ARID LAND GEOGRAPHY
Vol. 48 No. 7 Jul. 2025

Spatial Pattern and Evolution Trend of Agricultural Grey Water Footprint Intensity in the Yellow River Basin

CHENG Peng¹, PENG Haiyang¹, HOU Dingrong¹, SUN Mingdong², SONG Xiaowei¹
¹College of Resources and Environment, Shanxi University of Finance and Economics, Taiyuan 030006, Shanxi, China
²Institute of Water Ecology and Environment, Chinese Research Academy of Environmental Sciences, Beijing 100012, China

Abstract: Agricultural water pollution management is essential for alleviating water resource crises and promoting regional sustainable development. This study employs agricultural grey water footprint intensity (AGWFI), which integrates the agricultural grey water footprint with economic development levels, as an indicator of agricultural pollution intensity. We calculated the AGWFI for 112 prefecture-level cities in the Yellow River Basin from 2012 to 2021, comprehensively analyzing its spatial patterns and evolutionary trends. Panel quantile regression was used to investigate influencing factors. The results show that: (1) AGWFI decreased significantly across the entire Yellow River Basin and its upper, middle, and lower reaches from 2012 to 2021, with the decline in the upper reaches substantially greater than in the middle and lower reaches. (2) The spatial distribution exhibited a west-high, east-low pattern. The Gini coefficient for AGWFI in the basin overall and across its sub-regions was large and showed an upward trend, with intra-regional and inter-regional differences serving as the primary sources of disparity. Transfer paths mainly occurred between adjacent levels. (3) Agricultural economic development level had a significant negative effect on AGWFI, while the share of primary industry output and agricultural water resource utilization degree had significant positive effects. These findings provide scientific references for developing targeted agricultural water pollution management strategies in the Yellow River Basin.

Keywords: agricultural grey water footprint intensity; spatial pattern; evolution trend; influencing factors; Yellow River Basin

Water resources are crucial for regional agricultural production, economic growth, and sustainable development. With rapid societal development, water crises such as resource shortages and quality deterioration have become increasingly prominent. It is projected that by 2025, over half of the global population will face water scarcity, with more than 1.2 billion people experiencing severe water stress. Additionally, more than 80% of wastewater worldwide is discharged without effective treatment, posing significant challenges to drinking water safety. Agricultural production accounts for 70% of water resource development and utilization and represents a major source of water pollution. Therefore, accurately and reasonably evaluating the negative impacts of agricultural production on water resources is vital for alleviating water shortages.

Traditional water resource and environmental assessments have focused primarily on water quality as a pollution criterion, often neglecting the impact of pollution on water quantity. To address this limitation, Hoekstra proposed the "grey water footprint" theory, which quantifies water pollution levels through the volume of freshwater required for dilution, thereby integrating both water quality and quantity considerations. This approach has been widely applied in water environmental pollution assessments. As grey water footprint theory and methods have matured, increasing numbers of studies have utilized this concept to evaluate agricultural pollution levels. Current research on agricultural grey water footprints primarily focuses on calculation methods, spatiotemporal characteristics, and influencing factor analysis at national, provincial, municipal, and basin scales. Some studies have also examined grey water footprints for specific crops such as corn and sugarcane. However, most existing research evaluates regional agricultural pollution levels based solely on total agricultural grey water footprint, which cannot accurately reflect the pressure of regional agricultural pollution emission intensity on local aquatic ecosystems.

Grey water footprint intensity, expressed as the ratio of total grey water footprint to local GDP, incorporates economic factors and provides a more comprehensive examination of regional water pollution conditions than grey water footprint alone. Agricultural grey water footprint intensity (AGWFI) can be represented as the ratio of total agricultural grey water footprint to local agricultural GDP, reflecting agricultural water pollution intensity from an economic perspective. A smaller AGWFI value indicates less water pollution per unit of agricultural economic output, representing higher efficiency of agricultural grey water footprint. Compared with agricultural grey water footprint, AGWFI can more objectively reveal regional agricultural water pollution levels, though specialized studies on AGWFI remain relatively scarce.

