices, adopts binary values: “1” indicates permissible transitions between corresponding land use types, while “0” designates prohibited conversions. Based on a systematic analysis of both current land use patterns and projected development trajectories in the study region, transition matrices were established for each scenario as shown in Tables S3–S6. The second parameter, which ranges from 0 (easy conversion) to 1 (irreversible change), is determined on the basis of expert knowledge and observed behavior in recent years (Table S7).

The CLUE-S model operates in discrete time steps and uses conversion rules to simulate demand for all patterns and the most likely changes in the different types on the basis of Equation (5). For each grid i, the total probability of land use type u (${TPROP}_{i, u}$) was calculated according to the following equation:

$${TPROP}_{i,u}=P_{i,u}+{ELAS}_u+{ITER}_u$$

(6)

where $P_{i, u}$ is the spatial probability derived from logistic regression; ${ELAS}_u$ is the conversion elasticity coefficient; and ${ITER}_u$ is the iterative adjustment term.

2.3.3. Landscape-Scale Graph Metrics Selection

Landscape indices are well-established quantitative tools for characterizing landscape patterns, as they effectively capture both structural composition and spatial configuration features while enabling temporal monitoring of landscape dynamics [43,44]. Building upon these fundamental landscape ecology principles, our study systematically selected seven representative indices at both patch and landscape levels to quantitatively assess the spatial configuration characteristics of Guangzhou’s urban landscape under various development scenarios. At the patch-type level, the selected indices include the proportion of landscape (PLAND), largest patch index (LPI), and aggregation index (AI). At the landscape level, the selected indices include landscape shape index (LSI), Shannon’s diversity index (SHDI), Shannon’s evenness index (SHEI), and the connectance index (CONNECT). These indices comprehensively capture aspects such as landscape composition, dominance, spatial aggregation, shape complexity, and connectivity, providing a robust basis for evaluating urban landscape changes. The detailed definitions and calculation formulas for each index are provided in the Supplementary Material (Table S8).

2.3.4. Model Validation

The overall simulation accuracy was evaluated using the Kappa statistic. The Kappa index, which quantitatively measures the consistency between simulated and observed land use, is defined as follows:

$$Kappa={\displaystyle \frac{P_0-P_c}{P_p-P_c}}$$

(7)

where $P_p=1$ represents the perfect agreement under ideal conditions, $P_c$ is the expected agreement by chance, and $P_0$ is the observed proportion of correct simulations. A Kappa value greater than 0.75 is generally considered to indicate satisfactory agreement.

3. Results

3.1. Model Validation Results

Using 2010 land use data and predefined driving factors as inputs, we employed the CLUE-S model to simulate optimal land use allocation for Guangzhou in 2015. The simulation demonstrated strong consistency with the actual land use distribution, achieving an overall accuracy of 82.11% (146,921 correctly simulated grid cells out of 178,924 total). The Kappa coefficient of 0.785 further confirms the model’s reliability, surpassing the widely accepted threshold of 0.75 for satisfactory agreement. These validation results demonstrate that the parameter configuration provides a robust foundation for future land use simulation studies in Guangzhou, supporting the credibility of the 2035 scenario projections presented in the following sections.

3.2. Spatial Optimization Pattern of Land Use Under Different Scenarios

Land use structure optimization results under each scenario were input as demand data into the CLUE-S model, which, combined with various driving factors and conversion constraints, simulated spatial allocations. The optimal land use configuration for Guangzhou in 2035 varied across the four scenarios, reflecting differences in ecological and economic objectives (Figure 2) and land use types (

Table 4. Optimized area of different land use types in Guangzhou in 2035 under different objectives and scenarios (km²).

