Spatial-Temporal Evolution and Driving Mechanism of Urban Land Use Efficiency Based on T-DEA Model: A Case Study of Anhui Province, China

Abstract

As China’s urbanization has shifted from high-speed to high-quality development, Urban Land Use Efficiency (ULUE) has become an important scale for evaluating urban connotative development. However, existing research has paid less attention to errors caused by different urban environmental factors and random disturbances in ULUE. Therefore, the purpose of this study is to eliminate the impact of environmental factors and random disturbances on ULUE measurement results by placing different cities under the same environmental conditions. First, a three-stage DEA envelopment analysis (T-DEA) model is introduced to calculate the ULUE of 16 prefecture-level cities in Anhui Province from 2001 to 2020. On this basis, the kernel density estimation model, gravity center model, and geographic detector models are used to study the spatial and temporal evolution and driving factors. The results show that (1) the ULUE increases nonlinearly with time, with an increase of 12.74%; (2) the overall peak of ULUE is on the rise, and changes from a single peak to a multi-peak, indicating that ULUE is constantly improving and that there is multi-level differentiation between different cities; (3) during the study period, the center of gravity of the ULUE value moved 22.66 km to the northwest; the overall moving distance was small, and the moving rate was slow; and (4) the influence of the interaction of double driving factors on ULUE is significantly greater than that of a single driving factor, and the factors of urban built-up area and degree of openness, as the key driving factors affecting ULUE, also have a degree of duality. In addition, to achieve efficient urban land use and to coordinate the environmental differences faced by different cities, the government must formulate systematic policies and development strategies considering the spatial characteristics of urban land use efficiency and the complexity of the driving factors.

1. Introduction

As a material carrier of urban economic and social activities, urban land is also a basic guarantee of the healthy and stable development of a city [1]. In the past, during the process of rapid urbanization in China, a large agricultural population moved to non-agricultural areas, causing cities to expand rapidly over a short time [2,3]. In 1978, China’s urbanization rate was only 17.90%, growing to 63.89% by 2020 representing an annual growth rate of 1.095%, which is far higher than the global average [4,5]. However, this development comes at the expense of urban land resources and the ecological environment, resulting in large amounts of inefficient urban land. As China’s urbanization has shifted from high-speed to high-quality development, these inefficiencies have seriously restricted the efficient and sustainable development of cities [6,7]. The third United Nations Habitat Conference in 2016 emphasized that urban land-use efficiency is a necessary condition for promoting urban economic development and called for strengthening urban land planning, management, and supervision to accelerate the improvement of urban land-use efficiency. The Chinese government is also highly concerned about improving the efficiency of urban land use. In 2019, the Ministry of Natural Resources promulgated the “Outline of Territorial Spatial Planning” which was proposed to clarify the goals of achieving urban-rural integration, optimizing the spatial layout, and strengthening land conservation and intensive use. The plan is proposed to improve the efficiency of urban land utilization through land conservation and intensive use, urban renewal, and the promotion of urban–rural integration. In 2021, the “Regulations on the Administration of Urban Land Use,” as adopted by the State Council, emphasizes the strengthening of urban land conservation and intensive use, standardizing urban land use behavior, and promoting the efficient use of urban land. Therefore, obtaining the maximum utilization benefit with the minimum input of urban land elements while realizing connotative improvements has become a major problem and challenge for the sustainable development of land space in the new era.

Scholars have conducted extensive research on urban land-use efficiency. First of all, from the perspective of an in-depth understanding and definition of the concept of urban land use efficiency as a concept involving many aspects, scholars have offered different interpretations of the idea from different perspectives. Some scholars believe that urban land use efficiency is composed of two interrelated levels: The macro level (structural efficiency of land allocation) and the micro level (marginal efficiency of land use) and that the unit urban land output is a measure of urban land use efficiency [8,9]. However, most scholars believe that urban land-use efficiency is the process of putting the smallest urban built-up area, population, and production capital into urban land under natural and social urban environments, thereby generating the greatest economic, social, and ecological benefits through a series of complex economic and social activities [10,11,12,13]. Secondly, the methods for measurement of urban land-use efficiency mainly include data envelopment analysis (DEA) [14,15], geographic information systems (GIS) [16], regression analysis [17], and other mixed methods [18,19,20]. Among these, the most commonly used is data envelopment analysis, along with measurement models derived from the traditional DEA model, such as SBM-undesirable [21] and super-DEA models [22]. However, in practical applications, the traditional DEA method cannot solve the problem of negative output values and cannot be applied in the presence of uncontrollable factors. Fried (2002) proposed a three-stage DEA method, which introduces random error and environmental variable interference into the traditional DEA method, uses non-parametric methods to determine the weight of each DMU, and then uses these weights to remove the influence of uncontrollable parts. Finally, the traditional DEA method is used to calculate the efficiency value of each DMU [23]. Compared with the traditional DEA method, the three-stage DEA method can better reflect reality and has a more accurate evaluative efficiency. As a complex social giant system, the external environment and random disturbances of different cities cause errors in ULUE measurements. Therefore, this study uses a three-stage DEA model to measure urban land-use efficiency in Anhui Province. From the perspective of research area and scale, the research involves different regions, such as national scope [24], river basins [25], urban agglomerations [26], and provinces [27]; however, the current research primarily focuses on the urban scale and evaluates the overall efficiency of the city by analyzing the land-use efficiency of different regions found therein. To study the factors influencing urban land use efficiency, it is necessary to acknowledge that the factors influencing urban land use efficiency are multifaceted. Scholars do not have the same characteristics for different subjects and therefore the influencing factors also show different characteristics [28,29,30,31]. However, there are few studies on how to judge whether urban land use efficiency is the result of a single driving factor or multiple driving factors, which also affect the analysis of the driving mechanism of urban land use efficiency.