Existing research has investigated the spatiotemporal characteristics of agricultural grey water footprints in different regions. For example, Zhang et al. explored the spatial distribution patterns of China's agricultural grey water footprint, while Xu et al. analyzed the spatial correlation and agglomeration of provincial grey water footprint intensity in China from 1998 to 2012. These studies primarily employed spatial autocorrelation methods and parametric models such as the Gini coefficient and Theil index. However, parametric models are susceptible to interference from unknown parameters, and these methods are inadequate for revealing dynamic evolution characteristics and long-term transfer trends of spatial distributions. Kernel density estimation and Markov chain analysis can effectively overcome these limitations. Kernel density estimation uses non-parametric models to avoid parameter interference and provides intuitive, dynamic visualizations of distribution patterns, though it offers limited information on internal dynamics and cannot reveal long-term transfer trends. Markov chain analysis examines state transition probabilities over time, enabling in-depth investigation of long-term transfer trends and effectively compensating for this limitation. Currently, few studies have comprehensively applied kernel density estimation and Markov chain analysis to examine the dynamic evolution and long-term transfer trends of agricultural grey water footprints and their intensity.

The influencing factors of agricultural grey water footprints are complex and diverse. Most studies employ the Logarithmic Mean Divisia Index (LMDI) method to identify key driving factors. Chen et al. used the LMDI model to analyze factors affecting agricultural grey water footprint efficiency, finding that agricultural economic effects are the most critical drivers. However, these methods overlook the differential impacts of influencing factors across regions with varying agricultural grey water footprint levels. Panel quantile regression models can more comprehensively reflect the influence of driving factors by examining regression coefficients at different quantiles of the dependent variable, yielding more reliable conclusions. Currently, panel quantile regression is widely applied in fields such as carbon emissions, environmental taxes, and ecosystem service values, but its application in analyzing influencing factors of agricultural grey water footprints and their intensity remains limited.

The Yellow River Basin is a crucial ecological barrier and economic zone in China. In recent years, rapid industrialization and urbanization have intensified water pollution, becoming a major obstacle to ecological protection and high-quality development. The basin produces approximately 35% of the nation's grain and meat, generating enormous agricultural pollutant emissions that strongly constrain water environmental quality improvement. This study comprehensively analyzes the spatial patterns and evolution trends of AGWFI in the Yellow River Basin from 2012 to 2021 using Dagum Gini coefficient, kernel density estimation, and Markov chain analysis methods, and introduces a panel quantile regression model to explore influencing factors, aiming to provide scientific references for agricultural water pollution control in the basin.

1.1 Study Area Overview

The Yellow River originates from the Bayan Har Mountains on the Qinghai-Tibet Plateau and flows through nine provinces (regions): Qinghai, Sichuan, Gansu, Ningxia, Inner Mongolia, Shaanxi, Shanxi, Henan, and Shandong, with a total length of 5,464 km and a drainage area of 7.95×10⁵ km². Located in temperate zones with suitable climate and fertile soil, the basin is an important grain production base in China. Excessive fertilizer application has created substantial pressure on regional water resources, and improper discharge of livestock breeding pollutants has further exacerbated water environmental pressures.

Considering data availability and geographical unit continuity, this study excluded Golog Tibetan Autonomous Prefecture, Yushu Tibetan Autonomous Prefecture, and Haixi Mongolian and Tibetan Autonomous Prefecture in Qinghai Province, ultimately selecting 112 prefecture-level cities in the Yellow River Basin as study objects (Figure 1). The boundaries between upper-middle and middle-lower reaches were defined at Hekou Town in Inner Mongolia and Taohuayu in Henan Province, respectively, dividing the basin into upper, middle, and lower reaches.

1.2.1 Calculation of Agricultural Grey Water Footprint

Following relevant studies, the agricultural grey water footprint in the Yellow River Basin was divided into planting and livestock components. The planting grey water footprint refers to fertilizer and pesticide applications that enter surface water through runoff and subsurface leakage, causing water pollution. The livestock grey water footprint refers to untreated manure and urine from livestock breeding that severely pollutes water bodies. Total nitrogen and chemical oxygen demand represent the main pollutants from planting and livestock activities, respectively.

The agricultural grey water footprint is calculated as:

$$
AGWF = GWF_{pla} + GWF_{bre}
$$

where $AGWF$ is the total agricultural grey water footprint (m³), $GWF_{pla}$ is the planting grey water footprint (m³), and $GWF_{bre}$ is the livestock grey water footprint (m³).

The planting grey water footprint is calculated as:

$$
GWF_{pla} = \frac{\alpha \times Appl}{\max(TN) - nat(TN)}
$$

where $\alpha \times Appl$ is the total nitrogen pollution load entering water bodies from planting (kg), $\alpha$ is the nitrogen leaching rate (set at 7% based on literature), $Appl$ is the nitrogen content in compound fertilizer (kg), $\max(TN)$ is the maximum allowable concentration of total nitrogen in surface water environments (mg·L⁻¹), and $nat(TN)$ is the natural background concentration of total nitrogen in surface water environments (mg·L⁻¹).

