Evaluation and Prediction of the Water–Energy–Food–Land Nexus: A Case Study of Shanxi Province, China

Abstract

Water–energy–food (WEF) is fundamental for human survival, with land use profoundly impacting their supply-demand relationships. Integrating land into the WEF nexus is crucial for sustainable development. This study used a pressure–state–response model to establish its water–energy–food–land (WEFL) evaluation indicator system. The entropy method and coupling coordination degree (CCD) were applied to assess the WEFL nexus of Shanxi Province during 2000−2023. The obstacle degree model and Geodetector were utilized to identify internal constraints and external drivers, while the ARIMA model was employed to predict future CCD trends. The results show that (1) the comprehensive evaluation index and CCD increased over time, but overall coordination remained limited (average CCD = 0.575). Most regions were at bare to primary coordination levels, indicating persistent subsystem constraints. The spatial pattern evolved from “high in north and south, low in central region” to “high in north and west, low in south and east.” (2) Energy and land subsystems were the main sources of constraints, while the obstacle degrees of the water and food subsystems increased. External drivers shifted from being dominated by government scale and economic growth to being led by technological innovation and urbanization, with growing interaction between anthropogenic and natural factors. (3) The ARIMA model predicted further CCD improvement to intermediate coordination by 2030, although regional disparities persisted. These results provided a scientific basis for resource management and sustainable development in Shanxi Province.

1. Introduction

Water, energy, and food are key material resources essential for human survival and development. With the rapid growth of the global population, it is projected that by 2050, the global demand for water, energy, and food will increase by 55%, 80%, and 60%, respectively [1]. Problems such as water scarcity, energy depletion, and food security have gradually emerged, evolving into severe challenges that threaten sustainable development both regionally and globally [2,3,4]. The interrelationships among the water–energy–food (WEF) system are intricate and complex, and changes in any single variable within one system will trigger interactions and responses across the entire system [5,6]. Therefore, conducting comprehensive evaluations and managing the WEF nexus system is extremely important for achieving regional sustainable development.

The Bonn 2011 Conference (Stockholm Environment Institute) was the event at which water, energy, and food were first inextricably interlinked in a complex nexus, and it was noted that the growth of the global population and economic development would exert enormous pressure on this nexus relationship [7]. Since then, research on the WEF nexus has expanded and become more diversified. Previous research on the WEF nexus has largely focused on three issues. The first type of research focuses on the status assessment of the WEF nexus. Scholars have conducted quantitative evaluations of the WEF system characteristics, such as coupling coordination [8], security [9], pressure [10], efficiency [11], resilience [12], and sustainability [13]. These studies have laid a solid foundation for understanding the nexus’s operational state, but most remain confined to the three core elements of water, energy, and food. In recent years, additional elements such as ecology, land, and carbon have been gradually incorporated into the comprehensive evaluation framework [14,15,16]. For instance, Song et al. [17] analyzed the intricate dynamics within the water–energy–food–ecology (WEFE) nexus in Uzbekistan. Also, van den Heuvel et al. [18] integrated land into the WEF nexus to explore the interactions between the nexus sectors in Sweden from the perspective of the ecosystem services concept.

The second category of research explores the factors influencing the WEF nexus. Most studies have used methods such as the obstacle degree model and gray correlation analysis to identify internal driving factors [19,20,21]. For instance, Lv et al. [22] used the obstacle degree model to confirm that food self-sufficiency rate and average water consumption for energy production were the main obstacle factors affecting the nexus in Tianjin, China. However, the WEF nexus is not a fully independent system, as various external natural and anthropogenic factors all exert an impact on its development level. In recent years, the impact mechanisms of external influencing factors (e.g., climate change, natural disasters, and policy dynamics) on the nexus have also been gradually taken into account [23,24]. Guan et al. [11] applied a Tobit model to analyze the external drivers of WEF coupling efficiency and found that coupling efficiency was higher in the case of more intensive labor force and less precipitation. Wang et al. [25] employed correlation analysis to explore the external influencing factors of the WEF nexus, and the results indicated that precipitation, temperature, and population were strongly correlated with the WEF nexus, whereas nighttime lighting did not exert a direct and significant pressure on the nexus.

The third type of research focuses on simulation prediction and resource management of the WEF nexus. Tools such as the system dynamics (SD) model, life cycle assessment (LCA), back propagation (BP) model, and auto-regressive integrated moving average (ARIMA) model [26,27,28,29] have been used to conduct simulations and predictions for the WEF nexus system. For instance, Ren et al. [30] used the BP model to predict the future coupling coordination of the water–energy–food–land (WEFL) system in Xinjiang, China. Zhang and Wang [31] applied the ARIMA model to forecast the WEF score of various scenarios in the Yellow River Basin of China. Additionally, some studies have further optimized resource allocation schemes through model simulation [32,33]. Sánchez-Zarco and Ponce-Ortega [34] proposed a new integrated circular economy approach based on LCA to optimize generation, use, and distribution of resources in a given region. Du et al. [35] applied the SD model to predict the dynamic consumption of food, energy, and water under 37 sub-scenarios in Melbourne, Australia, and found that residents’ behavior adjustments were the most significant in reducing resource consumption, and the effects of replacing appliances and resource price adjustments were weak.

While advances have been made in the research on the interactions within the WEF nexus, several limitations still remain. First, land is a fundamental element for human survival and development, profoundly influencing water resource storage and recycling, energy development and utilization, as well as food production and transportation. Therefore, it is necessary to link the land element with WEF to explore the coupling mechanism between systems. However, current research on the coupling coordination relationship of the water–energy–food–land (WEFL) nexus is still insufficient. Second, most studies have analyzed the impact of the internal factors within the system on the nexus relationship, while relatively few studies have comprehensively evaluated the mechanisms by which external driving factors influence the nexus relationship. Finally, the existing studies have mainly focused on quantitative assessments of the historical nexus relationship, with no future predictions of the evolution of the nexus relationship.

Shanxi Province is located in the middle reaches of the Yellow River and serves as China’s core energy base [36]. The coal reserves of Shanxi account for 23% of the total of China, and its coal output has accounted for 29% of the national total in 2023, ensuring a stable supply of coal to 24 provinces across the country. While consolidating national energy security, Shanxi has also borne severe environmental costs, including land damage, ecological degradation, water resource depletion, and environmental pollution [37,38]. Notably, Shanxi’s resource endowment presents a distinct structural imbalance characterized by abundant coal, scarce water and limited land: it occupies merely 1.63% of China’s total land area and 0.4% of the national water resources, yet in 2023, it still provided domestic water for 2.45% of the national population, and supported 2.91% of the national arable land and 2.08% of the national grain production. The stark contrast between its limited resource endowment and the excessive socioeconomic carrying tasks has resulted in acute internal contradictions within the WEFL nexus. For a long time, capital and development factors in Shanxi have been excessively concentrated in the energy subsystem, leading to the marginalization of agricultural development and increased vulnerability of the water subsystem. The root cause of these issues lies in the province’s long-term reliance on a traditional resource exploitation model [39]. In addition, as a key region for the national strategy of Ecological Protection and High-Quality Development of the Yellow River Basin, the coordinated development of Shanxi’s WEFL nexus directly affects the overall ecological security and economic sustainability of the entire Yellow River Basin. In summary, Shanxi’s unique characteristics, national strategic importance, and prominent contradictions in the WEFL nexus make it a typical sample for investigating the coupling coordination development of the WEFL nexus. This study selects Shanxi Province as the research area, and the relevant findings are expected to provide valuable references for other energy-dependent regions worldwide.