As a typical unipolar development and labor outflow province in East China, Anhui Province has a large gap in the development level of different cities, and the internal and external factors faced by cities are similarly heterogeneous. Therefore, the contradiction between urban inefficiency and high-quality urban development is more prominent and complex in Anhui Province. In addition, after the Chinese government delineated the red line for 1.8 billion mu of cultivated land, it would have been difficult for cities to obtain more construction land. Therefore, making good use of urban stock land and reducing inefficient urban land has become an urgent problem to be solved in Anhui Province. In addition, Anhui Province is an important part of China’s “Rise of Central China” strategy. The gap between inefficient and efficient urban land seriously hinders the strategic layout of the Rise of Central China”. At present, few studies have placed cities in Anhui Province under the same environmental factors and conditions attributed to luck for urban land use efficiency measurement; it is impossible to clearly understand the development status of ULUE for 16 prefecture-level cities, with existing research not having determined the influencing factors of urban land use efficiency in Anhui Province.

Therefore, a three-stage DEA model is used to eliminate the interference of random errors and environmental variables on the efficiency value of decision-making units, and the kernel density measurement model is introduced. The gravity center model is used to analyze the spatial and temporal evolution characteristics of urban land-use efficiency in 16 prefecture-level cities in Anhui Province from 2001 to 2020. Finally, a geographical detector was used to explore and reveal the driving mechanisms of spatial and temporal evolution. The calculation, driving factors, and mechanism analysis of urban land use efficiency in Anhui Province are expected to provide a reference for the optimal allocation of land resources and a high-quality sustainable development for the cities in Anhui Province.

2. Data

2.1. Area of Study

Anhui Province (114°54′ E–119°37′ E, 29°41′ N–34°38′ N) is a significant province in East China, covering an area of approximately 140,100 km². It consists of 16 prefecture-level cities and nine county-level cities, with Hefei serving as the provincial capital. The province experiences a warm temperate semi-humid monsoon climate in the north and south of the Huaihe River, while a subtropical humid monsoon climate prevails. Geographically, Anhui is divided into five central geographical regions: The Huaibei Plain, Jianghuai Hilly Area, Plain Area along the Yangtze River, Dabie Mountain Area in Western Anhui, and Hilly Area in Southern Anhui, as shown in Figure 1.

In 2020, China’s per-capita GDP was 71,828 Chinese Yuan (CNY) per person, significantly lower than the national average of 63,426 CNY per person. Hefei City, despite covering only 10.32% of the province’s area and accounting for 15.34% of the population, contributed 26% of the province’s GDP, indicating a clear trend of single-core development. According to the seventh census data in 2020, Anhui Province had a population of 11.52 million people residing outside the province, which accounted for 18.9% of the permanent population. This represents an increase of 1.14 million people, equivalent to an 11% increase, compared to the outflow population of the sixth census. These statistics highlight Anhui Province as a typical province experiencing population outflow. Furthermore, the industrial structure of Anhui Province underwent tremendous changes during the study period, resulting in an imbalance in the spatial layout structure of construction land within the province. This resulted in significant differences in land-use efficiency among cities and regions, posing obstacles to achieving high-quality and sustainable development in Anhui Province [32,33,34,35]. Therefore, it is crucial to address the urgent issues of rational adjustment of the urban spatial layout, efficient allocation of land resources, and the realization of the connotative development in Anhui cities.

2.2. Data Sources

Statistical and administrative data used in this study were obtained from the 2001–2020 China Urban Statistical Yearbook, Anhui Statistical Yearbook, and Statistical Yearbooks of Cities in Anhui Province. These statistical yearbooks have a strong correlation and fewer data changes. Therefore, the interpolation method was used to obtain missing data for certain years. The details of these indicators are presented in

Table 1. Data information and sources.

Data	Data Description	Data Sources
Administrative division data	Anhui provincial boundary vector data	Resource Environmental Science and Data Center (https://www.resdc.cn/ accessed on 17 November 2021)
Socio-economic data	GDP per capita, investment in fixed assets, local fiscal revenue, etc.	Anhui Statistical Yearbook (2002–2021) (http://tjj.ah.gov.cn/ssah/qwfbjd/tjnj/ accessed on 17 November 2021)
Land use data	Built-up area, built-up area green rate, park green area, etc.	China Urban Statistical Yearbook (2002–2021) (http://www.stats.gov.cn/tjsj/ndsj/ accessed on 17 November 2021)

.

