Factors Influencing Water Use Efficiency in Agriculture: A Case Study of Shaanxi, China

Abstract

Enhancing agricultural water use efficiency can support Shaanxi Province’s transition to a sustainable agricultural economy. Using the data envelopment approach (DEA), we evaluated the water use efficiency in agriculture of various cities in the Shaanxi Province of China. Moran’s I and Dagum Gini coefficients were used to reveal spatial and temporal characteristics of agricultural water use efficiency. Furthermore, the geographically and temporally weighted regression (GTWR) model was used to investigate the effects of the agricultural economy, level of water-saving irrigation, cropping structure, environmental factors, and water supply structure on agricultural water use efficiency. The findings demonstrate that the average agricultural water use efficiency from 2011 to 2020 is low (0.796) in Shaanxi Province. Spatially there is considerable variation in water use efficiency between cities, with some highly efficient and some with very low efficiency. There is a strong negative spatial correlation between cities, with high efficiency cities tending to be adjacent to low efficiency cities. Moreover, the spatial differences in agricultural water use efficiency have increased over time, with the most significant increase within the northern region and between the north and central areas. Each influencing factor has a different impact from year to year and from city to city. Overall, the level of water-saving, the level of agricultural economics, the structure of the water supply, and environmental factors primarily have a negative influence. In contrast, the planting structure primarily has a favorable function. Therefore, to support the coordinated growth of agricultural water use efficiency, Shaanxi Province should improve its city-to-city cooperation and exchange of water use technology and experience, as well as develop differentiated water use measures appropriately for the development of each city according to local conditions.

1. Introduction

The water resources of the Yellow River Basin are linked to China’s ecological security since it is a key region for ecological function and a central hub for food production in that country [1]. The “Central Document No. 1” for 2022, published by China, outlines essential initiatives to promote rural revitalization comprehensively in that year. The document proposes that we further strengthen water conservation and water resource management in the Yellow River Basin agriculture and develop dry-land farming by increasing water use efficiency. To improve water conservation and water management in agriculture in the Yellow River Basin, it is critical to advance the efficiency of agricultural water use in Shaanxi Province, located in a portion of the Basin’s midsection. “We will stay true to the principle that lucid waters and lush mountains are invaluable assets, adhere to ecological priority and green development, and use Yellow River water resources as its capacity permits,” Xi Jinping emphasized while hosting a symposium on ecological protection and high-quality development of the Yellow River Basin. According to statistics from the China Statistical Yearbook, agricultural water use in Shaanxi Province increased from five billion m³ in 2011 to 5.3 billion m³ in 2020, rising at an average annual rate of 0.65%, much higher than the level in China as a whole (−0.30%). Shaanxi Province should promote ecological protection as a requirement for development and water resources as a limitation to support sustainable agricultural growth and effective utilization of water resources in light of the severe agricultural water shortage that is currently occurring.

The primary foundations of the measurement methods for agricultural water use efficiency are stochastic frontier functions (SFA) and DEA [1]. Because the DEA model does not require predefined production functions or standardized data processing, researchers primarily utilize it to assess the efficiency with which agriculture uses water. Academics primarily evaluate the effectiveness of agricultural water use on two separate scales: the macro scale and the micro scale. Researchers have developed micro-level research from the viewpoint of farmers [2]. They have considered national [3,4,5,6], provincial [7,8,9,10,11], and watershed [12,13,14] aspects at macro level. Considering the influencing elements, researchers used the Geographically and Temporally Weighted Regression Model [14], the Tobit model [15,16,17], the Spatial Durbin model [18,19], the Geographic detector [20,21], to investigate the factors that influence agricultural water use efficiency in terms of resource endowment, water resource structure, planting structure, agricultural economic level, degree of agricultural mechanization, and farmland water conservancy construction [22,23,24].

After reviewing the above literature, it is found that most recent studies on the factors that affect agricultural water use efficiency only pay attention to the macro level and ignore the micro level. To assess the efficiency of agricultural water use, the DEA approach was used on panel data collected from ten cities in Shaanxi Province from 2011 to 2020. Moran’s I and Dagum Gini coefficients were employed to demonstrate spatial correlation and inequality characteristics of agricultural water use efficiency. The GTWR model is also used to assess its affecting elements from a regional and temporal viewpoint and estimate the barriers that prevent improving agricultural water use efficiency in each city of Shaanxi Province.