The livestock grey water footprint is calculated as:

$$
GWF_{bre} = \sum_{j} \frac{L_{bre}}{\max(COD) - nat(COD)}
$$

where $L_{bre}$ is the chemical oxygen demand load entering water bodies from livestock (kg), $\max(COD)$ is the maximum allowable concentration of chemical oxygen demand in surface water environments (mg·L⁻¹), and $nat(COD)$ is the natural background concentration of chemical oxygen demand in surface water environments (mg·L⁻¹). The subscript $j$ represents cattle, sheep, pigs, and poultry; $N_j$ and $T_j$ are the quantity and breeding cycle of species $j$; $h_j$ and $u_j$ are the daily manure and urine excretion rates of species $j$ (kg); $D_j$ and $E_j$ are the pollutant contents in manure and urine of species $j$ (kg); and $\alpha_j$ and $\beta_j$ are the loss coefficients of pollutants in manure and urine of species $j$.

1.2.2 Calculation of Agricultural Grey Water Footprint Intensity

Agricultural grey water footprint intensity is defined as:

$$
AGWFI = \frac{AGWF}{GDP_{agr}}
$$

where $AGWFI$ is the agricultural grey water footprint intensity (m³·yuan⁻¹), $AGWFI$ represents the intensity of water pollution caused per unit of regional agricultural GDP, and a larger value indicates more severe pollution per unit of agricultural economic output.

1.2.3 Dagum Gini Coefficient

The Dagum Gini coefficient, proposed by Dagum, decomposes overall differences into intra-regional differences, inter-regional differences, and hypervariable density contributions. This method effectively resolves overlapping issues between sub-regions that traditional Gini coefficients cannot address.

The overall difference ($G$) can be decomposed as:

$$
G = G_w + G_{nb} + G_t
$$

where $G_w$ is the contribution of intra-regional differences, $G_{nb}$ is the contribution of inter-regional differences, and $G_t$ is the contribution of hypervariable density.

The formulas are:

$$
G_{jh} = \frac{\sum_{i=1}^{n_j} \sum_{m=1}^{n_h} |x_{ji} - x_{hm}|}{2n_j n_h (\bar{x}_j + \bar{x}_h)}
$$

where $k$ is the number of divided regions, $n$ is the total number of prefecture-level cities, $n_j$ ($n_h$) is the number of cities in region $j$ ($h$), $x_{ji}$ ($x_{hm}$) is the $AGWFI$ of city $i$ ($m$) in region $j$ ($h$) (m³·yuan⁻¹), and $\bar{x}_j$ ($\bar{x}_h$) is the average $AGWFI$ of region $j$ ($h$) (m³·yuan⁻¹).

1.2.4 Kernel Density Estimation

Kernel density estimation is a non-parametric method for analyzing spatial distribution non-equilibrium. It reveals distribution characteristics through density curve position, shape, and other information.

The kernel density estimation function is:

$$
f(x) = \frac{1}{Mh} \sum_{i=1}^{M} K\left(\frac{x_i - x}{h}\right)
$$

where $f(x)$ is the probability density function, $x_i$ is the $AGWFI$ of city $i$ (m³·yuan⁻¹), $h$ is the bandwidth, $M$ is the number of cities, and $K$ is the Gaussian kernel function.

1.2.5 Markov Chain Analysis

Traditional Markov chain analysis is a discrete-time stochastic process. This study constructs a Markov transition probability matrix to explore the evolution characteristics of $AGWFI$:

$$
Q = \begin{pmatrix}
q_{11} & q_{12} & \cdots & q_{1n} \
q_{21} & q_{22} & \cdots & q_{2n} \
\vdots & \vdots & \ddots & \vdots \
q_{n1} & q_{n2} & \cdots & q_{nn}
\end{pmatrix}
$$

where $Q$ is the Markov transition probability matrix and $q_{ij}$ is the probability of a city transitioning from type $i$ in year $t$ to type $j$ in year $t+1$.

1.2.6 Panel Quantile Regression

Panel quantile regression effectively excludes interference from extreme values and does not require normally distributed data. It provides more comprehensive and reliable results by examining regression coefficients at different quantiles of the dependent variable.