Optimization Objective	Scenario	Agricultural Land	Forest Land	Grassland	Water Bodies	Construction Land	Unutilized Land
	In 2020	2022.69	3022.73	93.13	473.65	1458.05	2.11
Ecological objective	EPS	1665.19	3232.45	82.03	554.41	1536.47	1.82
	CPS	1762.50	3121.40	82.17	507.42	1596.91	1.85
	EDS	1700.39	3137.71	82.34	507.30	1642.68	1.88
	BDS	1689.14	3184.62	82.01	515.30	1599.41	1.79
Economic objective	EPS	1661.25	3105.50	82.09	504.85	1716.78	1.90
	CPS	1758.6	3032.69	82.45	480.69	1715.76	1.83
	EDS	1680.12	3039.59	82.59	481.84	1786.11	1.77
	BDS	1670.69	3078.51	82.08	484.29	1754.82	1.78

). In addition, to evaluate the accuracy of the simulation results, land use transition matrices for the EPS under ecological objectives and the EDS under economic objectives were compared with the 2020 baseline land-use conditions. The corresponding transition matrices are presented in Tables S9 and S10, and the spatial distribution of land conversion is shown in Figure S4. The results indicate that the differences between the optimized land use areas and the predefined demand targets were minimal, confirming the reliability and consistency of the model’s simulation performance.

For the EPS under the ecological objective, the area of water bodies is projected to reach 554.41 km², largely due to the implementation of ecological protection measures. Major increases are observed in Panyu, Nansha, and the junction between Huadu and Baiyun districts (e.g., Liuxi River, Hongqi Asphalt Road, Jiaomen Waterway, and Humen Waterway) (Figure 2). The expected rise in water bodies was projected to further optimize the flood diversion and detention system of the water network, characterized by north storage, middle drainage, and south drainage. Additionally, it would achieve a balance between regulation and storage, reserve storage and buffer, and functional integration.

Construction land increases under all scenarios for the ecological objective when compared to 2020, with the greatest expansion (1642.68 km²) occurring under the EDS. This growth is primarily driven by the conversion of agricultural land and expansion of existing urban areas, particularly in the central city and Nansha District. The expansion of construction land would provide a foundation for industrial development, accelerate the growth of China-Singapore Knowledge City and Nansha Science City, and promote the construction of the Sci-tech innovation corridor linking Guangzhou–Shenzhen–Hong Kong and Guangzhou–Zhuhai–Macao Nansha District. Simultaneously, construction land would expand on its original foundation, while scattered residential areas would be converted into forest land or agricultural land to facilitate the integrated development of urban and rural areas.

The relationships between the four scenarios under the economic objective follow similar patterns. Lateral comparison of the two objectives revealed that the areas and spatial distributions of agricultural land, grassland, and unutilized land were comparable. The primary differences lie in the changes to forest land, construction land, and water bodies, as these land-use types generate greater ecological value than others, while construction land yields higher economic value. Consequently, depending on whether the focus was on ecological or economic objectives, the simulation process prioritized specific land use classes to meet transformation conditions and maximize target outcomes.

3.3. Comparison of Eco-Economic Value Under Different Scenarios

The ESV and economic benefit for Guangzhou in 2035 were quantitatively assessed under different objectives and scenarios using the ESV equivalent factor and land-based per capita GDP, respectively, to reflect ecological and economic outcomes (

Table 5. Optimized results of land use value under different objectives and scenarios in 2035 (10⁸ CNY).

Optimization Objective	Scenario	ESV	Variation	Economic Benefit	Variation
	2020	157.28	-	16,504.71	-
Ecological objective	EPS	172.76	9.85%	17,510.81	6.10%
	CPS	163.51	3.96%	17,999.38	9.06%
	EDS	163.65	4.05%	18,462.24	11.86%
	BDS	165.98	5.53%	18,042.15	9.32%
Economic objective	EPS	162.27	3.18%	19,209.66	16.39%
	CPS	157.31	0.02%	19,136.46	15.95%
	EDS	157.45	0.03%	19,853.08	20.29%
	BDS	158.64	0.87%	19,539.38	18.39%