2. Materials and Methods

2.1. Variable Selection and Data Sources

2.1.1. Evaluation System of Agricultural Water Use Efficiency

According to Jinping Tong [12], Junyi Fu [14], and Jianfeng Ma [19], agricultural water use efficiency can be defined as the ratio of optimal agricultural water use input to actual agricultural water use input (Equation (2)).

$$TFW{E}_{m,n}=\frac{TW{E}_{m,n}}{W{E}_{m,n}}$$

(1)

where $TFW{E}_{m,n}$ is the agricultural water use efficiency of the $m$th city in the $n$th year, $TW{E}_{m,n}$ is the optimal agricultural water input and $W{E}_{m,n}$ is the actual agricultural water input in the production process.

The procedure for evaluating the efficiency of Shaanxi Province’s agricultural water use was developed concerning prior studies and using available data. Four indicators are chosen for the inputs: agricultural water consumption, fertilizer application, the total power of agricultural machinery, and crop sown area. For the output, the total agricultural output value is chosen as the output indicator. To mitigate the impact of price factors, each city’s total agricultural output value based on 2011 is calculated based on the province’s total agricultural output value index.

Table 1. Descriptive statistics for the system that measures the water use efficiency in agriculture.

Type of Indicators	Variables	Mean	Standard Deviation
Input Indicators	Agricultural water consumption	5.01	3.92
	Crop seeding area	420.23	189.30
	Total power of agricultural machinery	230.52	131.88
	Fertilizer application amount	22.45	18.21
Output Indicators	Total agricultural output	140.69	83.31

displays the findings of descriptive statistics for input–output factors related to agricultural water use efficiency.

2.1.2. Influencing Factor Variables

Numerous intricate elements affect how effectively agricultural water is used. We examine the variables affecting agricultural water use efficiency from five angles, drawing on previous research: the level of water-saving irrigation, the agricultural economic status, the planting structure, environmental factors, and the structure of the water supply. Among them, the agricultural economic level is determined by the rural residents’ per capita disposable income, while the planting structure is determined by the ratio of wheat and vegetable planting area to crop sown area. The environmental factor is expressed in the amount of fertilizer pollution, which, drawing on Liu Tao’s study, accounts for 65% of fertilizer applications. The ratio of water-saving irrigation area to crop-sown area determines the water-saving irrigation level. The ratio of surface water supply to subsurface water supply represents the structure of the water supply. Rural dwellers’ per capita disposable income and fertilizer pollution were logarithmically handled to assure the correctness of the regression results because their units are incompatible with those of the other variables. The outcomes of descriptive statistics for the variable factors influencing agricultural water use efficiency are demonstrated in

Table 2. Descriptive statistics of influencing factors.

Factor Classification	Variable Description	Units of Measurement	Mean	Standard Deviation
Water-saving irrigation level	c1 Share of water-saving irrigation area	Percentage	20.93	12.60
Agricultural economic level	c2 Disposable income per rural resident	Yuan	9628.37	2505.11
Planting structure	c3 Share of wheat acreage	Percentage	24.39	17.78
	c4 Share of vegetable acreage	Percentage	11.68	4.41
Environmental factors	c5 Amount of fertilizer pollution	10 kt	26.18	12.56
Water supply structure	c6 Ratio of surface water to groundwater		3.58	4.80

The Shaanxi Provincial Water Resources Bulletin for 2011 to 2020 and the Shaanxi Provincial Statistical Yearbook for 2012 to 2021 provide all the data used in the article.

.

2.2. Research Methods

2.2.1. Data Envelopment Approach

The two main categories of DEA models are the CCR method, which has constant scale payoffs, and the BCC method, which has variable scale payoffs. To make the measurement results more consistent with the actual condition of Shaanxi Province’s use of agricultural water, the BCC model is chosen in this paper to estimate and explore Shaanxi Province’s efficiency in using water for agriculture. This is because the input factors and output scale in Shaanxi Province are not constant. The DEA model’s formula can be expressed as

$$\{\begin{array}{c}\mathrm{Min}\left[\theta -\epsilon \left({\displaystyle \sum _{i=1}^t}{s}_i^{+}+{\displaystyle \sum _{i=1}^m}{s}_r^{-}\right)\right]\\ {\displaystyle \sum _{j=1}^n}{x}_j{\lambda }_j+s^{-}=\theta x_0\\ {\displaystyle \sum _{j=1}^n}{y}_j{\lambda }_j-s^{+}=y_0\\ {\displaystyle \sum _j^n}{\lambda }_j=1\\ {\lambda }_j\ge 0,\left(j=1,2,\cdots ,n\right)\\ s^{+}\ge 0,s^{-}\ge 0 \end{array}$$

(2)

where $\epsilon >0$ is non-Archimedean infinitesimal; $\theta$ is the indicator used to evaluate agricultural water use efficiency, specifically, when $\theta =1$, the city’s agricultural water use efficiency is at its best; when $\theta <1$, the city’s agricultural water use scale and method both require more optimization; $t$ is the output type; $m$ is the input type; $x, y, n, s^{+}, s^{-}$ denote the slack variables of input quantity, output quantity, number of decision units, output and input, respectively.