3. Materials and Methods

3.1. ULUE Calculation

3.1.1. Three-Stage Dea Model

Data Envelopment Analysis (DEA) was first introduced in 1978 by A. Charnes, W.W. Cooper, and E. Rhodes, and was used to evaluate the relative effectiveness of the same sectors. The DEA model is essentially a black box model in which a non-parametric envelope frontier is constructed from the input and output indicators of the decision-making units (DMUs), and the effectiveness of the inputs and outputs of each decision-making unit is judged by comparing the degree of deviation of the DMUs from the DEA frontier. The traditional DEA model, however, does not aptly address the different environments in which different decision units are located and the situation of uncontrollable factors. Therefore, Fried (2002) proposed a three-stage DEA model that takes into account both environmental factors and random errors, which is a model that uses Stochastic Frontier Analysis (SFA) to remove environmental factors and random errors from the traditional DEA model [36,37,38] The results are again placed in the traditional DEA model with the original input factors. The results are then again placed in the traditional DEA model with the original input factors so that different decision units are placed under the same environmental factors and luck conditions, the model mechanism is shown in Figure 2. Therefore, this paper borrows the soft solution of Deap2.1 and Frontier4.1 to measure the urban land use efficiency in Anhui Province.

3.1.2. Index Selection

Based on the connotation of the ULUE value and the current research results [39,40,41,42,43], this study determined that the land use efficiency system includes input and output indices according to the principle of data representativeness and accessibility.

Regarding input factors, based on the composition of production input factors in the process of land use, this study selects three indicators: Land resource, capital, and labor inputs. It selects the urban built-up area to characterize land resource input, where fixed-asset investment represents capital investment and the number of employees in secondary and tertiary industries represents labor input. Concerning output indicators, based on the scientific and comprehensive nature of output benefits, the output indicators include economic, social output, and ecological and environmental benefits. The output value of the secondary and tertiary industries represents economic benefits, local fiscal revenue represents social output benefits, and the green coverage rate of built-up areas represents ecological and environmental benefits. The specific evaluation indicators are listed in

Table 2. ULUE input–output index table.

Factor	Index Types	Specific Indicators	Unit
Input index	Land resource input	Urban built-up area	km2
	Capital input	Fixed investment	Million CNY
	Labor input	Number of employees in secondary and tertiary industries	Million people
Output indicator	Economic output	The value of second and tertiary industries	Billion CNY
	Social output	Local fiscal revenue	Million CNY
	Ecological environment output	The green coverage rate of built district	%

.

3.1.3. Selection of Environmental Indicators

When selecting indicators, we examined the structural characteristics of land use efficiency and considered urban development conditions. Following the above principles and drawing on existing research, this study selected two indicators: Urban traffic conditions and openness [44,45].

The two indicators of urban traffic conditions and the degree of openness to the outside world in the construction of internal urban infrastructure and transportation facilities are significant. The development and expansion of cities must be supported by urban traffic networks. Urban growth also depends on external economic conditions. A comprehensive and open multifaceted town can allocate social resources in a rational and orderly manner. Therefore, this study selected the per capita urban road area to characterize urban traffic conditions. Hence, the total import and export volumes represent the degree of openness of the city. The specific evaluation indicators are listed in

Table 3. ULUE environmental factors index selection table.

Index Types	Specific Indicators	Unit
Urban traffic conditions	Per capita urban road area	m2
Degree of opening to the outside world	Total export-import volume	Million United States Dollar (USD)

.

3.2. Kernel Density Estimation

Kernel density estimation (KDE) is an economic data distribution detection model. Kernel density estimation has the objectivity of the preset function and accuracy of element state capture. The difference in the time-series evolution can be intuitively revealed [46]. The specific methods used are as follows:

$$F\left(x\right)=\left(1/nh\right){\displaystyle \sum }_{i=1}^{n}K\left(\left(x_i-x\right)/h\right)$$

(1)

where n is the number of samples; $h$ is bandwidth; $x_i$ is the sample observation value $x$ is the sample mean; $K\left(·\right)$ is a kernel function. Therefore, with the help of Matlab2020 a, based on the land use efficiency value of 16 prefecture-level cities in Anhui Province, through kernel density estimation, the temporal evolution of land use efficiency in Anhui Province is revealed by analyzing the peak shape, kurtosis, and peak center of gravity of the curve.

3.3. Gravity Center Model

The gravity center model was used to solve the problem of the spatial transition of regional attributes and to characterize the centralized, discrete distribution trends and migration trajectories of spatial qualities [47,48,49]. The spatial moving distance of the center of gravity of the land use efficiency value refers to the direct spatial distance between the gravity coordinate of the land use efficiency value in the jth year and the gravity coordinate of the j + 1 year [50]. The formula used is as follows:

$$D_{\left(j+1\right)-j}=R\times \sqrt{{(y_{j+1}-y_j)}^2+{(x_{j+1}-x_j)}^2}$$

(2)

where $D$ represents the moving distance of the center of gravity; ($x_j$,$y_j$), ($x_{j+1}$,$y_{j+1}$) denotes the $j$ year and $j+1$ year barycentric coordinates;$R$ is a constant value of 111.11 km. With the help of ArcGIS 10.2 software, this paper calculates the spatial center of gravity, migration direction, and moving distance of land use efficiency values of 16 prefecture-level cities in Anhui Province through the center of the gravity model to describe the spatial representation of land use efficiency values in the province.