). Under the ecological objective, all scenarios achieved varying degrees of improvement in ESV compared to the 2020 baseline. The EPS generated the highest ESV at 172.76 × 10⁸ CNY, marking a 9.85% increase, followed by the BDS at 165.98 × 10⁸ CNY (+5.53%), EDS at 163.65 × 10⁸ CNY (+4.05%), and CPS at 163.51 × 10⁸ CNY (+3.96%). Although EDS did not yield the highest ecological value, it produced the highest economic benefit among the ecological-objective scenarios at 18,462.24 × 10⁸ CNY, representing an 11.86% increase compared to 2020. This indicates that substantial economic gains can still be achieved under ecologically oriented planning through effective spatial optimization. In contrast, under economic objectives, land use configurations favored urban expansion, resulting in significantly higher economic benefits. EDS delivered the greatest economic return, reaching 19,853.08 × 10⁸ CNY, a 20.29% increase over the baseline. BDS and EPS followed closely with benefits of 19,539.38 × 10⁸ CNY (+18.39%) and 19,209.66 × 10⁸ CNY (+16.39%), respectively, while CPS reached 19,136.46 × 10⁸ CNY (+15.95%). However, improvements in ESV under economic objectives were relatively limited. The EPS achieved the highest ESV (162.27 × 10⁸ CNY; +3.18%), followed by BDS (158.64 × 10⁸ CNY; +0.87%), EDS (157.45 × 10⁸ CNY; +0.03%), and CPS (157.31 × 10⁸ CNY; +0.02%).

These findings confirm that although economic objectives can drive higher GDP growth, they offer limited improvements in ecological value. In contrast, the BDS under the ecological objective delivers high ecological outcomes with balanced economic gains, while the EPS under the economic objective provides a promising compromise that ensures ecological security without undermining economic performance. Thus, these two scenarios emerge as the most practical and sustainable options for rapidly urbanizing cities like Guangzhou in future development. The results support the use of integrated simulation approaches for guiding land use policy under rapid urbanization and complex eco-economic constraints.

3.4. Landscape Pattern Analysis Under Different Scenarios

Landscape pattern indices were employed to assess spatial structure, composition, and ecological integrity under each scenario. These include both patch-level metrics, such as proportion of landscape (PLAND), largest patch index (LPI), and aggregation index (AI), and landscape-level metrics, including landscape shape index (LSI), Shannon’s diversity index (SHDI), Shannon’s evenness index (SHEI), and connectance index (CONNECT).

3.4.1. Landscape Pattern Dynamics at the Patch Level

Table 6. Patch-level metrics under different objectives and different scenarios.

Optimization Objective	Type	Scenario	Agricultural Land	Forest Land	Grassland	Water Bodies	Construction Land	Unutilized Land
Ecological Objective	PLAND	2020	28.2592	42.3614	1.3322	7.5555	20.4626	0.0291
		EPS	23.3686	45.6512	1.3106	7.9967	21.6433	0.0296
		CPS	24.7463	44.0567	1.3134	7.3310	22.5230	0.0296
		EDS	23.8604	44.292	1.3151	7.3355	23.1674	0.0296
		BDS	23.7084	44.9699	1.3084	7.4411	22.5425	0.0296
	LPI	2020	7.9880	25.0578	0.1018	2.3124	10.9410	0.0067
		EPS	6.7621	29.6277	0.0246	2.2367	18.9203	0.0022
		CPS	7.2003	26.0384	0.0257	1.8499	19.4915	0.0028
		EDS	7.1013	26.1195	0.0263	1.8606	19.9515	0.0034
		BDS	6.8370	29.3711	0.0246	1.8751	19.5580	0.0017
	AI	2020	73.9246	86.7168	56.4966	63.6228	79.5281	47.1910
		EPS	79.3162	90.7642	24.6244	72.9137	86.5539	12.0879
		CPS	79.9474	90.9876	24.3102	71.4006	86.7574	16.4835
		EDS	79.7481	90.9766	23.8715	72.4981	87.0309	15.3846
		BDS	79.4938	90.8709	24.1876	72.6029	86.8799	9.8901
Economic Objective	PLAND	EPS	23.3680	45.6495	1.3173	7.9950	21.6405	0.0296
		CPS	24.7496	44.0567	1.3106	7.3294	22.5241	0.0296
		EDS	23.8738	44.2948	1.3145	7.3283	24.7647	0.0296
		BDS	23.4535	43.4156	1.3168	7.0197	23.1590	0.0296
	LPI	EPS	6.8845	26.7566	0.0240	2.0126	18.7907	0.0022
		CPS	7.2763	26.1849	0.0252	1.8036	19.6704	0.0022
		EDS	7.2064	26.2586	0.0246	1.8606	20.1147	0.0022
		BDS	7.1790	25.8339	0.0257	1.8404	20.8603	0.0039
	AI	EPS	79.5525	90.4198	22.4653	74.1170	87.0402	14.2857
		CPS	80.0692	90.6801	23.4487	72.4863	86.904	6.5934
		EDS	79.9419	90.6467	23.117	73.1784	87.1971	7.6923
		BDS	79.9313	91.0861	23.9705	72.7134	87.3887	13.1868