2.2.2. Dagum Gini Coefficient

Geographically, Shaanxi Province is split into three sections: the southern region, which consists of the cities of Ankang, Shangluo, and Hanzhong; the central region, which consists of the five cities of Xi’an, Baoji, Xianyang, Weinan, and Tongchuan; and the northern region, which consists of the cities of Yan’an and Yulin. This paper analyzes spatial differences in the province’s agricultural water use efficiency using each city as the basis. This spatial disparity can be divided into intra-regional disparities in the southern, central, and northern parts, disparities between the southern and central, central and northern, and the southern and northern areas, as well as the effect of spatial overlap between regions.

2.2.3. Spatial Autocorrelation Analysis

This research utilized Moran’s I to show the spatial relationship between agricultural water use efficiency. The Moran’s I calculation formula is denoted by

$$I=\frac{{{\displaystyle \sum }}_{i=1}^n{{\displaystyle \sum }}_{j=1}^n{w}_{ij}\left(x_i-\stackrel{\leftharpoonup }{x}\right)\left(x_j-\stackrel{\leftharpoonup }{x}\right)}{s^2{{\displaystyle \sum }}_{i=1}^n{{\displaystyle \sum }}_{j=1}^n{w}_{ij}}$$

(3)

$$I_i=\frac{\left(x_i-\stackrel{\leftharpoonup }{x}\right)}{S^2}{\displaystyle \sum }_{j=1}^n{w}_{ij}\left(x_j-\stackrel{\leftharpoonup }{x}\right)$$

(4)

where the global Moran’s I and the local Moran’s I are denoted by $I$ rand $I_i$, respectively;$S^2$ displays the sample variance; $n$ shows how many cities there are in Shaanxi Province; the agricultural water use efficiency of cities $i$ and $j$ is represented by $x_i$ and $x_j$, respectively; $w_{ij}$ represents the spatial weight matrix, this paper is based on the Queen adjacency matrix for analysis; $\stackrel{\leftharpoonup }{x}$ represents the sample mean value.

2.2.4. Geographically and Temporally Weighted Regression Model

The GTWR model, which Huang [25] proposed, involves the independent variables’ regression parameters varying over time and space. The GTWR method is used in this study to investigate the variables that affect each city’s agricultural water use efficiency and to investigate each variable’s spatial and temporal characteristics; it has the following mathematical formula:

$$y_i={\beta }_0\left(u_i,v_i,t_i\right)+{\displaystyle \sum }_k^p{\beta }_k\left(u_i,v_i,t_i\right)X_{ik}+{\epsilon }_i$$

(5)

where $\left(u_i,v_i,t_i\right)$ is the city’s spatial and temporal coordinates, $y_i$ is the efficiency of the city’s $i$ agricultural water use, and $X_{ik}$ is the influencing factor variable $k$ of city $i$. ${\epsilon }_i$ is the error term. The error term is assumed to be normally distributed with zero mean value [26,27,28] and constant variance [29,30,31].

3. Results and Discussion

3.1. Estimation of Water Use Efficiency

For each city in Shaanxi from 2011 to 2020,

Table 3. Shaanxi Province agricultural water use efficiency between 2011 and 2020.