3.4. Geographical Detector

3.4.1. Geographical Detector Model

Geographical detection is a crucial method used to study the formation mechanism of spatial differences in geographical elements based on geographical principles. The principle was to test whether the spatial differentiation of attributes was consistent with that of the factors [51,52]. Its advantage is that it can directly describe the interactions among different driving factors. This is discussed as follows:

$$P_D=1-\left(1/n{\sigma }^2\right){\displaystyle \sum }_{i=1}^k{n}_i{\sigma }_i^2$$

(3)

Among them, $P_D$ is the driving force of driving factor $D$; $n$ and ${\sigma }^2$are the number of decision-making units and the discrete variance of land use efficiency in Anhui Province; $n_i$ is the number of decision-making units; ${\sigma }_i^2$ is the discrete variance of land use efficiency of the secondary decision-making unit; $k$is the number of hierarchical partitions for driving factors. Therefore, this paper reveals the formation mechanism of spatial differences in urban land use efficiency from five aspects and six driving factors, including the degree of development, the degree of opening up, the industrial structure, the level of scientific and technological development, and the scale of the city.

3.4.2. Selection of Driving Factors of Land Use Efficiency

Complex social activities influence drivers of land use efficiency. Hence, it is essential to reveal the drivers of the spatial and temporal evolution of land-use efficiency to achieve value-added land-use efficiency. Drawing on existing studies [53], this study selects the urban built-up area as a proxy for city size, fixed asset investment as a proxy for economic capital input, the total population employed in secondary and tertiary industries as a proxy for human capital input, population urbanization rate, and gross domestic product (GDP) per capita as a proxy for urban development level, and the tertiary-industry-to-GDP ratio as a proxy for industrial structure. Furthermore, the ratio of the tertiary sector to GDP indicates the industrial format, and the amount of foreign capital utilized suggests the degree of openness to the outside world. This reveals the role of different factors in driving the spatial and temporal evolution of urban land use efficiency values. The specific evaluation indicators are listed in

Table 4. ULUE driving factor index table.

Driving Factors	Specific Indicators/Units	Variable Code
City size	Urban built-up area/km2	X1
Economic capital	Investment in fixed assets/Million CNY	X2
Human capital	Total number of people employed in secondary and tertiary sectors/Million people	X3
Level of urban development	Population urbanisation rate/%	X4
	GDP per capita/CNY/person	X5
Industrial structure	Tertiary sector to GDP ratio/%	X6
Degree of external openness	Actual amount of foreign investment utilised/Million USD	X7

.

4. Results

4.1. Results of ULUE

4.1.1. First Stage of DEA Results

An input-oriented DEA model was used to calculate the input and output indicators of 16 prefecture-level cities in Anhui Province from 2001 to 2020. To improve the accuracy of the calculations, the data were processed annually. However, because the impact of environmental factors and random errors on land use efficiency was not considered in the first stage, this study did not conduct a specific analysis; the results are shown in Figure 3.

4.1.2. Second-Stage SFA Regression Results

The second stage of the SFA extracts the slack variables in the first stage. The slack variables for each input factor, urban road area per capita, and total import and export volumes were selected as explanatory variables. We entered them into the SFA regression model to analyze and explore whether there was a significant difference between the actual and ideal input variables. The formula is then used to obtain the adjustment input after excluding environmental variables and random errors. To ensure the calculation is accurate, this paper uses Frontier 4.1 to calculate the data over ten years. The unilateral error likelihood ratio test results meet the essential requirements, indicating that the SFA regression model can be used for regression. The value tends toward 1 within ten years, indicating that the management inefficiency term primarily influences the factors influencing the slack variable. The relationship with random error is weak.

A difference exists between the actual and ideal inputs, referred to as the slack variable, caused by the management level. A positive coefficient between the environmental and slack variables shows that the external environment hinders land-use efficiency. Conversely, a negative coefficient suggests a favorable impact of the external environment on land use. This study listed the slack variables of three input factors for the years 2001, 2006, 2011, 2016, and 2020. The results, presented in

Table 5. Relaxation variable and environmental variable coefficient table.

Year	Slack Variables	Per Capita Urban Road Area	Total Export-Import Volume
2001	Urban built-up area	0.0613	0.00001
	Fixed investment	2100.5752	0.01448
	Number of employees in secondary and tertiary industries	−1.46	0.00005
2006	Urban built-up area	2.1108	0.00005
	Fixed investment	10,337.9200	0.06297
	Number of employees in secondary and tertiary industries	−0.6972	0.00003
2011	Urban built-up area	−0.0070	0.00001
	Fixed investment	−7547.0386	0.17256
	Number of employees in secondary and tertiary industries	2.0930	−0.00002
2016	Urban built-up area	−0.2888	−0.00001
	Fixed investment	−16,893.7300	0.16635
	Number of employees in secondary and tertiary industries	1.6806	0.00003
2020	Urban built-up area	0.0414	−0.00001
	Fixed investment	−6889.7430	−0.37872
	Number of employees in secondary and tertiary industries	0.8622	0.00001

, indicate that the per capita road area significantly impacts land-use efficiency regarding fixed asset inputs. Notably, the coefficient shifts from positive to negative, indicating that the external condition of fixed asset investment has increasingly become a key driving factor affecting urban land-use efficiency. However, the total import and export volumes are insignificant in improving land use efficiency. Furthermore, the correlation coefficient between the metropolitan built-up area and fixed asset investment also changes from positive to negative, indicating that the improvement of openness has enhanced the urban scale efficiency and asset investment structure in the 16 prefecture-level cities in Anhui Province, ultimately improving land use efficiency.