displays the changes in patch-level landscape indices for different land use types in 2020 and their projected values in 2035 under both ecological and economic objectives. Among all land use types, agricultural land exhibited the most pronounced changes. Under the economic objective, its PLAND is projected to reach 23.3680% in the EPS and 24.7496% in the CPS, reflecting a notable decline from the 2020 level. In contrast, forest land, owing to its high ecological and economic value, shows a consistent increase in PLAND across all scenarios, with the highest value of 45.6512% observed under the ecological-objective EPS. Construction land is projected to expand under all scenarios due to development pressure, with its PLAND reaching 24.7647% under the economic objective EDS. Meanwhile, the PLAND of grassland and unutilized land remains relatively stable, given their limited ecological and economic contributions.

For agricultural land, LPI is projected to decrease across scenarios, while AI increases, indicating that although large continuous patches are being reduced, the remaining agricultural land becomes more spatially aggregated. This change is driven by the conversion of scattered or marginal farmland into other land-use types, while new agricultural areas are more compactly distributed. In the case of construction land, both LPI and AI increase significantly under all simulations, especially under the EDS. This suggests a trend toward concentrated urban growth, with the emergence of sub-centers and enhanced spatial cohesion. Overall, these changes reflect the contrasting landscape restructuring processes under ecological and economic planning objectives.

3.4.2. Landscape Pattern Dynamics at the Landscape Level

At the landscape level, four indices (i.e., LSI, SHDI, SHEI, and CONNECT) were used to evaluate fragmentation and structural complexity across all land use types (

Table 7. Landscape-level metrics under different objectives and different scenarios.

Optimization Objective	Scenario	LSI	SHDI	SHEI	CONNECT
	2020	45.5721	1.3007	0.7259	0.1059
Ecological objective	EPS	34.2216	1.2902	0.7201	0.1048
	CPS	33.9923	1.2933	0.7218	0.1071
	EDS	33.669	1.2924	0.7213	0.1086
	BDS	33.8877	1.2889	0.7194	0.1065
Economic objective	EPS	34.0798	1.2904	0.7202	0.1055
	CPS	34.003	1.2932	0.7218	0.1100
	EDS	33.7175	1.2923	0.7212	0.1118
	BDS	33.1738	1.2939	0.7221	0.1140

). The LSI reflects the degree of dispersion or aggregation of patches, and higher values indicate greater fragmentation. By 2035, the LSI values across all scenarios are lower than those in 2020, suggesting a reduction in landscape fragmentation. However, it is important to note that this reduction may be driven by the consolidation of construction land into larger, more contiguous urban areas rather than improvements in ecological habitat integrity. The decreased LSI primarily reflects more spatially ordered urban development patterns.