City	2011	2012	2013	2014	2015	2016	2017	2018	2019	2020	Mean
Shangluo City	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000
Ankang City	0.736	0.639	0.637	0.628	0.615	0.628	0.617	0.603	0.599	0.595	0.629
Hanzhong City	1.000	1.000	1.000	0.960	0.963	0.992	0.977	0.918	0.904	0.903	0.962
Baoji City	0.609	0.642	0.652	0.616	0.620	0.605	0.586	0.555	0.555	0.559	0.600
Tongchuan City	0.793	0.758	0.768	0.776	0.771	0.844	0.865	0.888	0.828	0.827	0.812
Weinan City	0.507	0.545	0.557	0.478	0.484	0.486	0.428	0.397	0.522	0.533	0.494
Xi’an City	0.788	0.856	0.900	0.882	0.896	0.905	0.836	0.839	0.864	0.854	0.862
Xianyang City	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000
Yan’an City	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000
Yulin City	0.677	0.632	0.664	0.661	0.652	0.649	0.606	0.526	0.475	0.469	0.601
Mean	0.811	0.807	0.817	0.800	0.800	0.811	0.792	0.773	0.775	0.774	0.796

lists the agricultural water use efficiency figures. With an average value of just 0.769, Shaanxi Province’s overall agricultural water use efficiency from 2011 to 2020 is low, suggesting that under the current output conditions, the province needs to invest in more water resources and other production factors. Additionally, there is an absence of coordination between the growth of the agricultural industry and the preservation of water resources. Only Shangluo, Xianyang, and Yan’an have an average value of one, indicating that the ratio of inputs to outputs is optimal for these three cities. The remaining seven cities, Hanzhong being the highest with a mean value of 0.962 and Weinan being the lowest with a mean value of 0.494, all have agricultural water use efficiencies that are less than one, indicating that the ratio of input to output in these cities is irrational and requires more water input for a given output. Shangluo, Xianyang, and Yan’an, which have the highest efficiency values, are 0.506 higher than Weinan, which has the lowest efficiency. The gap between cities is large, which may be due to the lack of knowledge exchange of water use technology and experience between cities, which may cause uneven development of agricultural water use efficiency between cities.

Shaanxi Province’s agricultural efficiency in using water fluctuated, falling by an average of 0.51% annually. In particular, the agricultural water use efficiency remained above 0.800 from 2011 to 2016, and then it sharply reduced after 2017. This decline in efficiency may be attributed to the aggressive promotion of agricultural supply-side reform in Shaanxi Province in 2017, which emphasizes improving the quality of agricultural products rather than encouraging water conservation. Due to variations in each city’s capacity for allocating resources and level of control of agricultural production factors, cities in Shaanxi Province did not change their agricultural water use efficiency in the same way: agricultural water use efficiency grew in three cities—Tongchuan, Weinan, and Xi’an—and fell in the remaining four cities—Ankang, Hanzhong, Baoji, and Yulin.

3.2. Spatial and Temporal Characteristics of Agricultural Water Use Efficiency

3.2.1. Correlation Characteristics

The global Moran’s I value for Shaanxi Province’s agricultural water use efficiency from 2011 to 2020 is shown in

Table 4. Shaanxi Province agricultural water use efficiency (global Moran’s I).

Year	2011	2012	2013	2014	2015	2016	2017	2018	2019	2020
Moran’s I	−0.535	−0.585	−0.581	−0.581	−0.567	−0.565	−0.552	−0.535	−0.542	−0.540
Z-value	−2.055	−2.234	−2.213	−2.213	−2.179	−2.171	−2.126	−2.032	−2.036	−2.031
p-value	0.020	0.013	0.013	0.013	0.015	0.015	0.017	0.021	0.021	0.021

, along with their significance levels. The global Moran’s I value for Shaanxi Province’s agricultural water use efficiency from 2011 to 2020 was negative and passed the significance test at the 5% level, demonstrating a substantial inverse relationship between agricultural efficiency of water use in nearby cities. According to the changing trend, the absolute value of the global Moran’s I rose from 0.535 to 0.585 between 2011 and 2012 before falling to 0.540 in 2020, showing that the spatial aggregation of agricultural water use efficiency in Shaanxi Province initially climbed before declining.

Moran scatter diagrams in 2011 and 2020 were created with the aid of the Geode platform to visually depict the spatial agglomeration status of Shaanxi Province’s agricultural water use efficiency (Figure 1). In the Moran scatter diagram, the first and third quadrants, which correspond to high-efficiency and high-efficiency cities as neighbors and low-efficiency and low-efficiency cities as neighbors, respectively, show a positive correlation between the two dimensions of agricultural water use efficiency. The second and fourth quadrants, which show low and high and high and low-efficiency cities as neighbors, respectively, suggest a negative correlation between agricultural water use efficiency. According to Figure 1, there were six cities in quadrant two in 2011 (Xi’an, Ankang, Baoji, Tongchuan, Weinan, and Yulin) and four cities in quadrant four (Hanzhong, Shangluo, Xianyang, and Ankang); Tongchuan and Xi’an will switch to quadrants one and four respectively in 2020.