The objectives of this study were therefore (1) to establish a quantified framework and evaluate the coupling coordination relationship of the WEFL nexus in Shanxi Province, (2) to identify the effect of internal and external factors on the coupling coordination relationship of the WEFL nexus, and (3) to predict the development trend of coupling coordination level during the period of the 15th Five-Year Plan of China (2026–2030).

2. Materials and Methods

2.1. Study Area

Shanxi Province (34°35′–40°45′ N, 110°14′–114°33′ E) is located in central China, with a total land area of approximately 156,700 km² [40]. The province administers 11 prefecture-level cities: Datong, Shuozhou, Xinzhou, Lvliang, Taiyuan, Yangquan, Jinzhong, Linfen, Changzhi, Yuncheng, and Jincheng (Figure 1). The altitude of the region ranges from 231 to 3055 m, and the overall topography is characterized by higher elevations in the east and west and lower elevations in the center. The terrain is primarily composed of mountains, hills, and basins, with mountains and hills accounting for over 80% of the total area. Shanxi has a temperate continental monsoon climate, with an average annual precipitation of approximately 508.8 mm. There is a spatial distribution pattern of higher precipitation in the southeast and lower precipitation in the northwest. The per capita water resources in Shanxi Province stand at 381 m³ in 2023, merely one-sixth of the national average in China. As a major energy supply base in China, 10 of the 11 prefecture-level cities (excluding Taiyuan) are resource-based cities, and the coal-bearing area accounts for 45.33% of the total area [41]. Restricted by the topography, both the cultivated land area and grain output of Shanxi Province do not rank highly nationally.

2.2. Data Sources

The data used in this study covered the period from 2000 to 2023 and included water, energy, food, and land system indicator data, together with socioeconomic and meteorological data. The data were mainly obtained from the Shanxi Statistical Yearbook, the Shanxi Water Resources Bulletin, the Shanxi Rural Statistical Yearbook, and the Statistical Bulletins of National Economic and Social Development of each prefecture-level city.

2.3. Methods

The research framework of this study is divided into the following four steps (Figure 2). Firstly, the WEFL evaluation indicator system was constructed based on the pressure–state–response (PSR) model, covering 4 subsystems and 40 indicators. Secondly, the coupling coordination relationship of the WEFL system was quantitatively assessed using the entropy weight model, comprehensive evaluation index (CEI) model, and coupling coordination degree (CCD) model. Thirdly, the obstacle degree model and Geodetector were utilized to identify internal constraints and external drivers. Finally, the ARIMA model was employed to forecast the CCD from 2026 to 2030.

2.3.1. Construction of the WEFL Evaluation Indicator System

Constructing a scientifically sound and appropriate indicator evaluation framework is crucial for ensuring the accuracy of evaluation results. With population growth, unsustainable resource use, and the intensification of climate change, the WEFL system is facing various pressures. These pressures alter the state of resource availability, which in turn forces human beings to conduct responsive regulation and management. The PSR model, a problem analysis framework based on a causal transmission relationship, is well-suited to analyzing the complex interactions between human activities and resource-environment systems [42,43]. It has been used to evaluate the sustainable development of water, energy, and food resources [44,45]. Therefore, this study adopted the PSR model as the theoretical basis for constructing the WEFL nexus evaluation indicator system, adhering to four core selection principles to ensure the rationality and scientificity of indicators: scientific rigor, data accessibility, system comprehensiveness, and regional pertinence.

The specific principles for indicator selection are elaborated as follows: (1) Scientific rigor: All indicators are closely aligned with the connotation of the PSR framework, accurately reflecting the causal relationship between human activities and the WEFL system. Specifically, pressure indicators characterize human demand and external stress on the system, state indicators reflect the resource endowment and status of each subsystem, and response indicators represent human management measures and policy implementations for the system. (2) Data accessibility: Indicators are selected based on the availability of long-term continuous data from official statistical yearbooks, government reports, and authoritative databases, ensuring the operability of data collection and calculation. (3) System comprehensiveness: Indicators cover all four subsystems and the three dimensions of the PSR model, avoiding the omission of key processes and ensuring the full reflection of the overall characteristics of the WEFL nexus. (4) Regional pertinence: Combined with the actual regional characteristics of Shanxi Province—a typical energy-rich region with a fragile ecological environment, prominent soil erosion, and unbalanced resource allocation—indicators that can reflect its regional attributes are specially incorporated to enhance the applicability of the evaluation system.

Based on the above principles, integrated with the existing literature on WEF nexus evaluation [19,21] and the actual conditions of Shanxi Province, this study ultimately identified 40 evaluation indicators in total for the integrated WEFL nexus, comprising 11, 8, 10, and 11 indicators for the water, energy, food, and land subsystems respectively, with a roughly balanced indicator distribution across subsystems to avoid bias caused by uneven indicator allocation. Specifically, the indicators for the water subsystem are selected from four aspects: consumption structure, utilization efficiency, resource endowment, and ecological management. The indicators for the energy subsystem are chosen from three aspects: consumption intensity and structure, supply and self-sufficiency, and environmental response. The indicators for the food subsystem are derived from three aspects: input intensity and consumption, production capacity and security, and agricultural modernization and efficiency. The indicators for the land subsystem are selected from four aspects: land pressure and environmental stress, land endowment and structural allocation, economic output intensity, and ecological and managerial response.

The exclusion of certain theoretically relevant variables was mainly due to two reasons: First, the lack of long-term continuous data. For example, water quality parameters and industrial water pollution discharge intensity were excluded because consistent monitoring data for the entire study period were not available in official databases, and discontinuous data would lead to inaccurate evaluation results. Second, inconsistent data caliber. Individual indicators had inconsistent statistical standards in different years, making it impossible to conduct unified quantitative calculation and comparison.

Detailed information of all indicators is presented in

Table 1. The evaluation indicator system for the coupling coordination relationship of the WEFL nexus using the PSR model.