For landscape diversity, both the SHDI and SHEI exhibited consistent trends. Higher SHDI values indicate greater land use richness and heterogeneity, often associated with increased fragmentation. Compared to 2020, the SHDI and SHEI values change only slightly by 2035, indicating that landscape heterogeneity remains relatively stable or declines slightly. The results indicate that through CLUE-S model simulations, scattered patches of agricultural and construction land are reallocated into more cohesive land-use types, thereby improving overall land use efficiency. Furthermore, the CONNECT index reveals enhanced landscape connectivity and spatial cohesion compared to the 2020 baseline. Nonetheless, as with LSI, part of the connectivity increase may be attributable to the aggregation of non-ecological land (e.g., construction land), so interpretations related to habitat integrity should be made with caution and in conjunction with patch-level ecological metrics.

4. Discussion

4.1. Policy Implications of Scenario-Based Land Use Optimization

The simulation results reveal that Guangzhou’s future land use trajectory is highly sensitive to the policy orientation embedded in scenario design. It is important to note that these four scenarios represent extreme policy orientations designed to explore the full range of possible outcomes under different development priorities. In practice, land use planning would likely adopt more moderate approaches that blend elements from different scenarios based on specific local conditions and emerging challenges. Each scenario produced markedly different spatial and functional outcomes, underscoring the crucial role of integrated planning in achieving sustainable development. Notably, under the EPS, forest land expanded to 3232.45 km², primarily through conversion from agricultural and construction land, particularly in ecologically important zones such as Conghua and Huadu Districts. This expansion aligns with Guangzhou’s ecological conservation goals but comes at the cost of reduced agricultural land, which may threaten regional food security. However, the conversion of 357.61 km² of agricultural land to forest requires extensive farmer compensation mechanisms and coordination across multiple administrative jurisdictions. The administrative capacity needed to manage such large-scale land conversion while maintaining social stability represents a considerable institutional challenge. Additionally, the reduction in agricultural land conflicts with China’s food security policies, requiring a careful balance between ecological goals and agricultural preservation requirements.

Conversely, the EDS prioritized urban expansion, with construction land increasing to 1786.11 km², surpassing the allowable development boundary (1772 km²). This scenario generated the highest increase in economic output (20.29%) but simultaneously caused severe landscape fragmentation, as reflected by elevated LSI values and declined landscape connectivity (Table 5). These outcomes suggest that unrestrained urbanization, while beneficial economically in the short term, compromises long-term ecological sustainability and may face regulatory constraints due to exceeding planning boundaries. The implementation feasibility of the EDS is constrained by infrastructure investment requirements and regulatory compliance issues. Supporting such extensive urban expansion would require substantial investment in transportation networks, utilities, and public services that may exceed current municipal fiscal capacity.

The BDS demonstrated the strongest potential for achieving multiple planning objectives, with moderate gains in both economic output (18.39%) and ecosystem service value (5.53%). This scenario preserved key ecological corridors and high-quality agricultural land while supporting compact urban growth. Such results reinforce the argument that scenario-based simulations can effectively illuminate trade-offs and inform spatial strategies that promote synergistic outcomes. These findings are consistent with the broader literature advocating for integrated land use frameworks that accommodate ecological constraints within development planning [45,46,47]. However, the balanced approach requires unprecedented interdepartmental coordination between urban planning, environmental protection, agricultural, and economic development agencies.

4.2. Landscape Pattern Dynamics and Spatial Mechanisms of Change

The landscape pattern analysis revealed distinct spatial dynamics under each scenario, providing insights into the mechanisms driving land use change. Under the EPS, forest land not only increased in quantity but also exhibited improved aggregation (LPI = 29.63) and enhanced connectivity (Table 6 and Table 7), particularly in the northern mountainous and riparian areas (Figure 2). These improvements stem from the protection of ecological corridors and the restoration of water bodies, with the total water area expanding to 554.41 km². While these spatial shifts are consistent with ecological objectives, emphasizing the importance of large, contiguous habitat patches for biodiversity conservation, they came at the cost of displacing urban growth into agriculturally valuable areas, revealing a classic ecological-agricultural trade-off [48,49]. However, the implementation of these changes faces practical constraints related to land ownership patterns and existing development commitments. Creating large, contiguous forest areas requires coordinating across multiple land parcels with different ownership structures, potentially involving lengthy negotiation and acquisition processes.