The model is specified as:

$$
Q_{Y_{it}}(\tau_k | X_{it}) = \alpha_i + \beta(\tau_k) X_{it}
$$

where $Y_{it}$ is the $AGWFI$ of city $i$ in period $t$, $Q_{Y_{it}}$ is the conditional quantile function, $X_{it}$ is the matrix of explanatory variables, $\alpha_i$ is the city-specific constant term, and $\beta(\tau_k)$ is the coefficient at quantile $\tau_k$.

Based on relevant literature and data availability, this study selects five representative indicators as potential influencing factors of $AGWFI$ in the Yellow River Basin: agricultural economic development level, agricultural water resource utilization degree, share of primary industry output, grain production, and crop planting area (Table 1).

Table 1 [TABLE:1] Influencing factors of AGWFI

Influencing Factor Variable Definition Unit Agricultural Economic Development Level Per capita agricultural GDP yuan·person⁻¹ Agricultural Water Resource Utilization Degree Ratio of agricultural water use to total water resources % Share of Primary Industry Output Ratio of primary industry output to GDP % Crop Planting Area Total crop planting area 10³ hm² Grain Production Total grain output 10⁴ t

1.3 Data Sources

The study period spans 2012–2021. Agricultural water use and total water resources data were obtained from provincial Water Resources Bulletins. Data on nitrogen fertilizer application, livestock inventory, crop planting area, grain production, and other relevant indicators were sourced from the China Environmental Statistical Yearbook, China Rural Statistical Yearbook, provincial statistical yearbooks, and municipal statistical bulletins. Missing data were interpolated using linear interpolation. GDP and agricultural GDP were adjusted to constant 2012 prices. Livestock breeding cycles, daily excretion rates, pollutant contents, and loss coefficients were obtained from the China Large-scale Livestock and Poultry Breeding Pollution Survey Technical Report. Maximum allowable concentrations of total nitrogen and chemical oxygen demand in surface water were based on Class III standards of the Environmental Quality Standard for Surface Water (GB 3838-2002), with natural background concentrations generally set at zero.

2.1 Calculation of AGWFI in the Yellow River Basin

The overall $AGWFI$ in the Yellow River Basin declined rapidly from 2012 to 2015, then entered a gradual decline phase from 2016 to 2021 (Figure 2). The $AGWFI$ in the upper, middle, and lower reaches all showed downward trends similar to the basin-wide pattern. The upper reaches exhibited the largest decline at 46.70%, followed by the middle reaches at 40.79%, while the lower reaches showed a more moderate decline of 26.44%. Moreover, the $AGWFI$ in the upper reaches was significantly higher than in the middle and lower reaches, likely due to lagging economic development, insufficient agricultural infrastructure, and lack of modern agricultural technology support, resulting in higher pollutant generation per unit of agricultural economic benefit.

Figure 2 [FIGURE:2] AGWFI in the Yellow River Basin and its upstream, midstream, and downstream regions from 2012 to 2021

2.2.1 Spatial Distribution Characteristics of AGWFI

Using ArcGIS 10.8 natural breaks classification, $AGWFI$ was categorized into five levels: low (0.246–0.728 m³·yuan⁻¹), relatively low (0.729–0.994 m³·yuan⁻¹), medium (0.995–1.376 m³·yuan⁻¹), relatively high (1.377–2.982 m³·yuan⁻¹), and high (2.983–12.544 m³·yuan⁻¹). The spatial distribution showed a clear west-high, east-low pattern. Low-level $AGWFI$ areas were concentrated in eastern cities (Weihai, Yantai, Qingdao), central cities (Taiyuan, Xi'an, Yangquan), and southwestern cities (Chengdu, Neijiang, Meishan). High and relatively high $AGWFI$ areas were mainly distributed in western regions including Aba Tibetan and Qiang Autonomous Prefecture, Gannan Tibetan Autonomous Prefecture, and Garzê Tibetan Autonomous Prefecture (Figure 3).

Figure 3 [FIGURE:3] Spatial distribution of AGWFI in the Yellow River Basin

The number of cities in low, relatively low, medium, relatively high, and high $AGWFI$ levels was 48, 21, 9, 7, and 2 in 2012, accounting for 42.86%, 18.75%, 8.04%, 6.25%, and 1.79%, respectively. By 2021, low-level cities increased to 68 (60.71%), while relatively low and medium-level cities decreased to 31 (27.68%) and 8 (7.14%), respectively. The number of high and relatively high-level cities remained essentially unchanged, indicating that $AGWFI$ transfer primarily occurred from relatively low and medium levels to low levels.