Criterion Layer	Sub-Criteria Layer	Indicator Layer	Attribute	Weight
Water Subsystem	Pressure	Per capita water consumption (W1)	−*	0.062
		Proportion of agricultural water consumption (W2)	−	0.163
		Proportion of industrial water consumption (W3)	−	0.033
		Proportion of domestic water consumption (W4)	−	0.049
		Water consumption per 10,000 yuan of GDP (W5)	−	0.032
		Water resources development and utilization rate (W6)	−	0.019
	State	Per capita water resources amount (W7)	+	0.169
		Water yield modulus (W8)	+	0.124
	Response	Proportion of ecological water consumption (W9)	+	0.201
		Urban sewage treatment rate (W10)	+	0.054
		Urban industrial water reuse rate (W11)	+	0.095
Energy Subsystem	Pressure	Per capita energy consumption (E1)	−	0.025
		Energy consumption per 10,000 yuan of GDP (E2)	−	0.022
		Industrial sulfur dioxide (SO2) emissions (E3)	−	0.051
	State	Per capita primary energy production (E4)	+	0.212
		Energy self-sufficiency rate (E5)	+	0.126
		Proportion of coal consumption (E6)	−	0.166
	Response	Decline rate of energy consumption per unit of GDP (E7)	+	0.033
		Energy conservation and environmental protection expenditure in general public budget (E8)	+	0.366
Food Subsystem	Pressure	Intensity of fertilizer application (F1)	−	0.055
		Intensity of pesticide use (F2)	−	0.055
		Per capita grain consumption (F3)	−	0.020
	State	Per capita grain yield (F4)	+	0.134
		Per unit area food production (F5)	+	0.054
		Engel’s coefficient (F6)	−	0.043
		Grain self-sufficiency rate (F7)	+	0.152
	Response	Per unit area total agricultural machinery power (F8)	+	0.173
		Effective irrigation rate of farmland (F9)	+	0.207
		Irrigation water use efficiency coefficient (F10)	+	0.107
Land Subsystem	Pressure	Per unit area wastewater discharge rate (L1)	−	0.014
		Multiple cropping index (L2)	−	0.045
		Soil erosion intensity (L3)	−	0.007
	state	Per capita urban construction land area (L4)	+	0.118
		Per capita cultivated land area (L5)	+	0.072
		Forest coverage rate (L6)	+	0.065
		Per unit area GDP (L7)	+	0.258
		Per capita road area (L8)	+	0.088
	Response	Green coverage rate in urban built-up areas (L9)	+	0.030
		Soil erosion control rate (L10)	+	0.219
		Mechanized farming area ratio (L11)	+	0.083

* − represents a negative indicator (the higher the value, the worse the system status); + represents a positive indicator (the higher the value, the better the system status).

.

2.3.2. The Comprehensive Evaluation Index Model

Because the selected indicators differed in numerical magnitude, attribute, and dimension, the original data had to be standardized. The specific formulas are as follows:

$$X_{ijt}={\displaystyle \frac{x_{ijt}-\min (x_i)}{\max (x_i)-\min (x_i)}} (\mathrm{positive} \mathrm{indicators})$$

(1)

$$X_{ijt}={\displaystyle \frac{\max (x_i)-x_{ijt}}{\max (x_i)-\min (x_i)}} (\mathrm{negative} \mathrm{indicators})$$

(2)

where X_ijt is the standardized data of indicator i in region j in year t, x_ijt is the original data of the indicator i, and max(x_i) and min(x_i) are the maximum and minimum values of the original data of the indicator i, respectively.

To eliminate the inherent randomness and subjective conjecture associated with subjective weighting methods, we adopted the entropy weight method [46,47] to objectively assign weights to each indicator. The weight was calculated as follows:

$$p_{ijt}={\displaystyle \frac{X_{ijt}}{{\displaystyle \sum _{j=1}^{11}{\displaystyle \sum _{t=1}^{24}{X}_{ijt}}}}}$$

(3)

$$e_i=-{\displaystyle \frac{{\displaystyle \sum _{j=1}^{11}{\displaystyle \sum _{t=1}^{24}{p}_{ijt}}}\ln (p_{ijt})}{\ln (11\times 24)}}$$

(4)

$$w_i=-{\displaystyle \frac{1-e_i}{{\displaystyle \sum _{i=1}^I(1-e_i)}}}$$

(5)

where p_ijt is the proportion of the value of the indicator i in region j in year t, and e_i and w_i are the information entropy and weight of indicator i, respectively. The weights of the indicators of the four subsystems are presented in Table 1.

Then, the comprehensive evaluation index of each subsystem was calculated using the standardized value and the corresponding weight for each indicator [48]. Given that the four subsystems were of equal importance in the WEFL nexus, the comprehensive evaluation index of the WEFL system was calculated as the arithmetic mean of the four subsystems [17]. The corresponding formulas are as follows:

$$\begin{array}{l}CE{I}_w={\displaystyle \sum _{i=1}^{10}{p}_{ijt\_w}{w}_{i\_w}}\\ CE{I}_e={\displaystyle \sum _{i=1}^{12}{p}_{ijt\_e} w_{i\_e}}\\ CE{I}_f={\displaystyle \sum _{i=1}^{13}{p}_{ijt\_f} w_{i\_f}}\\ CE{I}_l={\displaystyle \sum _{i=1}^{14}{p}_{ijt\_l}{w}_{i\_l}}\end{array}$$

(6)

$$CEI={\displaystyle \frac{CE{I}_w+CE{I}_e+CE{I}_f+CE{I}_l}{4}}$$

(7)

where CEI is the comprehensive evaluation index of the WEFL system, CEI_w, CEI_e, CEI_f, and CEI_l are the water, energy, food, and land subsystems, respectively.

2.3.3. The Coupling Coordination Degree Model

The WEFL system is a complex and dynamic coupled system. Although the coupling degree can measure the degree of interaction between subsystems, it fails to reflect the overall coordinated development level of the system [49,50]. In contrast, CCD better reflects the overall function of the system and the synergistic development among subsystems [51,52]. Therefore, we adopted the CCD model [53] to characterize the coupling coordination relationship of the WEFL system. The specific formula is as follows:

$$C={\displaystyle \frac{4\times \sqrt[4]{CE{I}_w\times CE{I}_e\times CE{I}_f\times CE{I}_l}}{CE{I}_w+CE{I}_e+CE{I}_f+CE{I}_l}}$$

(8)

$$D=\sqrt{C\times T}$$

(9)

where C is the coupling degree of the WEFL system, D is the CCD of the WEFL system, and T is the CEI of the system, which reflects the overall coordination effect of each subsystem of the system.

It should be clarified that although the CCD model does not directly quantify physical resource flows, a higher CCD value indicates a system state more conducive to the sustainable management of such inter-sectoral resource flows, and can effectively reflect the actual interdependencies among the WEFL subsystems. Based on relevant research results [54,55,56], the CCD of the WEFL nexus can be divided into 10 levels and three stages (

Table 2. Classification standards for coupling coordination levels and stages.

D Interval	Coupling Coordination Level	Coupling Coordination Stage
0.0 < D ≤ 0.1	Extreme disorder	Dysfunctional stage
0.1 < D ≤ 0.2	Severe disorders
0.2 < D ≤ 0.3	Moderate disorder
0.3 < D ≤ 0.4	Mild disorders
0.4 < D ≤ 0.5	Near-disorder	Transitional stage
0.5 < D ≤ 0.6	Barely coordinated
0.6 < D ≤ 0.7	Primary coordination
0.7 < D ≤ 0.8	Intermediate coordination	Coordinated stage
0.8 < D ≤ 0.9	Virtuous coordination
0.9 < D ≤ 1.0	Quality coordination

).