3.2.2. Inequality Characteristics

With the help of Stata 15.0 software, the overall Gini coefficient, inter-regional Gini coefficient, intra-regional Gini coefficient, and contribution rate of agricultural water use efficiency in Shaanxi Province from 2011 to 2020 were measured, aiming to reveal the magnitude and sources of spatial disparities in agricultural efficiency of water use in Shaanxi Province, whose outcomes are displayed in

Table 5. Sources and contribution of regional disparities in agricultural water use efficiency in Shaanxi Province.

Year	Total Gini	Within			Between			Contribution Rate/%
		South	Central	North	South-Central	North-South	North-Central	Between	Overlap	Within
2011	0.118	0.064	0.127	0.096	0.122	0.082	0.126	41.947	24.765	33.288
2012	0.120	0.091	0.118	0.113	0.121	0.103	0.122	28.317	37.085	34.598
2013	0.115	0.092	0.117	0.101	0.119	0.099	0.117	25.531	39.073	35.395
2014	0.128	0.096	0.140	0.102	0.134	0.102	0.135	26.129	37.74	36.130
2015	0.129	0.100	0.139	0.105	0.135	0.106	0.134	24.230	39.507	36.263
2016	0.128	0.095	0.138	0.106	0.136	0.103	0.135	23.576	40.151	36.273
2017	0.141	0.098	0.153	0.123	0.148	0.112	0.151	25.124	38.867	36.009
2018	0.155	0.105	0.167	0.155	0.155	0.133	0.170	19.258	44.417	36.326
2019	0.145	0.107	0.134	0.178	0.130	0.146	0.155	17.356	48.351	34.294
2020	0.145	0.108	0.130	0.181	0.127	0.148	0.154	17.503	48.590	33.908
Mean	0.132	0.096	0.136	0.126	0.133	0.113	0.140	24.897	39.855	35.248

.

The overall spatial disparity in agricultural water use efficiency in Shaanxi Province tends to rise, having a 2.32% average yearly growth rate, which shows that the uneven and inadequate spatial development of agricultural water use efficiency in Shaanxi Province has not been addressed in recent years. The central area of Shaanxi Province had the largest spatial discrepancies in agricultural water use efficiency before 2018. After 2018, the northern region replaced the central region as the region with the most apparent development imbalance, while the spatial disparities between the northern and central parts of Shaanxi Province were usually more significant than those in the southern region.

Regarding the spatial disparities between regions, before 2018, the spatial disparities between central and northern Shaanxi Province and central and southern Shaanxi Province alternately became the largest, while the disparities between northern and southern Shaanxi Province were the smallest. After 2018, the differences between northern and central Shaanxi Province were the smallest, while the differences between central and southern Shaanxi Province were the largest. The spatial disparities between the north and south exhibit an upward fluctuation in terms of evolutionary tendencies, but the spatial discrepancies between the south and center of Shaanxi Province exhibit a rising trend followed by a decreasing trend. The spatial inequalities between the northern and southern sections of Shaanxi Province have grown more pronounced than those between the central and southern regions, the north and the center. In summary, Shaanxi Province’s unequal agricultural water use efficiency development is getting worse. There should be more focus on how this issue is growing in tandem with other regions in the province’s northern region.

With a mean value of 39.85%, the spatial overlap effect between areas accounts for the biggest percentage of the source of spatial variances in agricultural water use efficiency. The contribution rate of the spatial overlap effect displayed an upward trend, the contribution rate of the net difference between regions displayed an opposite trend to that of the contribution rate of the spatial overlap effect between areas, and the contribution of spatial disparities within regions tended to climb and then diminish. Net interregional differences and intraregional disparities had average contribution rates of 24.89% and 35.24%, respectively. Spatial disparities between areas [32], which include two components—the contribution of inter-regional spatial overlap effects and net inter-regional disparities—are the primary cause of the widening spatial disparities in Shaanxi Province’s agricultural water use efficiency.

3.3. Influencing Factors of Agricultural Water Use Efficiency

To further investigate the spatial and temporal heterogeneity of each influential factor since agricultural water use efficiency in Shaanxi Province appears to have negative spatial correlation and inequality characteristics, the GTWR method was employed in the article to estimate the regression coefficients of the variables affecting agricultural water use efficiency in different years with each city in Shaanxi Province.