2.2.2 Spatial Differences and Decomposition of AGWFI

The Dagum Gini coefficient was used to analyze spatial differentiation characteristics (Figure 4). At the basin scale, the Gini coefficient showed a fluctuating upward trend from 0.416 in 2012 to 0.466 in 2021, indicating increasing spatial disparities. Regionally, the Gini coefficient in the upper reaches (mean: 0.616) was significantly higher than in the middle (mean: 0.413) and lower reaches (mean: 0.330), primarily due to extreme values in the upper reaches that expanded internal differences.

Figure 4 [FIGURE:4] Difference of AGWFI and its contribution rates in the Yellow River Basin from 2012 to 2021

Decomposition revealed that intra-regional differences contributed most to overall disparity (mean: 45.09%), followed by inter-regional differences (mean: 36.38%), while hypervariable density contributed the least (mean: 18.53%). Intra-regional and inter-regional differences are thus the main sources of $AGWFI$ disparity in the Yellow River Basin.

2.3 Evolution Trend Characteristics of AGWFI

Kernel density estimation illustrates the dynamic evolution of $AGWFI$ (Figure 5). The density curve peaks shifted leftward overall, indicating declining $AGWFI$. The upper reaches showed a similar pattern to the basin total, while the middle and lower reaches exhibited distinct differences, with the middle reaches showing a more pronounced leftward shift, indicating a greater decline.

Figure 5 [FIGURE:5] Dynamic evolution of AGWFI in the Yellow River Basin

The distribution shape revealed that the basin-wide and upper-reach density curves showed rising peaks with stable widths, while the middle and lower reaches showed rising peaks with widening distributions, indicating increasing absolute internal differences. All regions exhibited right-tail characteristics, with the upper reaches showing the most prominent tail, indicating cities with exceptionally high $AGWFI$. The middle reaches displayed a side peak that gradually increased over time, suggesting intensifying polarization.

Markov chain analysis examined long-term transfer trends (Table 2). The diagonal elements of the transition probability matrices were all greater than off-diagonal elements, indicating strong "self-locking" and path dependence. Low-value areas showed extremely stable states with maintenance probabilities no lower than 0.90. Transitions primarily occurred between adjacent levels, with minimal probability of cross-level transitions. Only the upper reaches showed a small probability (0.03) of transitioning from relatively high to low levels, while the basin total and upper reaches showed a 0.01 probability of transitioning from relatively high to low levels.

Table 2 [TABLE:2] Markov transition probability matrix of AGWFI in the Yellow River Basin

t/(t+1) Low Relatively Low Medium Relatively High High Basin-wide Low 0.92 0.08 0.00 0.00 0.00 Relatively Low 0.08 0.84 0.08 0.00 0.00 Medium 0.00 0.09 0.82 0.09 0.00 Relatively High 0.01 0.00 0.10 0.79 0.10 High 0.00 0.00 0.00 0.13 0.87 Upper Reaches Low 0.90 0.10 0.00 0.00 0.00 Relatively Low 0.10 0.80 0.10 0.00 0.00 Medium 0.00 0.11 0.78 0.11 0.00 Relatively High 0.03 0.00 0.12 0.73 0.12 High 0.00 0.00 0.00 0.17 0.83

2.4 Influencing Factors of AGWFI

Panel quantile regression results are presented in Table 3. Agricultural economic development level showed significant negative effects on $AGWFI$ across the entire basin and all sub-regions, with greater impacts at higher quantiles. This indicates that in agriculturally underdeveloped areas with high $AGWFI$, economic development can generate economies of scale that significantly reduce $AGWFI$.

Table 3 [TABLE:3] Panel quantile regression results

Variable Basin-wide Upper Reaches Middle Reaches Lower Reaches 0.25 0.50 0.75 0.25 Agricultural Economic Development Level -0.32*** -0.28*** -0.35*** -0.41*** Share of Primary Industry Output 0.28*** 0.31*** 0.35*** 0.35*** Agricultural Water Resource Utilization Degree 0.15*** 0.12** 0.10** 0.18*** Crop Planting Area -0.08* -0.05 -0.03 -0.12** Grain Production 0.05 0.03 0.02 0.08*

Note: *, , * indicate significance at 1%, 5%, and 10% levels, respectively.