4.1.3. The Third Stage DEA Results

After the second stage of the SFA regression, the input variables and original output values were entered into DEAP 2.1 software to calculate the efficiency value. When comparing the average calculation results for the comprehensive, purely technical, and scale efficiencies of land use with the first-stage results, significant changes were observed in the majority of cities. Consequently, it becomes crucial to adjust the input variables for each city to ensure a uniform external environment and luck conditions across all cities. The results are shown in Figure 4.

After the second regression stage, Ma‘anshan remained the leading city regarding ULUE efficiency value, with a ULUE value of 1 over 20 years, signifying it as the top-performing city concerning ULUE in Anhui Province. Hence, the ULUE values for eight prefecture-level cities increased. Hefei, Suzhou, Bengbu, Huainan, Lu’an, Wuhu, Chizhou, and Huangshan exceeded their first-stage levels, indicating that the economic environment contributed to promoting land use efficiency. Notably, Hefei exhibited the most significant increase in ULUE value, rising from 0.947 in the first stage to 0.947, representing a growth rate of 2.8%. This suggests that the external environment had the most substantial positive impact on Hefei’s ULUE value. Conversely, Huaibei, Bozhou, Fuyang, Chuzhou, Xuancheng, Tongling, and Anqing experienced a decrease in ULUE values, indicating that the economic environment of these seven cities hindered land use efficiency. Among them, Xuancheng displayed the most substantial reduction, declining from 0.99 in the first stage to 0.98, reflecting an attenuation rate of 1.04%. This indicates that the external environment had the most substantial inhibitory effect on Xuancheng’s ULUE value. The results are shown in Figure 5.

4.2. The Temporal Evolution of ULUE

4.2.1. ULUE Timing Evolution

Using ArcGIS 10.2 software to classify the comprehensive efficiency of land use in 16 prefecture-level cities, it is divided into six intervals: Low-value interval, lower-value interval, median interval, higher-value interval, high-value interval, and efficiency frontier. The results are shown in Figure 6. According to the map, in 2001, the land use efficiency of Anhui Province was at a low level at the beginning of the century. Only five cities in the province were at the efficiency frontier, and the average ULUE was only 0.85, indicating that the utilization and transformation of land resources were inefficient. However, after 20 years of development, 10 cities in 2020 have reached the frontier of efficiency value, with an average ULUE value of 0.959, an increase of 12.74% compared with the ULUE value of 20 years ago showed a non-linear increasing trend, which was primarily related to the social and economic activities of Anhui Province. During the rapid development of Anhui Province from 2001 to 2020, a large agricultural population transferred to non-agricultural areas, causing the city to develop rapidly. It has also led to a gradual increase in urban land use efficiency.

4.2.2. ULUE Kernel Density Estimation

Based on the land use efficiency values of cities in Anhui Province from 2001 to 2020, Matlab2020a software was used to draw the kernel density curve of ULUE in Anhui Province from 2001 to 2020 (Figure 7), and the temporal evolution characteristics of different land use efficiency values of cities can be intuitively seen. The details are as follows: (1) From the shift in the peak center of gravity, the distribution center of ULUE in Anhui Province continues to move to the right, indicating that the ULUE level of each city shows a rising trend. (2) From the perspective of kurtosis, the maximum value of kurtosis appeared in 2014, which indicates that there are cities with a high level of ULUE development, and most cities are clustered in high-value areas. (3) From the shape and number of peaks, the distribution of ULUE values in Anhui Province from 2001 to 2013 had only one single peak, and it was not steep, indicating that there was no noticeable polarization trend in the development of ULUE in Anhui Province during this period. Since 2014, the distribution of ULUE values in the Anhui Province has shown a single-peak–multi-peak transition, and the peaks have suddenly become steeper. This indicates that the development gap between cities in the high- and low-value areas of ULUE in Anhui Province is widening, and the “ULUE gap” and multilevel differentiation have begun to appear. Therefore, in general, the development of ULUE values in Anhui Province shows a dynamic evolutionary trend of continuous improvement at the overall level, but the multi-level differentiation and gaps increase among cities.

The results of the kernel density model revealed a non-linear trend in the ULUE value of Anhui Province. The reason can be attributed to the implementation of the “Rise of Central China” development strategy by the central government in 2008. The strategy fostered the balanced development of ULUE among cities. With lower resistance from governments at all levels in Anhui Province and faster implementation of the new policy, land use efficiency between cities has continuously improved. Considering the lag between the policy formulation and its effect, it is reasonable to expect a peak in 2014. Furthermore, in 2016, Anhui Province introduced the “Opinions on promoting the redevelopment and utilization of inefficient construction land (Trial),” which when, combined with the adjustments and upgrades in the industrial structure and overall urban and rural development, led to innovative and improved land use management mechanisms. These efforts promoted the redevelopment and utilization of inefficient construction land, resulting in another peak in 2016. This indicates a gradual maturation and stabilization of land use benefits transformation across various cities, and a more rational allocation of land resources.

Existing research shows that the land use mode in Anhui Province has undergone changes owing to the implementation of the national and regional development strategy, as well as a series of laws and regulations aimed at improving land use efficiency. The findings of this study, in conjunction with other research, suggest that policies such as the “Rise of Central China” strategy, high-quality development, and promotion of inefficient construction land redevelopment and utilization have proven to be highly effective in significantly reducing the emergence of low-value clusters in ULUE.