2.3.4. The Obstacle Degree Model

To identify the key factors restricting the coupling coordination development of the WEFL nexus system, we employed the obstacle degree model [57] to analyze the impact of internal system factors on its coupling coordination development. The specific formula is as follows:

$$d_{ijt}=1-X_{ijt}$$

(10)

$$O_{ijt}={\displaystyle \frac{d_{ijt}{w}_i}{{\displaystyle \sum _{j=1}^n{d}_{ijt}{w}_i}}}\times 100\%$$

(11)

$$U={\displaystyle \sum _{i=1}^I{O}_{ijt}}$$

(12)

where d_ijt is the deviation degree of indicator i in region j in year t, O_ijt is the obstacle degree of the indicator i, n is the number of evaluation units, and U is the obstacle degree of each subsystem. A higher value of O_ijt indicates a greater hindrance of the indicator to the coupling coordination development of the WEFL nexus system.

2.3.5. The Geodetector Model

The Geodetector is a statistical method for detecting the spatial differentiation of a certain element and revealing its underlying driving factors [58]. Its core principle is to quantify the spatial correlation between independent and dependent variables to identify the driving forces causing spatial heterogeneity of the dependent variable. It has been widely applied to study multi-dimensional systems such as the natural environment and land use [59]. Therefore, we used the factor detection and interaction detection functions of the Geodetector to reveal the influence degree and interaction effects of external driving factors on the coupling coordination development of the WEFL nexus. The primary purpose of factor detection was to assess the explanatory power of individual factors on the spatial heterogeneity of the dependent variable, which is typically measured by the q statistic. It was calculated as follows [60]:

$$q=1-{\displaystyle \frac{{\displaystyle \sum _{h=1}^L{N}_h{\sigma }_h^2}}{N{\sigma }^2}}$$

(13)

where q is the explanatory power of driving factors on the spatial differentiation of the dependent variable Y, h is the number of factor stratifications, with a total of L layers. N_h and N are the sample sizes of the h-th layer and the whole area, respectively, while ${\sigma }_h$ and $\sigma$ are the variances of Y values in the h-th layer and the entire study area, respectively. The larger the q value, the stronger the explanatory power of the driving factors.

Interaction detection was used to analyze the interactive effects of two or more factors on the spatial heterogeneity of the dependent variable. There were five types of factor interactions [15], as shown in

Table 3. Types of interactive effects between two independent variables.

Judgment Basis	Interaction Types
$q(x_1\cap x_2)<\min (q\left(x_1\right),q\left(x_2\right))$	Nonlinear weakening
$\min (q\left(x_1\right),q\left(x_2\right))<q(x_1\cap x_2)<\max (q(x_1),q(x_2\left)\right)$	Single-factor nonlinear weakening
$\max (q\left(x_1\right),q\left(x_2\right))<q(x_1\cap x_2)<q(x_1)+q(x_2)$	Dual-factor enhancement
$q(x_1\cap x_2)=q(x_1)+q(x_2)$	Mutual independence
$q(x_1\cap x_2)>q\left(x_1\right)+q\left(x_2\right)$	Nonlinear enhancement

.

Drawing on relevant studies [11,25] and combining the characteristics of the study area, we selected nine external influencing factors as independent variables and used the CCD as the dependent variable to determine the impacts of external influencing factors on the coupling coordination level of the WEFL system. These nine influencing factors reflected the driving effects of three dimensions (climate, topography, and human activity), as presented in

Table 4. External driving factors of the coupling coordination level of the WEFL system.

Types	Driving Factors	Factor Interpretation
Climatic factors	Precipitation (X1)	Annual precipitation
	Temperature (X2)	Mean annual temperature
Topographic factors	Slope (X3)	Topographic slope
	Elevation (X4)	Average elevation
Human factors	Per capita GDP (X5)	Regional GDP/total population
	Population density (X6)	Total population/total area of the region
	Urban population share (X7)	Urban population/total population
	Level of government size (X8)	Government expenditure/regional GDP
	Intensity of scientific and technological investment (X9)	Scientific and technological investment/regional GDP

. All continuous driving factor variables were discretized into strata using the natural breaks method, and the number of strata for each variable was set to three.

2.3.6. The ARIMA Model

The ARIMA model is a commonly used time series forecasting model. It can achieve favorable fittings for stationary sequences and is characterized by high prediction accuracy and simple operation [31,61]. Therefore, we adopted the ARIMA model to simulate the CCD time series. The CCD data from 2000 to 2018 were used as the training set, and the data from 2019 to 2023 were used as the test set to verify the simulation performance of the model. Finally, the model was used to forecast the CCD of the whole province of Shanxi and its 11 prefecture-level cities for the period of 2026–2030. The formula of the model is as follows:

$${Y_t}^{\prime }=c+{\displaystyle \sum _{i=1}^p{\phi }_i{Y_{t-i}}^{\prime }}+{\epsilon }_t-{\displaystyle \sum _{j=1}^q{\theta }_j{\epsilon }_{t-j}}$$

(14)

where Y_t′ is the stationary series derived from the original series after order differencing; c represents the constant term; p and q are the orders of the autoregressive and moving average components, respectively; ϕ_i and θ_j are the autoregressive and moving average coefficients; and ε_t is the white noise random error term at time t.

3. Results

3.1. Analysis of the Comprehensive Evaluation Index for the WEFL System

The variations in the comprehensive evaluation indexes of the WEFL subsystems and the coupled system in Shanxi Province from 2000 to 2023 are shown in Figure 3.

In the water resource system, the CEI_w of all cities and the whole of Shanxi Province increased by varying degrees over the study period, with the growth ranging from 16.10% to 114.60%. This was closely related to the implementation of water resource development and utilization policies in Shanxi Province, as well as the increase in precipitation. The multi-year average CEI_w in Shanxi Province was 0.415, and the ranking of CEI_w among cities was as follows: Jincheng > Yangquan > Changzhi > Taiyuan > Datong > Jinzhong > Lvliang > Shuozhou > Xinzhou > Linfen > Yuncheng. The relatively high CEI_w values in Jincheng (0.544) and Changzhi (0.450) were attributed to their large natural water resources, high proportion of ecological water use, and low per capita water consumption. In Yangquan and Taiyuan, the key contributing factors included a low proportion of agricultural water use and a high urban industrial water reuse rate. In contrast, Yuncheng, Linfen, and Xinzhou recorded lower CEI_w scores (0.316, 0.367, and 0.368, respectively), primarily due to their excessively high proportion of agricultural water use (exceeding 60%) and a notably lower urban industrial water reuse rate than other cities.