3.3.1. Time Series Variation of Influencing Factors

ArcGIS 10.8 was utilized to calculate the regression parameters of each urban efficiency of agricultural water use influencing factors at various stages and to create a box plot of their changes over time. The ArcGIS has also been used in various studies for generating study area plots [33,34,35]. The regression coefficients of each influential factor were examined to determine their temporal evolution trend (Figure 2).

(1) Water-saving irrigation level: the share of water-saving irrigation area (c1) in Shaanxi Province contributes primarily negatively to agricultural water use efficiency. Its regression coefficient exhibits a trend of initially growing and then falling. Early on, environmental preservation was not well understood, water-saving equipment was archaic, and financial investment was insufficient. Farmers would use more chemical fertilizers and pesticides to assure consistent production as the water-saving irrigation area grew, resulting in water contamination. Later, as the economy developed and agricultural water conservation and production technologies advanced, crop yields were guaranteed while using less water. People were no longer required to increase fertilizer and pesticide inputs to maintain stable production. In 2017, Shaanxi Province prioritized enhancing agricultural goods’ quality while neglecting the management of agricultural water resources, which led to a steady gap between agricultural water conservation efforts and the production level. As a result, the province’s agricultural water consumption increased by a total of 453 million m3 from 2017 to 2020, which laterally conveys that the general level of water-saving irrigation has decreased.

(2) Agricultural economic level: the per capita discretionary income (c2) of rural dwellers in Shaanxi Province has a detrimental impact on agricultural water use efficiency. Its adverse impact has been worse in recent years; this tendency suggests that despite the rise in living standards and the quickening of technology upgrading, farmers and agricultural firms still do not have a greater understanding of water conservation, and there is insufficient sustainable growth in the agricultural sector.

(3) Planting structure: while the proportion of vegetable planted area (c4) primarily has beneficial effects on agricultural water use efficiency, the negative effect has gradually increased in recent years, the proportion of wheat planted area (c3) has both benefits and disadvantages on agricultural water use efficiency, and its degree of action is relatively stable. People started following a light, nutritious diet as residents’ consumption levels rose, and as a result, the market value of fruits and vegetables continued to rise. Farmers are more likely to plant vegetables that have significant financial yields. Although wheat requires a lot of water to grow, the area of wheat production in Shaanxi Province was reduced by 141.81 thousand hectares from 2011 to 2020 due to farmers’ lack of enthusiasm for growing the crop, and the water intensity per unit area of wheat will gradually decrease. This adjustment will lessen the negative effects of the share of wheat acreage. However, the area dedicated to vegetable agriculture increased by 64.19 thousand hectares between 2011 and 2020, resulting in a continuing rise in the water intensity per unit area of vegetables and a slow onset of the inhibitory impact.

(4) Environmental factors: in the province of Shaanxi, the amount of chemical fertilizer pollution (c5) mostly has a detrimental impact on agricultural water use efficiency, and its regression coefficient exhibits an upward trend. The province has expanded its efforts since the 18th Communist Party of China Congress to develop an ecological civilization and gradually lessen its reliance on chemical fertilizers. Shaanxi Province saw a 370,000-ton decrease in chemical fertilizer use between 2012 and 2020, and the harmful consequences of chemical fertilizer pollution steadily subsided.

(5) Water supply structure: Shaanxi Province’s agricultural water use efficiency is impacted chiefly negatively by the ratio of surface water supply to underground water supply (c6), and the size of this regression coefficient shifts from an increasing trend to a steady one. Rural irrigation in central and northern Shaanxi Province traditionally relied on groundwater supply due to a lack of surface water. The shortage of surface water resources has been reduced, people have begun to replace groundwater with surface water, and the issues of groundwater overdraft and pollution have improved thanks to ongoing improvements in policies and regulations, and the gradual promotion of water conservation projects like the Yellow Diversion Project.