In contrast, the EDS significantly altered the spatial composition and configuration of land use by converting large areas of agricultural and forest land into construction land. This scenario yielded higher SHDI values (Table 7), indicating increased land use heterogeneity. However, the concurrent reduction in functional connectivity and elevated LSI suggests that such fragmentation may reduce ecological resilience and compromise landscape function, particularly in peri-urban zones where patch cohesion is critical.

The BDS emerged as the most spatially optimized scenario, achieving a delicate balance between land use intensity and ecological coherence. It maintained relatively low landscape fragmentation and a higher CONNECT index, signaling effective retention of structural connectivity among ecological patches. These results underscore the need to consider both compositional aspects (e.g., land use proportions) and configurational characteristics (e.g., patch arrangement) in landscape planning, rather than focusing solely on the total area of land-use types. Policies focused solely on expanding total forest land without addressing spatial configuration may fall short in achieving biodiversity or climate adaptation goals [50,51].

4.3. Practical Recommendations for Sustainable Land Use Planning

Based on the simulation results and implementation challenges identified, differentiated planning strategies should be adopted to support sustainable territorial development. It is crucial to emphasize that our four scenarios represent theoretical extremes designed to illuminate trade-offs and possibilities rather than prescriptive blueprints for implementation. Real-world planning would benefit from combining elements across scenarios and adapting strategies to local conditions, resource constraints, and evolving priorities. Under the EPS, it is essential to enhance the conservation of key water resource areas and increase afforestation efforts, but implementation should be phased to allow for farmer transition support and compensation system development. Priority should be given to protecting existing high-value ecological areas while developing innovative financing mechanisms for restoration activities. For the CPS, given that Guangzhou’s grain self-sufficiency rate remains low (approximately 30%), expanding agricultural land area faces natural constraints. Therefore, efforts should focus on consolidating fragmented construction land in suburban areas, promoting land reclamation and “village-in-city” renovation, improving agricultural land quality, and enhancing rural spatial structure. These measures can help mitigate the land-use conflict between agriculture and urbanization while improving the overall rural environment. Under the EDS, strategies should emphasize improving land development intensity and enhancing the efficient use of existing construction land while recognizing fiscal and regulatory constraints. Compact urban growth and vertical development can alleviate pressure on ecological and agricultural land, but require supporting infrastructure investments that must be carefully sequenced to avoid overextending municipal resources. The BDS represents the most promising but also the most complex implementation scenario. Land use planning should aim to harmonize ecological, agricultural, and urban functions, but success requires building the institutional capacity for integrated planning. We recommend beginning with pilot projects in selected districts to test coordination mechanisms and build administrative expertise before attempting city-wide implementation.

Since the implementation of the Guangdong Province general land use plan, urban expansion in the Pearl River Delta, particularly in Guangzhou, has accelerated significantly. The strategic development of the GBA has further fueled the demand for construction land. However, given spatial constraints, Guangzhou must avoid inefficient and unutilized land use by strengthening land consolidation and fully implementing the “Three Old” renovation policy (i.e., old towns, old factories, and old villages). We recommend controlling the total scale of urban land, delineating strict urban growth boundaries, and establishing flexible and rigid land use control zones across short-, medium-, and long-term horizons, aligned with the goals of integrated regional development in the GBA.

Finally, it is critical to enforce the protection of permanent basic agricultural land and ecological redlines, promote internal urban greening, return ecologically unsuitable agricultural land to forest or other ecological land, and adopt stringent protection measures for rivers and drinking water source areas to prevent further reduction in their spatial extent. These spatial governance measures are necessary to ensure long-term ecological security and sustainable land use in Guangzhou, but their success depends on a realistic assessment of implementation constraints and systematic capacity-building efforts.