In the energy system, the CEI_e of all cities and the whole of Shanxi Province exhibited a distinct upward trend over the study period, with increases ranging from 36.91% to 161.24%. This growth was attributed to rising energy demand and the transformation and upgrading of industries. The multi-year average CEI_e for Shanxi Province was 0.290, and the ranking of CEI_e among cities was as follows: Shuozhou > Yangquan > Taiyuan > Linfen > Jincheng > Datong > Lvliang > Changzhi > Xinzhou > Jinzhong > Yuncheng. Shuozhou and Yangquan are the most important energy bases in Shanxi Province, with both having high energy production and energy self-sufficiency rates, which explained their top rankings in the CEI_e values. Taiyuan and Linfen also achieved relatively high CEI_e scores (0.317 and 0.307), which was due to their low proportion of coal consumption and high proportion of energy conservation and environmental protection expenditure, reflecting strong performances in energy saving and emission reduction. Conversely, Yuncheng recorded the lowest CEI_e (0.161) due to its relatively poor energy production. Its per capita primary energy production was only one-sixtieth of that of Shuozhou, and its energy self-sufficiency rate was only 39.75%. Jinzhong and Xinzhou also had lower CEI_e values, primarily because of their relatively low proportion of energy conservation and environmental protection expenditure.

In the food system, the CEI_f of Taiyuan and Jincheng exhibited a slight decreasing trend interannually, while the CEI_f of other cities presented an increasing trend with a growth range of 2.01% to 59.86%. This was mainly attributed to the implementation of the national food security strategy and agricultural modernization. The multi-year average CEI_f of Shanxi Province was 0.371, with obvious variations among different cities. The ranking was as follows: Yuncheng > Shuozhou > Jinzhong > Linfen > Xinzhou > Lvliang > Datong > Changzhi > Jincheng > Taiyuan > Yangquan. Yuncheng, Shuozhou, Jinzhong, and Linfen had high per capita grain output and grain self-sufficiency rates, as well as a large total agricultural machinery power per unit area and a high effective farmland irrigation rate, thus achieving relatively high CEI_f values (0.605, 0.481, 0.439, and 0.435). In contrast, due to low per capita grain output and food self-sufficiency rates, as well as relatively weak agricultural modernization and irrigation levels, Yangquan and Taiyuan had notably lower CEI_f values (0.210 and 0.207) than other cities.

In the land system, the CEI_l of various cities and the entire Shanxi Province exhibited a considerable increasing trend interannually, with the largest growth among the four subsystems, ranging from 78.84% to 168.16%. This was mainly attributed to the implementation of land management policies and socioeconomic development. The multi-year average CEI_l of Shanxi Province was 0.310, and the ranking of CEI_l among cities was as follows: Linfen > Shuozhou > Xinzhou > Lüliang > Taiyuan > Datong > Jinzhong > Changzhi > Jincheng > Yuncheng > Yangquan. Linfen ranked first in CEI_l due to its high soil erosion control rate and large ratio of mechanized farming area, as well as its high forest coverage. Shuozhou also achieved a relatively high CEI_l value (0.357) because of its high gross domestic product (GDP) per unit area, and large per capita road area. Yangquan had the lowest CEI_l value (0.226) due to its small per capita cultivated land area, as well as low levels of soil erosion control and mechanized farming. The low CEI_l value of Yuncheng was due to its small per capita construction land and road areas.

Because the comprehensive evaluation indexes of the four subsystems mostly showed an increasing trend during the study period, the CEI of the WEFL nexus system in Shanxi Province also exhibited an interannual growth trend, with a growth range of 31.33% to 95.87%. The multi-year average CEI for Shanxi Province was 0.346, and the CEI ranking among cities was as follows: Shuozhou (0.398) > Linfen (0.371) > Jincheng (0.353) > Lvliang (0.348) > Xinzhou (0.347) > Jinzhong (0.342) > Datong (0.341) > Yuncheng (0.339) > Changzhi (0.330) > Yangquan (0.322) > Taiyuan (0.321). The CEI values of Shuozhou and Linfen were obviously higher than those of other cities, which was mainly attributed to their high scores in the energy, food, and land subsystems, while the differences in CEI values among the other cities were relatively small.

3.2. Spatio-Temporal Variations in the Coupling Coordination Relationship Within the WEFL Nexus

The annual CCD results of the WEFL system for Shanxi Province as a whole and its various cities are presented in Figure 4. Overall, the coupling coordination level of Shanxi’s WEFL nexus was relatively low, with the CCD values varying from 0.458 to 0.728 and a multi-year average of 0.575 for the entire province. This indicated a strong mutual constraint among the four systems of water resources, energy, food, and land, and that synergistic development remains suboptimal, being in a transitional coordination stage for much of the study period. Driven by the interannual growth trend of CEI, the coupling coordination development level of Shanxi’s WEFL system also improved, with CCD growth across cities, ranging from 17.74% to 39.76%. This trend suggested that, despite ongoing conflicts and tensions among the water, energy, food, and land systems in Shanxi Province, factors such as policy regulation, optimized resource allocation, agricultural modernization, and strengthened ecological protection measures contributed to a gradual improvement in coupling coordination among the elements, with system operations progressively moving toward positive interaction.

In terms of spatial variation, the coupling coordination level of Shanxi Province’s WEFL nexus exhibited specific regional differences (Figure 5). In 2000, the ranking of cities by CCD was: Shuozhou (0.535) > Linfen (0.532) > Jincheng (0.526) > Datong (0.518) > Changzhi (0.501) > Taiyuan (0.500) > Jinzhong (0.485) > Xinzhou (0.480) > Yangquan (0.477) > Yuncheng (0.473) > Lvliang (0.471). The overall spatial distribution pattern was characterized as “high in the north and south, low in the central region” (with the exception of Yuncheng). Among these cities, Shuozhou, Linfen, Jincheng, Datong, and Changzhi, all located in northern and southern Shanxi, had CDD values at the barely coordinated level. In contrast, most central cities recorded CCD values below 0.5, falling into the near-disorder category. By 2023, the CCD ranking across cities had changed to Shuozhou (0.723) > Lvliang (0.691) > Linfen (0.682) > Jinzhong (0.678) > Xinzhou (0.676) > Datong (0.666) > Taiyuan (0.665) > Changzhi (0.655) > Jincheng (0.648) > Yuncheng (0.639) > Yangquan (0.583). With the decline in the CCD rankings of Jincheng and Changzhi and the substantial increase in Lvliang, the spatial distribution pattern of the system CCD experienced changes, generally presenting the characteristics of “high in the north and low in the south, high in the west and low in the east.” Nevertheless, CCD values improved in all prefecture-level cities, with most reaching the primary coordination level.

3.3. The Impact of Internal Influencing Factors on the Coupling Coordination Level

The 10 main obstacles influencing the CCD of the WEFL nexus system in Shanxi Province from 2000 to 2023 are presented in

Table 5. The main obstacles affecting the coupling coordination level of the WEFL system in Shanxi Province from 2000 to 2023.