The share of primary industry output showed significant positive effects across all regions, with stronger impacts at higher quantiles. Agricultural water resource utilization degree also had significant positive effects, though its impact coefficient decreased with increasing quantiles, suggesting weaker effects in high $AGWFI$ areas. Crop planting area had significant negative effects on the basin-wide and upper-reach $AGWFI$, with greater impacts at higher quantiles. Grain production had limited effects, showing significant positive impacts only in the upper reaches and weak negative impacts in middle and lower reaches.

Ordinary least squares (OLS) regression was used to verify robustness (Table 4). Results were consistent with panel quantile regression, confirming that the findings are not dependent on estimation methodology.

Table 4 [TABLE:4] OLS test results

Variable Coefficient t-statistic Agricultural Economic Development Level -0.30*** -4.21 Share of Primary Industry Output 0.32*** 5.83 Agricultural Water Resource Utilization Degree 0.12** 2.45 Crop Planting Area -0.06* -1.78 Grain Production 0.03 0.85

3 Discussion

From 2012 to 2021, $AGWFI$ in the Yellow River Basin showed a continuous downward trend. This decline can be attributed to two main factors. First, the agricultural grey water footprint decreased substantially during this period, primarily due to implementation of policies such as the 2015 Water Pollution Prevention and Control Action Plan. Second, stable growth in the basin's agricultural economy provided funding for developing efficient and environmentally friendly agricultural technologies, creating a virtuous cycle of ecological-economic development. Recent studies have found that most regions in the Yellow River Basin have achieved strong decoupling between agricultural grey water footprint and economic growth, meaning agricultural pollution has not increased with economic growth, indirectly confirming the decline in $AGWFI$.

Based on these findings, we propose three recommendations. First, prioritize the upper reaches as the key region for $AGWFI$ reduction. The upper reaches should strengthen agricultural infrastructure, optimize industrial structure, and develop green organic agriculture with low pollution and high added value while reducing high-pollution livestock activities. Second, middle and lower reaches with lower $AGWFI$ should focus on efficient water use and allocation, promoting water-saving irrigation technologies and adjusting planting structures. Finally, enhance cooperation among upper, middle, and lower reaches, with middle and lower reaches providing financial and technical support to upper reaches to leverage neighborhood effects and achieve coordinated $AGWFI$ reduction.

4 Conclusions

This study analyzes the spatial patterns, evolution trends, and influencing factors of $AGWFI$ in the Yellow River Basin from 2012 to 2021. The main conclusions are:

1. $AGWFI$ decreased significantly across the Yellow River Basin and its upper, middle, and lower reaches from 2012 to 2021, with the upper reaches showing the largest decline. The spatial distribution exhibited a west-high, east-low pattern. The Gini coefficient was large and increasing, with intra-regional and inter-regional differences as primary sources. Transfer paths mainly occurred between adjacent levels, with minimal cross-level transitions.

2. Agricultural economic development negatively affected $AGWFI$, with stronger effects in high $AGWFI$ areas. The share of primary industry output and agricultural water resource utilization degree positively affected $AGWFI$. The share of primary industry output had greater impacts on high $AGWFI$ areas, while agricultural water resource utilization degree had stronger effects on low $AGWFI$ areas. Crop planting area and grain production had limited effects on $AGWFI$.

References

[1] Li Ruzi, Bai Yi, Zhou Yunnan, et al. Decoupling and decomposition of driving factors of water resources utilization and economic growth in the Yellow River Basin[J]. Scientia Geographica Sinica, 2023, 43(1): 110-118.

[2] Distefano T, Kelly S. Are we in deep water? Water scarcity and its limits to economic growth[J]. Ecological Economics, 2017, 142: 130-147.

[3] Jin Yinxue. Water crisis under global warming[J]. Ecological Economy, 2020, 36(6): 5-8.

[4] Hoekstra A Y, Chapagain A K, Aldaya M M, et al. The water footprint assessment manual: Setting the global standard[M]. London: Routledge, 2011.

[5] Kong Y, He W J, Zhang Z F, et al. Spatial-temporal variation and driving factors decomposition of agricultural grey water footprint in China[J]. Journal of Environmental Management, 2022, 318: 115601, doi: 10.1016/j.jenvman.2022.115601.

[6] Zeng Zhao, Liu Junguo. Historical trend of grey water footprint of Beijing, China[J]. Journal of Natural Resources, 2013, 28(7): 1169-1178.

[7] Chu Tianshu, Huang Yixuan, Lai Shixuan, et al. Analysis and evaluation of grey water footprint for planting-animal husbandry industry in provincial regions, China[J]. Chinese Journal of Agricultural Resources and Regional Planning, 2023, 44(11): 53-64.