Having analyzed the temporal trends in the magnitude of the effects of the influencing factors, it is necessary to analyze the spatial variation in the effects of the influencing factors on agricultural water use efficiency in Shaanxi Province. In Figure 2, the length of the box represents the degree of concentration of the regression coefficients of the influencing factors, which can measure the spatial difference of the effect of the influencing factors on agricultural water use efficiency. The longer the box is, the more discrete the regression coefficients are, and the greater the spatial difference of the effect of the influencing factors on agricultural water use efficiency. The horizontal line in the box represents the median of the regression coefficients. The black squares above or below the box represent the outliers of the regression coefficients, i.e., the values that are significantly different from the other regression coefficients. As can be seen from Figure 2, from 2011 to 2020, only the length of the box for the ratio of surface water to groundwater becomes shorter, the gap between the anomalous values and the other regression coefficients narrows, and the spatial difference in the effect of the ratio of surface water to groundwater on agricultural water use efficiency narrows. On the contrary, the regression coefficients of other influencing factors have the problem of longer box lengths or an increase in the gap between the anomalies and other regression coefficients, indicating that the spatial differences in the effects of other influencing factors on agricultural water use efficiency have widened. This indicates the unbalanced development of water-saving irrigation level, agricultural economy level, environmental protection level and planting structure optimization ability in Shaanxi Province. The reason for this problem may be that Shaanxi Province has a large north-south span, and there are differences in the economic base, resource possession, governance ability and customs of each region, resulting in each city being at a different stage of development.

3.3.2. Spatial Distribution of Influencing Factors

In the above paper, through the analysis of the time trends of the regression coefficients of the influencing factors, it was found that there were obvious spatial differences in the influence of the influencing factors on agricultural water use efficiency. Therefore, this section examines the spatial distribution characteristics of the influencing factors of agricultural water use efficiency in Shaanxi Province by selecting the regression coefficients of each influencing factor in each city in 2020 (

Table 6. Spatial distribution of GTWR regression coefficients in 2020.

City	c1	c2	c3	c4	c5	c6
Shangluo City	−0.007	−0.178	0.501	1.695	−0.198	−0.017
Ankang City	1.315	0.032	1.229	−2.415	−0.166	−0.010
Hanzhong City	−2.945	0.062	−0.913	0.777	0.036	−0.008
Baoji City	−1.821	0.102	−0.866	5.529	0.152	−0.039
Tongchuan City	−2.194	−0.037	−0.164	1.071	0.303	0.020
Weinan City	0.219	−0.135	−2.015	2.450	−0.238	−0.031
Xi’an City	−0.081	−0.127	0.407	−1.202	0.455	−0.029
Xianyang City	−0.083	−0.214	0.510	−1.656	0.571	−0.030
Yan’an City	−0.611	0.066	1.939	4.423	−0.666	−0.080
Yulin City	−2.464	−0.486	32.969	10.909	−0.557	−0.016

).

(1) Water-saving irrigation level: the share of water-saving irrigation area (c1) only shows a beneficial impact in Ankang and Weinan and a suppressive effect in the other eight cities. This finding suggests that most cities may have poor planning, outdated technology, and poor management in water-saving irrigation, which prevents the current level of conservation from achieving a great match with the scale of agricultural production.

(2) Agricultural economic level: the positive areas of rural per capita disposable income (c2) are distributed across four cities: Ankang, Hanzhong, Baoji, and Yan’an, indicating that these cities can invest more surplus funds into water control and conservation work. The hostile areas are distributed across the remaining six cities: Shangluo, Tongchuan, Weinan, Xi’an, and Yulin; this shows that the agricultural economic growth of these cities is not sustainable and that more money needs to be invested in water conservation.

(3) Planting structure: Baoji, Tongchuan, and Weinan are some of the cities in the Shaanxi Province’s central region where the share of wheat cultivation area (c3) exhibits a suppressive effect. This is likely because the central region is the province’s primary wheat-producing area, and the large wheat cultivation area results in a higher water consumption intensity. Due to the monoculture structure and increased intensity of agricultural water use in Ankang, Xi’an, and Xianyang, the proportion of vegetable acreage (c4) has a detrimental impact on these cities.

(4) Environmental factors: in five cities, including Baoji, Hanzhong, Tongchuan, Xi’an, and Xianyang, the level of fertilizer contamination (c5) has aided making agricultural water use more effective. The cause of this is that these cities have boosted the effectiveness of agricultural water use by reducing the pollution caused by fertilizers and improving environmental management. Contrarily, the amount of fertilizer pollution impeded the growth of the five cities of Ankang, Shangluo, Yulin, Yan’an, and Weinan. This may be because farmers who use less fertilizer experience lower crop yields and higher environmental management expenses.

(5) Water supply structure: the current issue of groundwater over-exploitation in central and northern Shaanxi Province still needs to be further addressed because the ratio of surface water supply to groundwater supply (c6) in most cities in those regions is less favorable than in the south for the efficiency of agricultural water use. By boosting groundwater management and improving the development of water supply projects, the limiting effect of the water supply structure can be reduced.