However, while the ULUE values in Anhui Province are growing non-linearly, the development gap between cities in the high-value and low-value areas is widening, and a “ULUE divide” and multi-level differentiation are beginning to emerge. The main reason for this is that in 2019, the Anhui provincial government proposed the “One Circle, Five Regions” strategy. The “One Circle, Five Regions” strategy is more biased towards the priority development of regional central cities, although the implementation of the “One Circle, Five Regions” strategy has improved the overall economic and social development of the province. Although the implementation of the “One Circle, Five Regions” strategy has improved the overall economic and social development of the province, generating certain economic and social benefits, there are also problems and challenges. For example, the uneven distribution of resources and the imbalance in development between cities, which has led to a large gap in development levels between cities in Anhui Province, has led to the emergence of a multi-level division in the province.

4.3. Spatial Evolution of ULUE

Firstly, ArcGIS 10.2 software was used to extract the barycentric coordinates of land-use efficiency of cities in Anhui Province from 2001 to 2020, and its latitude and longitude values were obtained. The center of gravity of the ULUE in Anhui Province was between 117.265° E–117.369° E and 31.772° N–31.931° N. In 2020, the barycenter coordinates migrated from 2001 (117.368° E, 31.773° N) to (117.303° E, 31.884° N), moving northward and westward, with a total migration of 22.6626 km.

The change in the barycenter coordinates of the spatial trajectory can be divided into five stages. (1) First stage (2001–2004): In this stage, the main direction of movement of the barycentric coordinates was northwest, with a total migration of 17.99 km; (2) Second stage (2005–2007): During this stage, the barycenter coordinates were primarily to the south, with a total migration of 2.5104 km; (3) Third stage (2008–2011): In this stage, the main direction of movement of the barycentric coordinates was northwest, with a total migration of 1.7984 km; (4) Fourth stage (2012–2016): In this stage, the main direction of movement of the barycentric coordinates was east, with a total migration of 0.0767 km; (5) Fifth stage (2017–2020): In this stage, the main direction of movement of the barycentric coordinates was the northeast, with a total migration of 0.2781 km. Moreover, the migration rate of the barycenter coordinates in the east–west direction is significantly greater than that in the north–south direction, the results are shown in Figure 8.

Among them, in the first stage (2001–2004) and the second stage (2005–2007), the ULUE center of gravity moved the most, 20.5004 km, accounting for 90.45% of the overall migration distance. In 2004, the ULUE center of gravity moved the most in a year, with a migration distance of 17.2047 km, accounting for 75.91% of the overall migration distance. In the third stage (2008–2011), fourth stage (2012–2016), and fifth stage (2017–2020), although the ULUE center of gravity moved in the north–south and east–west directions, the moving distance, and rate were relatively slow and remained stable for a long time.

From the perspective of the moving trajectory of the center of gravity of land use efficiency from 2001 to 2020, the center of gravity of land use efficiency in Anhui Province shifted in both the north–south and east–west directions. However, the difference in movement in the north–south direction was significantly greater than that in the east–west direction. During the study period, the largest migration occurred in the first stage (2001–2004) and second stage (2005–2007). In the first stage (2001–2004), the center of gravity of the ULUE rapidly moved to the northwest, indicating that the efficiency difference between the southeast and northwest urban agglomerations expanded rapidly. The main reason for this is that, in 2004, the Anhui provincial government implemented the “861” plan. This government established nearly 800 projects in northern and western Anhui, with an investment of approximately trillions of yuan, making the social fixed asset investment in northern and western Anhui significantly increase by 29.6% compared to 2003, which significantly promoted the development of northern and western Anhui. During the second stage (2005–2007), the center of ULUE shifted southeastward because the Anhui provincial government earnestly implemented the eastward development strategy and strengthened cooperation with the Yangtze River Delta region to enhance openness. Although the center of gravity of ULUE experienced subsequent movement from 2009 to 2020, the migration rate was low, and the migration distance per unit year was also small, maintaining a relatively stable state.

Overall, the center of gravity of the land use space in the province moved slowly, and the moving distance was also small. This indicates that the center of gravity of the land use efficiency value in Anhui Province was relatively balanced and stable during the study period. However, from the perspective of the center of gravity migration rate and distance, the efficiency center of gravity migration rate and distance in the north–south direction of Anhui Province are much larger than those in the east–west direction because of the status quo of Anhui Province and its economic development. Anhui Province has a large north–south span and a small east–west span. When the unit variable changed, the movement in the north–south direction was far greater than that in the east–west direction. At the same time, the southern cities of Anhui Province are closely connected with Jiangsu, Zhejiang, and other places, which are strongly affected by radiation and have a relatively high level of development. The northern cities of Anhui Province are primarily agricultural. The surrounding provinces had a small driving effect on the northern cities of Anhui Province, and the development level was low. The large gap in urban development level will lead to a gap in urban land use efficiency. However, from 2001–2020, the Anhui provincial government adopted a series of policy measures and development strategies for the northern region, which made the cities in the northern region develop rapidly, continuously narrowed the development gap between the northern and southern regions, and made the gap between the land-use efficiency of the northern and southern cities smaller. Therefore, the center-of-gravity migration rate and distance in the north–south direction were also excellent. In contrast, in the eastern and western parts of Anhui Province, the difference in the development level among cities was small, and the development rate during the study period was also relatively close, making the direct gap in urban land use efficiency unclear.