[8] Yan Y A, Wang R, Chen S, et al. Three dimensional agricultural water scarcity assessment based on water footprint: A study from a humid agricultural area in China[J]. Science of the Total Environment, 2023, 857: 159407, doi: 10.1016/j.scitotenv.2022.159407.

[9] Wang Danyang, Li Jingbao, Ye Yaya, et al. Analysis of agricultural grey water footprint in Hunan Province based on different receiving water bodies[J]. Water Resources Protection, 2016, 32(4): 49-54.

[10] Cheng Peng, Li Xuyong, Sun Mingdong. Calculation method and illustration of seasonal agricultural grey water footprint in monsoon region based on river reference conditions[J]. Acta Scientiae Circumstantiae, 2020, 40(9): 3453-3462.

[11] Du Huijuan, Wang Guangyao, Ran Guangyan, et al. Agricultural grey water footprint in the Tarim River Basin using SDGs analysis[J]. Arid Zone Research, 2023, 40(7): 1184-1193.

[12] Zhang Xin, Li Lei, Zhen Zhilei, et al. Analysis of agricultural grey water footprint in Fenhe River Basin based on the perspective of space-time and efficiency[J]. China Environmental Science, 2019, 39(4): 1502-1510.

[13] Li H Y, Wang Y F, Qin L J, et al. Effects of different slopes and fertilizer types on the grey water footprint of maize production in the black soil region of China[J]. Journal of Cleaner Production, 2020, 246: 119077, doi: 10.1016/j.jclepro.2019.119077.

[14] Barreto M D L P, Netto A M, Da Silva J P S, et al. Gray water footprint assessment for pesticide mixtures applied to a sugarcane crop in Brazil: A comparison between two models[J]. Journal of Cleaner Production, 2020, 276: 124254, doi: 10.1016/j.jclepro.2020.124254.

[15] Xu C X, Liu Y, Fu T B. Spatial-temporal evolution and driving factors of grey water footprint efficiency in the Yangtze River Economic Belt[J]. Science of the Total Environment, 2022, 844: 156930, doi: 10.1016/j.scitotenv.2022.156930.

[16] Li Shengnan, Wang Yuan, Luo Jin, et al. Spatio-temporal variations and driving factors of grey water footprint in Fujian Province[J]. Acta Ecologica Sinica, 2020, 40(21): 7952-7965.

[17] Cheng Peng, Sun Mingdong, Song Xiaowei. Study on the spatial and temporal dynamic evolution and driving factors of grey water footprint in China[J]. Ecology and Environmental Sciences, 2024, 33(5): 745-756.

[18] Zhao Liangshi. Study on spatial convergence of grey water footprint intensity on provincial scale in China[J]. Journal of Liaoning Normal University (Natural Science Edition), 2017, 40(4): 541-547.

[19] Xu Ruifan, Zhao Minjuan, Gao Jianzhong. Measurement and spatial-temporal variation of agricultural grey water footprint in China[J]. Chinese Journal of Agricultural Resources and Regional Planning, 2023, 44(10): 52-66.

[20] Zhang L L, Zhang R, Wang Z Z, et al. Spatial heterogeneity of the impact factors on gray water footprint intensity in China[J]. Sustainability, 2020, 12(3): 865, doi: 10.3390/su12030865.

[21] He W J, Zhang K, Kong Y, et al. Reduction pathways identification of agricultural water pollution in Hubei Province, China[J]. Ecological Indicators, 2023, 153: 110464, doi: 10.1016/j.ecolind.2023.110464.

[22] Chen Yan, Tong Guoping, Wang Lei. Spatial-temporal distribution and driving models of agricultural grey water footprint efficiency in the Huai River Basin[J]. Water Resources Protection, 2020, 36(6): 60-66, 85.

[23] Han Qin, Sun Caizhi, Zou Wei. Grey water footprint efficiency measure and its driving pattern analysis on provincial scale in China from 1998 to 2012[J]. Resources Science, 2016, 38(6): 1179-1191.

[24] Chen Minghua, Liu Huajun, Sun Yanan. Research on the spatial differences and distributional dynamic evolution of financial development of five megalopolises from 2003 to 2013 in China[J]. Journal of Quantitative & Technological Economics, 2016, 33(7): 130-144.

[25] Chen Minghua, Zhang Xiaomeng, Liu Yuxin, et al. Dynamic evolution and trend prediction of green TFP growth: Empirical research based on five urban agglomerations in China[J]. Nankai Economic Studies, 2020(1): 20-44.