Year		2000	2003	2006	2009	2012	2015	2018	2021	2023
1st	OF */ OD (%)	E8/ 12.398	E8/ 12.490	E8/ 12.665	E8/ 12.882	E8/ 12.748	E8/ 11.927	E8/ 10.415	E8/ 8.812	L7/ 8.443
2nd	OF/ OD (%)	L7/ 8.674	L7/ 8.721	L7/ 8.711	L7/ 8.733	L7/ 8.513	L7/ 8.809	L7/ 8.751	L7/ 8.675	E8/ 7.622
3rd	OF/ OD (%)	L10/ 6.888	L10/ 6.844	L10/ 6.768	L10/ 6.585	W9/ 6.374	E4/ 6.605	E4/ 6.683	E4/ 6.666	F9/ 6.741
4th	OF/ OD (%)	E4/ 6.701	E4/ 6.563	W9/ 6.391	E4/ 6.257	L10/ 6.290	L10/ 6.064	F8/ 6.201	F9/ 6.487	E4/ 6.521
5th	OF/ OD (%)	F9/ 5.482	W9/ 6.299	E4/ 6.348	W9/ 5.890	E4/ 6.113	W7/ 5.891	F9/ 6.000	F8/ 6.472	E6/ 6.515
6th	OF/ OD (%)	W7/ 5.044	F9/ 5.527	F9/ 5.608	F9/ 5.686	F9/ 5.822	F9/ 5.824	L10/ 5.898	E6/ 6.402	F8/ 6.457
7th	OF/ OD (%)	F8/ 4.866	F8/ 4.724	W7/ 5.252	W7/ 5.498	W7/ 5.478	W9/ 5.432	W7/ 5.700	L10/ 6.049	L10/ 5.993
8th	OF/ OD (%)	W9/ 4.673	W7/ 4.652	F8/ 4.540	F8/ 4.414	F8/ 4.238	F8/ 4.279	E6/ 5.458	W7/ 4.961	W7/ 5.192
9th	OF/ OD (%)	F7/ 4.058	F7/ 3.862	W8/ 3.787	W8/ 3.967	W8/ 3.930	W8/ 4.202	W9/ 5.289	W9/ 4.495	W9/ 4.865
10th	OF/ OD (%)	W8/ 3.650	W8/ 3.334	F7/ 3.774	F7/ 3.899	W2/ 3.547	W2/ 3.929	W8/ 4.140	W2/ 3.849	W2/ 3.997

* OF represents the obstacle factor. OD represents the obstacle degree. The red color in a cell indicates that the ranking of the OD of this obstacle in the current year decreased compared with the previous year, while green indicates a corresponding rise in the ranking.

. Based on the multi-year average results, energy conservation and environmental protection expenditure in the general public budget (E8), GDP per unit area (L7), per capita primary energy production (E4), soil erosion control rate (L10), and effective irrigation rate of farmland (F9) were the key obstacles affecting the coupling coordination development of the Shanxi WEFL nexus, with their obstacle degrees being 11.58%, 8.69%, 6.48%, 6.37%, and 5.84%, respectively. The proportion of ecological water consumption (W9), per capita water resources (W7), total agricultural machinery power per unit area (F8), percentages of coal consumption (E6), and water yield modulus (W8) were relatively secondary internal influencing factors, with their obstacle degrees at 5.53%, 5.41%, 5.04%, 4.24%, and 3.89%, respectively. From the perspective of interannual changes, the obstacle degree rankings of L10, W7, and grain self-sufficiency rate (F7) declined. In contrast, the rankings of three indicators, effective irrigation rate of farmland (F9), proportion of agricultural water consumption (W2), and proportion of coal consumption (E6), rose from 5th, 11th, and 13th in 2000 to 3rd, 10th, and 5th in 2023, respectively, making them important obstacles restricting the sustainable development of the system.

Figure 6 presents the obstacle degrees of the water, energy, food, and land subsystems in Shanxi Province for 2000, 2008, 2016, and 2023. The multi-year average obstacle degrees for the province water, energy, food, and land subsystems were 22.42%, 27.17%, 24.09%, and 26.31%, respectively. This indicated that the energy and land subsystems exerted stronger restrictive effects on the coupling coordination development of the provincial WEFL nexus than the food and water subsystems. However, the dominant obstacle subsystems varied slightly across the different cities. For example, the coupling coordination development of Taiyuan, Yangquan, and Jincheng was more notably constrained by the food and land subsystems, while the obstacle degrees of the land and water resources subsystems in Shuozhou were higher than those of the energy and food subsystems. In terms of interannual changes, the obstacle degrees of the energy and land subsystems decreased in most regions, whereas those of the water and food subsystems increased. This trend indicated that the restrictive effects of the water and food subsystems on the regional coupling coordination development in Shanxi Province were gradually becoming prominent.

3.4. Analysis of External Driving Factors on the Coupling Coordination Level

Table 6. The explanatory power of the external driving factors of the CCD in Shanxi Province from 2000 to 2023.

2000		2008		2016		2023
Driving Factors	q-Value	Driving Factors	q-Value	Driving Factors	q-Value	Driving Factors	q-Value
X8	0.399	X9	0.492	X4	0.553	X9	0.589
X5	0.224	X4	0.284	X9	0.513	X7	0.583
X3	0.171	X7	0.224	X2	0.276	X1	0.380
X9	0.140	X2	0.206	X3	0.244	X4	0.321
X7	0.138	X1	0.185	X6	0.226	X5	0.165
X4	0.137	X5	0.183	X5	0.194	X2	0.152
X2	0.107	X6	0.151	X1	0.017	X3	0.137
X6	0.016	X8	0.066	X7	0.017	X6	0.083
X1	0.001	X3	0.016	X8	0.015	X8	0.013

presents the explanatory power of external driving factors on the spatial differentiation of CCD in Shanxi Province over the study period. There were obvious disparities in the explanatory power of each influencing factor on the CCD, and the ranking of driving factors by their explanatory power also changed considerably over time. In 2000, the ranking of explanatory power for the driving factors was X8 (0.399) > X5 (0.224) > X3 (0.171) > X9 (0.140) > X7 (0.138) > X4 (0.137) > X2 (0.107) > X6 (0.016) > X1 (0.001). The system coupling coordination was primarily driven by the level of government scale, with per capita GDP as a secondary driving factor, while the influence of other factors was relatively weak. This suggested strong characteristics of administrative dominance and reliance on economic growth during the early stages of development. However, over time, the explanatory power of the level of technological investment and urbanization rate continued to rise, gradually assuming dominant roles. Climate and topographic factors evolved from having minimal influence in the early stages to becoming important driving forces, whereas the driving effects of government scale level and per capita GDP declined. By 2023, the ranking of explanatory power for each factor had shifted to: X9 (0.589) > X7 (0.583) > X1 (0.380) > X4 (0.321) > X5 (0.165) > X2 (0.152) > X3 (0.137) > X6 (0.083) > X8 (0.013). The increasing explanatory power of natural factors (especially topographic factors) over time did not imply significant changes in the natural conditions themselves, but rather indicated that the coordination state of the WEFL system had become increasingly dependent on natural conditions with the advancement of socioeconomic development. Overall, the core driving forces for the coupling coordination development of the system gradually shifted from administrative regulation and economic growth to technological empowerment and urbanization, while the driving dimension transitioned from human-dominated to a synergistic human–nature dual dimension.