4.4. Implications and Limitations

The integrated LP-CLUE-S framework represents a significant methodological advancement by combining mathematical optimization with spatial simulation to overcome the limitations of single-model approaches in land use planning. The framework’s reliability is validated through robust metrics including ROC values exceeding 0.75, a Kappa coefficient of 0.785, and strong consistency between LP optimization targets and CLUE-S spatial allocations. This integration addresses critical gaps in demand forecasting and spatial allocation by leveraging the complementary strengths of both mathematical optimization and GIS-based spatial analysis. The modular design enables transferability to other rapidly urbanizing regions through the adaptation of ecosystem service coefficients, economic indicators, and spatial constraints to local contexts.

In terms of application potential, this framework is highly adaptable to other cities and regions experiencing rapid urbanization and competing policy priorities. By adjusting the driving factors, constraint settings, and ESV parameters, planners can tailor the model to different socio-economic and ecological contexts. Its scenario-based structure enables stakeholders to explore trade-offs between ecological protection, agricultural preservation, and economic growth in a transparent, quantitative manner, facilitating evidence-based decision-making. Moreover, the method aligns with current policy directions in China, such as territorial spatial planning and the “Three Lines” policy (ecological redlines, permanent basic farmland, and urban growth boundaries). The approach can serve as a decision-support tool for implementing integrated land management strategies, supporting interdepartmental coordination, and testing the implications of alternative development pathways before committing to costly or irreversible changes. Its capacity to bridge global methodological advances with localized planning needs makes it a valuable addition to the planning toolkit for sustainable urban and regional development.

Despite these contributions, several limitations should be acknowledged. The model relies on static driver assumptions and inadequately captures feedback mechanisms between land use change and socio-economic dynamics, potentially limiting predictive accuracy over the 15-year planning horizon. The exclusion of climate change impacts represents a critical omission, as sea-level rise, extreme weather events, and hydrological shifts could significantly reshape land use feasibility, particularly in vulnerable areas like the Nansha District. Additionally, the focus on intra-city dynamics overlooks cross-boundary interactions within the broader GBA, where regional connectivity and institutional coordination likely influence metropolitan-scale development patterns. While the study addresses ESV changes, it did not quantify other critical ecosystem functions such as biodiversity maintenance, carbon storage, or hydrological regulation in a spatially explicit way. Future research should therefore expand the evaluation framework through dynamically coupling land use simulation with ecosystem process models, integrating climate adaptation strategies, and incorporating multi-scalar spatial interactions, thereby enhancing both its predictive power and its applicability to diverse regional contexts.

5. Conclusions

This study developed an integrated land use optimization framework by coupling the CLUE-S model with the LP model to simulate spatial patterns and evaluate eco-economic trade-offs under alternative policy scenarios in Guangzhou. By incorporating ecological constraints, economic development demands, and spatial planning regulations, we simulate future landscape changes and regional benefits under four policy scenarios (EPS, EDS, BDS, and CPS), demonstrating how varying spatial configurations of land use generate distinct ecological and economic benefits. The results demonstrate that land use patterns and regional benefits in 2035 vary significantly across scenarios. Under the ecological objective, the EPS significantly improved forest connectivity and increased ESV by 9.85%, though at the expense of agricultural land. Under the economic objective, the EDS maximized economic benefit (20.29%) but also resulted in increased landscape fragmentation and ecological degradation. The CPS successfully preserved agricultural land to support food security, but this led to a reduction in ecological benefits due to urban expansion into environmentally sensitive areas. In contrast, the BDS achieved a more balanced outcome, yielding a 0.87% improvement in ESV and an 18.39% increase in GDP while maintaining overall landscape connectivity and reducing fragmentation. Notably, the BDS under ecological objectives and the EPS under economic objectives emerged as the most effective pathways for promoting sustainable urban development, offering a reasonable balance between environmental protection and economic growth.

The integrated LP–CLUE-S framework overcame the limitations of individual models by combining demand forecasting with spatial allocation, providing a robust tool for land use decision-making. This modeling approach offers valuable theoretical and technical support for strategic land use planning and policy formulation in rapidly urbanizing regions. Future urban development in fast-growing cities like Guangzhou should be guided by adaptive, scenario-based planning strategies that respect ecological thresholds, prioritize compact urban forms, and balance competing land use priorities to ensure long-term regional sustainability.