[26] Bai Tianjiao, Sun Caizhi. Regional inequality and factor decomposition of the per capita grey water footprint in China[J]. Acta Ecologica Sinica, 2018, 38(17): 6314-6325.

[27] Chen J, Gao Y Y, Qian H, et al. Insights into water sustainability from a grey water footprint perspective in an irrigated region of the Yellow River Basin[J]. Journal of Cleaner Production, 2021, 316: 128329, doi: 10.1016/j.jclepro.2021.128329.

[28] Feng H Y, Sun F Y, Liu Y Y, et al. Mapping multiple water pollutants across China using the grey water footprint[J]. Science of the Total Environment, 2021, 785: 147255, doi: 10.1016/j.scitotenv.2021.147255.

[29] Fu T B, Xu C X, Yang L H, et al. Measurement and driving factors of grey water footprint efficiency in Yangtze River Basin[J]. Science of the Total Environment, 2022, 802: 149587, doi: 10.1016/j.scitotenv.2021.149587.

[30] Wang Shaojian, Xie Zihan, Wang Zehong. The spatiotemporal pattern evolution and influencing factors of CO₂ emissions at the county level of China[J]. Acta Geographica Sinica, 2021, 76(12): 3103-3118.

[31] Bi Qian, Yu Lianchao. Relationship between environmental taxes and enterprise green investment behavior: A panel quantile regression approach[J]. China Population, Resources and Environment, 2016, 26(3): 76-82.

[32] Xu Chunxia. The spatial disequilibrium and dynamic evolution of financial services industry development in Shandong Province: Based on an empirical analysis of banking industry[J]. Review of Economy and Management, 2016, 32(6): 109-117.

[33] Liu Zhitao, Wang Shaojian, Fang Chuanglin. Spatiotemporal evolution and influencing mechanism of ecosystem service value in the Guangdong-Hong Kong-Macao Greater Bay Area[J]. Acta Geographica Sinica, 2021, 76(11): 2797-2813.

[34] Liu Tongbing, Fang Yu, Huang Feng. Dynamic evaluation of the matching degree and utilization condition of generalized agricultural water and arable land resources in China[J]. Transactions of the Chinese Society of Agricultural Engineering, 2023, 39(10): 56-65.

[35] Dagum C. A new approach to the decomposition of the Gini income inequality ratio[M]. Berlin: Springer, 1998.

[36] Liu Shuaibin, Yang Shan, Wang Zhao. Characteristics and formation mechanism of China's provincial urbanization spatial correlation based on population flow[J]. Acta Geographica Sinica, 2019, 74(4): 648-663.

[37] Zhang X Y, Xiao Y A, Ramsey T S, et al. Decoupling agricultural grey water footprint from economic growth in the Yellow River Basin[J]. Water, 2024, 16(8): 1129, doi: 10.3390/w16081129.

[38] Huang Wanhua, Wang Tingting, Gao Honggui. Transforming of river chief system from district unit to basin unit: A quasi-natural experiment based on the policy of river chief system in the Yellow River Basin[J]. Journal of Dongbei University of Finance and Economics, 2022(6): 74-84.

[39] Luo Jingyi, Dong Mei. Green efficiency and rebound effect of water for grain production in China: Based on the three-stage super efficiency SBM Malmquist model[J]. Arid Land Geography, 2024, 47(9): 1508-1517.

[40] Mu Mingqingzhong, et al. Dynamic evolution characteristics and driving factors of tourism ecological security in the Yellow River Basin[J]. Acta Geographica Sinica, 2022, 77(3): 714-735.

[41] Guo Wei, Hu Xueqing, Guo Xiangyang, et al. Spatial-temporal pattern evolution and driving factors of agricultural grey water footprint efficiency in Yellow River Basin[J]. Bulletin of Soil and Water Conservation, 2024, 44(2): 437-445.

[42] Liu Jiaqi, Zhou Luhong, Xi Xiaoya. Land ecological quality and its change trend prediction in the Yellow River Basin from 2000 to 2020[J]. Arid Land Geography, 2023, 46(10): 1654-1662.

[43] Xu Ruifan, Liu Wenxin. Relationship between agricultural grey water footprint and economic growth in the Yellow River Basin: Based on decoupling and rebound effects[J]. Chinese Journal of Agricultural Resources and Regional Planning, 2023, 44(11): 79-91.