The interaction detection results for the various external driving factors in Shanxi Province are shown in Figure 7. The interactions between driving factors mostly exhibited a significant nonlinear enhancement or dual-factor enhancement characteristics, with nonlinear enhancement being the dominant type. This suggests that, compared to the effect of individual factors, the interactions between factors generally enhanced the explanatory power for the coupling coordination of the WEFL system. We ranked the explanatory power of the 36 cross-combinations each year and defined the top 12 as the high-value zone. The results showed that in 2000, the high-value zone was observed seven, three, and two times in the combinations of natural + anthropogenic, natural + natural and anthropogenic + anthropogenic factors, respectively; the corresponding frequencies in 2008, 2016 and 2023 were six, three, and three times; ten, one, and one times; and ten, zero and two times, respectively. The explanatory power of the natural + anthropogenic cross combinations increased obviously, which indicated that the coupling coordination of the WEFL system in Shanxi Province was primarily driven by the synergistic interaction between anthropogenic and natural factors, with this effect becoming increasingly prominent.

3.5. Prediction of the Coupling Coordination Level

The CCD prediction performance in Shanxi Province from 2019 to 2023 based on the ARIMA model is presented in

Table 7. Simulation performance of the ARIMA model for the CCD from 2019 to 2023.

Region	Absolute Relative Error (%)
	2019	2020	2021	2022	2023	Average
Datong	0.551	0.862	1.207	0.059	0.273	0.591
Shuozhou	3.053	0.769	0.318	1.509	0.201	1.170
Xinzhou	2.606	2.580	2.068	2.572	4.189	2.803
Lvliang	5.796	0.395	1.644	2.090	1.591	2.303
Taiyuan	0.787	1.019	2.427	3.523	4.329	2.417
Yangquan	0.615	1.892	0.999	0.825	4.848	1.836
Jinzhong	0.737	0.674	1.805	0.547	1.449	1.042
Linfen	0.448	1.253	4.070	3.222	4.548	2.708
Changzhi	0.146	1.377	5.018	3.385	3.873	2.760
Yuncheng	3.381	0.082	3.675	0.892	1.511	1.908
Jincheng	3.079	0.080	5.166	2.746	5.110	3.236
Shanxi	2.244	0.371	1.414	0.486	0.208	0.945

. The absolute relative error of the ARIMA model predictions across different regions and years ranged between 0.059% and 5.796%, indicating that the model performed well with high levels of accuracy and could be used to predict the future CCD in Shanxi Province (Figure 8). The coupling coordination development trend in Shanxi from 2026 to 2030 generally followed the trajectory observed from 2000 to 2023, with the CCD exhibiting an increasing trend. The average CCD for the province from 2026 to 2030 was projected to increase by 8.26% compared to the 2019−2023 period, advancing from a primary to intermediate coordination level. The CCD of each city will ultimately reach either a primary or intermediate coordination level, with Lvliang and Xinzhou exhibiting relatively faster growth rates, while Jincheng and Linfen exhibited slower growth. Overall, the synergistic effects of the Shanxi WEFL system are expected to further strengthen in the future.

4. Discussion

4.1. Comparison with the Previous Literature

This study reveals that the CCD of the WEFL nexus system in Shanxi Province exhibited an upward trend from 2000 to 2023, with the provincial CCD increasing from 0.500 in 2000 to 0.664 in 2023. This growth trend aligned with the region’s long-term policy orientation of resource conservation, ecological protection, and industrial transformation, and is also consistent with the findings of Zhang et al. [62], who investigated the CCD of the WEF nexus across major provinces in China and identified a fluctuating upward trend in Shanxi’s CCD. However, discrepancies existed between the results of this study and those of some existing studies in terms of specific values and growth margins. For instance, Zhang and Wang [31] found that the CCD of Shanxi’s WEF system increased from 0.430 to 0.607 during 2000–2020, with overall values lower than those of this study. Such differences stemmed primarily from two key extensions of our research: first, we systematically incorporated the land subsystem into the evaluation framework, which more comprehensively captured the spatial carriers and ecological constraints of the resource system, potentially raising the baseline value of the system’s CCD; second, in indicator selection, we emphasized specific indicators associated with regional resource transformation and ecological protection (e.g., energy environmental protection expenditure, soil and water loss control), which reflected the unique development path of Shanxi as a major energy base. These discrepancies precisely highlighted the importance of integrating the land element into traditional WEF nexus analysis and adopting a regionalized indicator system, rendering the evaluation results more consistent with local realities.

In terms of spatial distribution, the CCD of Shanxi’s WEFL system evolved from a pattern of “high in the north and south, low in the central region” in 2000 to “high in the north and west, low in the south and east” in 2023. This spatial characteristic underscored the heterogeneous impacts of regional development strategies and resource endowments, and complemented existing spatial analysis studies on the WEF nexus system. Most existing WEF studies at the scale of the Yellow River Basin or Chinese provinces focused on inter-provincial differences in coupling coordination levels, lacking in-depth characterization of intra-provincial heterogeneity at the prefectural level [63]. This study filled this gap by identifying typical regional coordination development models: Shuozhou (an energy-driven coordination model) and Linfen (an ecological–agricultural synergy model) had consistently maintained leading positions in CCD; meanwhile, Lvliang had achieved a substantial increase in CCD driven by ecological governance and industrial transformation, providing a new case for exploring how late-developing regions can optimize the WEFL nexus—a finding rarely documented in the existing research on energy-intensive regions. This refined analysis at the prefectural level offered empirical evidence for understanding how macro policies exert heterogeneous effects at the local level, serving as an important supplement to the existing literature that mostly took provinces as the basic research unit. It should be noted that although inter-city resource transfers (e.g., coal exports, water diversion projects, and food inflows) exerted a certain influence on the spatial pattern of coupling coordination levels, they were not directly incorporated into the assessment framework of this study due to the difficulty in obtaining relevant data.

By integrating the analysis of internal obstacle factors and external driving factors, this study further deepened research on the driving mechanisms of resource nexus systems. Most existing studies adopted obstacle degree models to identify internal obstacle factors: for example, Chang et al. [64] identified the food self-sufficiency rate, electricity generation, and ecological water use as the core internal constraints of the WEF nexus in Northeast China; Jing et al. [19] found that pollution from fertilizer and pesticide application, investment in industrial solid waste treatment, waterlogging control area, and urban water-saving measures were significant obstacle factors affecting the coupling coordination development of the WEFL nexus in the Yangtze River Economic Belt, China. This study found that the core obstacle factors of Shanxi’s WEFL system had long been concentrated in the energy and land subsystems (e.g., energy environmental protection expenditure, GDP per unit area), highlighting the strong regional specificity of obstacle factors. Additionally, this study used the Geodetector model to analyze the external driving mechanism, revealing that the external driving factors of Shanxi’s WEFL system had evolved from government scale and per capita GDP in 2000 to scientific and technological investment and urbanization rate in 2023. These findings were basically consistent with those of Huang and Han [36], who pointed out that key factors influencing the resilience of the WEF system in Shanxi Province included population density, technological innovation, and industrial structure.