4. Conclusions and Policy Implications

The article calculates the agricultural water use efficiency for Shaanxi Province using the DEA method. Then, using Moran’s I and Dagum’s Gini coefficients, it describes the correlation characteristics and inequality features of the efficiency of agricultural water use. Finally, it reveals the heterogeneity of the factors, such as the water supply structure and the planting structure, from two aspects of time and space using the GTWR method. The article then comes to the following conclusions.

(1) With an average score of just 0.796 from 2011 to 2020, Shaanxi Province’s agricultural water use efficiency is low. Yan’an, Shangluo, and Xianyang, which have the highest efficiency values, are 0.506 higher than Weinan, which has the lowest efficiency value. This demonstrates the apparent disparity in agricultural water use efficiency amongst the cities.

(2) In Shaanxi Province, there is a negative spatial correlation between the agricultural water use efficiency of different cities; adjacent cities display low and high efficiency on Moran’s I scatter plot. Shaanxi Province exhibits growing geographical inequalities, with the most noticeable widening occurring in the northern region and between it and the central region.

(3) The regression coefficients of each contributing factor’s change patterns were categorized into four scenarios based on a temporal perspective: regression coefficients for several variables shift in a steady direction, including the regression coefficient for the percentage of wheat cultivation area, which went from increasing to stable as well as the ratio of surface water supply to the underground water supply. The regression coefficient for the level of fertilizer pollution is rising. The water-saving irrigation degree and per capita disposable income of rural people showed an initial increase and then a subsequent decline in their relationship coefficients. The fraction of the land devoted to vegetable cultivation shows a decreasing and subsequently increasing regression coefficient.

(4) In terms of spatial disparity, only Ankang’s level of water-saving irrigation contributes positively to agricultural water use efficiency, and Shangluo, Ankang, Xi’an, and Yan’an all see positive effects on rural households’ per capita discretionary income. On the other hand, most of Shaanxi Province contributes negatively to the proportion of wheat cultivation. In Ankang, Xi’an, and Xianyang, the proportion of vegetable cultivation is detrimental; in Shangluo, Ankang, Weinan, Yan’an, and Yulin, fertilizer pollution is detrimental; and in Shaanxi Province’s central and northern parts, the surface water supply to groundwater supply ratio is unfavorable.

In response to these findings, the following recommendations are made.

(1) Optimize the input ratio of agricultural water resources and other production elements. Ankang, Hanzhong, Baoji, Tongchuan, Weinan, and the other seven cities where agricultural efficiency in water use has not been effectively realized should, by their actual situation, reasonably control all input factors within a certain range and improve their ability to allocate production factors.

(2) We encourage the rational flow of factors between the northern and central regions and the northern and southern regions of Shaanxi Province and reduce the spatial disparity by enhancing the exchange of experience between neighboring cities and areas in the utilization of agricultural water resources. Coordinating experience-sharing in water efficiency and conservation methods between Yan’an and Yulin in the northern region needs special attention. Cities with high agricultural water use efficiency, like Xianyang, Yan’an, and Shangluo, work together to aid cities with low agricultural water use efficiencies, such as Ankang, Baoji, Weinan, and Yulin.

(3) Shaanxi Province as a whole should place a high priority on the development, study, and use of new crop varieties and water-saving irrigation technologies; promote agricultural water conservation by doing more than just reducing the area planted to high water-use crops, like wheat and rice, and commit to educating farmers and agribusinesses about water conservation and providing them with incentives and penalties to motivate them to take the necessary steps. We must specify the maximum amount of groundwater that can be extracted from each city, enhance the design of water supply projects, promote the use of surface water and unconventional water, and do an excellent job of preventing and controlling water pollution. We must also increase research and development of low-pollution fertilizers to protect the water environment. We must also strictly set the quality standards for fertilizer use to guide farmers to use fertilizers scientifically and sensibly.

(4) Since the spatial and temporal heterogeneity in the specific effects of each influencing factor on each city, differentiated measures ought to be put into place by each city’s current situation regarding its agricultural economic level, planting structure, water supply structure, environmental factors, and level of water-saving irrigation: Using Weinan City as an example, four factors—the proportion of wheat cultivation area, rural per capita disposable income, the amount of fertilizer pollution, and the ratio of surface water supply to groundwater supply—impacted the city’s agricultural efficiency of water use negatively, among them, the share of wheat acreage’s suppressive effect was the greatest. As a result, Weinan should actively promote new crop varieties, such as water-saving wheat varieties.