4.4. Driving Factors of ULUE

4.4.1. Detection of Single Driving Factor

Based on the spatial and temporal evolution characteristics of land-use efficiency in Anhui Province, the driving mechanisms of different driving factors on the evolution of land-use efficiency were studied using a geographical detector model. Each driving characteristic passed the 1% significance test, indicating that each element had a driving effect on the change in land use efficiency, but the driving impact was different.

In the years 2001, 2006, 2016, and 2020, the urban built-up area (X₁) made the most significant contribution to ULUE; in 2016, the coefficient was 0.4795, ranking fourth. In 2011, the influence coefficient of urban built-up areas on ULUE reached 1, making it the most critical factor affecting ULUE that year. Second, the contributions of openness (X₇) and development (X₅) to ULUE were relatively high. The contribution rates of human capital (X₃) and industrial structure (X₆) to ULUE were the lowest, and the average contribution rate to ULUE during the study period was below 0.35. However, the contribution rate of the human capital factor (X₃) to ULUE gradually decreased over 20 years, while the contribution rate of the industrial structure factor (X₆) decreased by 0.0966 in 2016.

If the maximum value of the factor contribution rate in each year is taken as the main driving factor of the clues, the main driving factor in 2001, 2006, 2016, and 2020 is the urban built-up area (X₁), and the main driving factor in 2016 is the degree of openness (X₇). If the main driving factors of the clues are determined by the average value of the contribution during the study period, the urban built-up area (X₁) is 0.8094 as the main driving factor, and the degrees of openness (X₇) and development (X₅) are 0.6616 and 0.6615, ranking 2, 3, the results are shown in

Table 6. The ULUE driving factors index of Anhui Province from 2001 to 2020.

Year	X1	X2	X3	X4	X5	X6	X7
2001	0.8908	0.7679	0.3741	0.4112	0.7784	0.4354	0.8254
2006	0.7668	0.5871	0.4359	0.4329	0.6215	0.3236	0.5504
2011	1.0000	0.4042	0.4828	0.4446	0.7272	0.4115	0.5675
2016	0.4795	0.3524	0.2173	0.6383	0.3534	0.0966	0.8522
2020	0.9097	0.4192	0.1768	0.7347	0.8270	0.3250	0.5126

.

The analysis of the empirical study revealed a clear duality in the size of the urban built-up area as one of the key factors influencing ULUE. On the one hand, the expansion of urban built-up areas can provide more urban land use space to meet the needs of urban development, thereby improving urban land use efficiency; on the other hand, blindly increasing the urban built-up areas will also bring some adverse effects. For example, an increase in many urban built-up areas will lead to the overexploitation and waste of urban land, leading to the excessive consumption of urban land resources and seriously affecting the sustainability of urban land use. From 2001 to 2020, Anhui Province was in a stage of rapid urbanization, and the urban built-up area of each city increased. This development mode relies primarily on expanding urban built-up areas to improve urban land-use efficiency. However, the disorderly expansion of urban built-up areas has led to the waste of urban land resources and reduced the efficiency of urban land use. The impact of the degree of openness (X₇) was more indirect than that of the urban built-up areas. The degree of urban openness can promote economic development and urbanization, optimize and transform urban land use to a certain extent, and improve urban land use efficiency. However, the degree of urban opening up to the outside world makes urban land tense. For example, the excessive development of industrial areas and inefficient urban commercial centers will lead to a limited distribution of land resources. In summary, the two main controlling factors are complex and dual. In urban development, it is crucial to comprehensively consider the relationship between different influencing factors and urban land-use efficiency and formulate corresponding policy measures and planning methods to optimize urban land-use structure and improve urban land-use efficiency.

4.4.2. Double Drive Due to Interaction Detection

This study conducts an interactive detection of different driving forces to determine whether the interaction between the driving forces increases or decreases compared with a single driving force. The results showed that the interaction between the two factors was nonlinearly enhanced. No changes or decreases were observed, indicating that the influence of the two-factor interaction was more significant than that of the two single independent factors. The enhancement of the two-factor impact indicates that the driving force is a collection of interrelated components rather than a single element. This nonlinear enhancement indicates that the evolution of ULUE is not limited to a single part but is determined by the synergy of multiple factors.

The non-linear enhancement of the two-factor interaction dominated ULUE in 2016. With the exception that the interaction between human capital (X₃) and industrial structure (X₆) was 0.36103 at a low level, the remaining interaction coefficients were between 0.9 and 1.0 at a high level. In addition, among the ULUE influencing factors, the interaction between the urban built-up area (X₁), the openness factor (X₇), and other factors was more significant than the interaction between different factors, and the interaction factor value was high. This shows that urban built-up area (X₁) and degree of openness (X₇) are significant factors influencing ULUE. However, the interaction between human capital (X₃), development level (X₅), and other factors was significantly lower than that between the different elements, the results are shown in Figure 9. Therefore, this study believes that it is necessary to ensure that urban built-up areas meet the requirements of sustainable development by reasonably controlling the speed and scale of urban expansion: Improve the degree of urban opening-up, promote regional economic development, and accelerate the transformation and upgrading of cities; integrate and optimize science and technology education resources; improve the technical level and quality of the labor force; enhance the city’s innovation ability and competitiveness; optimize the industrial structure to change the inherent mode of land use efficiency that initially depended on land resources, capital investment, and labor resources; open up new ideas for the development of land use efficiency; and effectively improve the efficiency of urban land input and output.