The coupling coordination level of the Shanxi WEFL system during the upcoming 15th Five-Year Plan period (2026−2030) was also predicted. The ARIMA model exhibited a good simulation performance, and the forecast results show that Shanxi’s CCD will rise from 0.685 (primary coordination level) in 2026 to 0.716 (intermediate coordination level) in 2030. Zhao et al. used the ARIMA model to forecast the CCD of nine provinces in the Yellow River Basin, showing that the Shanxi WEF CCD will increase from 0.648 in 2025 to 0.674 in 2030 [65]. Zhang and Wang demonstrated that the Shanxi WEF CCD is expected to reach approximately 0.74 by 2027 [31]. The prediction results from the present study was within the range of these two studies, confirming the rationality of the projections obtained in this research.

In summary, the analysis results of the WEFL nexus system in Shanxi Province presented in this study are highly consistent with the existing research literature and the actual conditions of the study area, while also filling the research gaps in the current studies on the WEF nexus system.

4.2. Recommendations for Improving the Coupling Coordination Level of the WEFL System

In response to the core internal obstacles, it is essential to first increase investment in energy conservation and environmental protection. It is recommended to raise the proportion of energy conservation and environmental protection expenditures in the general public budget, support technological upgrades in energy-intensive industries, and promote a low-carbon transition in the energy structure. Particular emphasis should be placed on reducing the share of coal consumption in regions with higher obstacle levels. Second, efforts should be made to enhance land resource utilization efficiency by increasing investment in soil and water loss management, expanding mechanized farming, and improving land output efficiency to alleviate constraints in the land subsystem. Third, accelerating the transformation of the energy structure is critical. While ensuring energy security, renewable energy should be vigorously developed to reduce dependence on coal and mitigate the dual pressures of energy supply and the environmental footprint on the resource-linked system. Finally, strengthening the synergistic development of the water and food subsystems is necessary. This includes improving agricultural water use efficiency, enhancing farmland water infrastructure, and rationally adjusting agricultural planting structures in accordance with water resource endowments. These measures will reduce the pressure of agricultural water use on the water resource system while ensuring food security.

In response to the evolving external drivers, priority should be given to leveraging the leading role of technological innovation. Establishing a scientific and technological innovation platform for the interconnected WEFL system is advisable, along with enhancing the research, development, and application of key technologies such as water-saving agriculture, clean energy, and efficient land use. This will strengthen the technical foundation for the coordinated development of the system. Simultaneously, promoting high-quality urbanization development and optimizing the spatial layout of urbanization will develop the agglomeration effect of population and industries, and help reduce per capita resource consumption and improve resource allocation efficiency. Additionally, attention should be given to the impacts of climate change. Establishing an early warning mechanism for climate-related resource security risks will enhance the adaptability of the WEFL system to climate change.

Finally, differentiated regulatory strategies should be formulated for different regions. For northern regions represented by Shuozhou, the focus should be on maintaining the advantages of the energy and land subsystems, strengthening water resource protection, and preventing the over-exploitation of water resources from hindering sustainable development. For southern regions, such as Yuncheng and Linfen, efforts should be made to improve agricultural water use efficiency and strengthen land subsystem management while stabilizing grain production capacity. For central regions, such as Taiyuan and Yangquan, the core objectives are to enhance grain production capacity, optimize urban industrial structures, and improve the coordination between the food and land subsystems. For western regions with strong growth momentum, such as Lvliang, it will be crucial to consolidate the achievements of ecological governance and promote the coordinated development of energy exploitation and ecological protection.

4.3. Study Limitations and Future Prospects

Although this study systematically analyzed the spatiotemporal evolution and driving mechanisms of the WEFL nexus system in Shanxi Province, it had several limitations. First, despite covering the main aspects of each subsystem, the comprehensive evaluation index system could not fully characterize all dimensions of the nexus system. A systematic sensitivity analysis was not performed in this study to test the robustness of results against changes in indicator selection. Second, this study focused on the provincial and prefecture-level scales, lacking an analysis at smaller spatial scales, such as the county level, limiting our ability to refine the characteristics of the coordinated development of the WEFL system. Third, the ARIMA model adopted for CCD prediction mainly relied on historical data trends [61], without fully considering the impacts of sudden factors such as extreme climate events and major policy adjustments. The prediction results should therefore be interpreted as trend projections under a business-as-usual scenario rather than definitive predictive conclusions.

To address the above limitations, future research should focus on (1) exploring the sensitivity of model outcomes to variations in indicator selection, optimizing the indicator system, and strengthening the analysis of interaction mechanisms, (2) expanding the research scales and conducting refined studies to explore the more detailed heterogeneity and dynamic evolution laws within the resource nexus system [54], and (3) optimizing the prediction model and scenario settings. Methods such as system dynamics and topic modeling should be introduced to simulate the complex feedback relationships between systems and the implementation effects of different policy scenarios [66,67], thereby improving the scientific rigor and reliability of prediction results.

5. Conclusions

5.1. Temporal and Spatial Evolution Characteristics of the WEFL Nexus

The comprehensive evaluation index of most subsystems exhibited an upward trend from 2000 to 2023, among which the land subsystem registered the fastest growth, followed by the energy subsystem. The average value of the overall CEI of the nexus was 0.346, with Shuozhou and Linfen performing better than the other cities. The CCD of the nexus ranged from 0.458 to 0.728, with an average of 0.575, indicating that the whole system was in the transitional coordination stage while exhibiting a steady improving trend. Its spatial pattern shifted from “high in the north and south, low in the central region” to “high in the northwestern part, low in the southeastern part.” By 2023, most prefecture-level cities in the province upgraded to the primary coordination level or above.

5.2. Key Internal Constraints and External Drivers of the WEFL Nexus

The driving mechanism for the coupling coordination development of Shanxi’s WEFL nexus has undergone a notable transformation. The key internal obstacles included public budget expenditure on energy conservation and environmental protection, GDP per unit area, per capita primary energy output, soil erosion control rate, and effective irrigation rate of farmland. Furthermore, the restrictive effects of the water and grain subsystems gradually became prominent. In terms of external driving forces, the core driving factors shifted from government scale and economic growth in the early stage to technological investment and urbanization. Climate and terrain conditions also gradually evolved into important influencing factors, forming a dual-dimensional driving pattern of human–nature interaction.

5.3. Prediction of the Coupling Coordination Level of the WEFL Nexus

The prediction results based on the ARIMA model revealed an absolute relative error that ranged from 0.059% to 5.796%, indicating good prediction accuracy. From 2026 to 2030, the CCD of the Shanxi WEFL nexus maintained an upward trend, with an average increase of 8.26% compared with the mean value from 2019 to 2023, and the coordination level will upgrade from primary to intermediate coordination. Among the cities, Lvliang and Xinzhou were projected to witness relatively rapid growth, while Jincheng and Linfen will likely experience relatively slow growth.

This study clarified the operating rules and influencing mechanisms of the coupling coordination relationship of the Shanxi WEFL nexus. The results provided a scientific basis for optimizing regional resource allocation and promoting coordinated and sustainable development.