5. Discussion

Studying the efficiency of urban land use in Anhui Province is of great significance for promoting the high-quality and connotative development of cities in the province. However, existing research does not consider the deviation caused by urban environmental factors and random disturbances in the calculation of urban land use efficiency. This study considers the urban land-use efficiency of 16 prefecture-level cities in Anhui Province to provide an empirical reference for the efficient allocation and utilization of urban land resources in Anhui Province.

Empirical research shows that urban land use efficiency in Anhui Province is relatively high, while the ULUE shows a non-linear growth trend during 2001–2020. This is consistent with the level of urban economic and social development in Anhui Province. As economic and social development increases the input of production factors for urban land use efficiency, an increase in urban land output is to follow, thereby improving urban land use efficiency. This also confirms the view emphasized by some scholars of the importance of urban production factors in urban land-use efficiency [54,55]. However, it should be noted that because of the differences in the attributes of the research objects brought about by different cities, regions, and times, the selection of urban land-use efficiency measurement indicators will be focused on—thus, the measurement results will be different. Therefore, it is necessary to select appropriate evaluation and calculation methods for different research objects to make the calculation of urban land use efficiency more accurate and scientific.

From the results of the time series evolution, although urban land use efficiency in Anhui Province shows a non-linear growth trend, there is also a “ULUE gap” and multilevel differentiation. However, this was not an example of this phenomenon. In the studies of other scholars, although there are differences in the choice of research objects, the results show that there are gaps in the land-use efficiency of different cities in the region, and that these gaps do not show a decreasing trend [56,57]. Bridging the gap between ULUE in different cities is also a direction for future research.

From the perspective of spatial evolution, the center of gravity of the ULUE value in Anhui Province moves in the north–west direction as a whole, but the overall relatively stable moving distance remains small, with the migration rate and distance of the efficiency center of gravity in the north–south direction appearing much larger than those in the east–west direction. According to the research, the center of gravity movement of ULUE in Anhui Province is related to the current situation of its own province and the economic development situation. The main reason for the lack of an obvious center of gravity migration in the region may be that the selection scale and research period of the study area are relatively short, and that it is difficult to find its evolution law from the macro level, with the potential to conduct in-depth research on the topic in the future.

Finally, the driving factors and mechanisms of ULUE in Anhui Province were explored. The comprehensive influence of any two driving factors is greater than that of a single factor, and the urban built-up area (X₁) and degree of openness (X₇) are the main control factors at both levels. This also confirms that the development mode of cities in Anhui Province primarily depends on the input factors of urban built-up areas during the study period. However, if a city wants to achieve high-quality and sustainable development, it must change its original development mode. Therefore, the Anhui provincial government needs to further promote new ideas for urban land use efficiency development and realize a virtuous cycle of urban economic, social, and ecological development, and urban land use.

This study has some limitations. On the one hand, this study is limited to urban areas, with uncovered areas for the whole Anhui Province present; therefore, it is necessary to adopt a smaller scale in future research to improve the accuracy and scientific nature of these results. On the other hand, the selection of input-output and external environmental factors needs to be further explored and must be comprehensively considered from the natural environment, policy factors, and other aspects. Moreover, although this study evaluates ULUE, it is impossible to determine the quantitative impact of changes in input factors on output factors. Future studies should also focus on this aspect of evaluation.

6. Conclusions

By studying the spatial and temporal evolution and driving factors of urban land use efficiency in Anhui Province, optimizing urban land allocation in the Province takes on great significance, promoting the full transformation of urban land use efficiency, and realizing high-quality and connotative development of cities. Therefore, this study selected 16 prefecture-level cities in Anhui Province as samples for studying ULUE in Anhui Province from 2001 to 2020. First, the urban land-use efficiency of 16 prefecture-level cities was calculated based on the T-DEA. On this basis, the kernel density estimation model and center of gravity model were used to quantitatively analyze the spatial and temporal evolution characteristics. Finally, combined with the calculation of urban land use efficiency in Anhui Province and the results of spatial and temporal evolution analyses, a geographical detector model was used to explore the driving factors and mechanisms of ULUE values.

The conclusions are as follows:

• From the results of urban land use efficiency, the overall level of ULUE in Anhui Province is high, but the gap between the different cities remains large. Compared to the results of the third stage, the ULUE values of most cities in the first stage changed significantly. To ensure the accuracy of the results, it is necessary to ensure that each city has the same external environment and luck conditions.
• From the perspective of a time-series evolution, ULUE generally shows a nonlinear growth trend over time, which is consistent with the economic and social development of Anhui Province. However, at the same time of growth, there is also a trend of a ‘ULUE gap’ and multilevel differentiation.
• From the perspective of spatial and temporal evolution, the center of gravity of the ULUE space moves slowly, and the moving distance remains small. The migration rate and distance of the efficiency center of gravity in the north–south direction are far greater than those in the east–west direction, which is related to the current situation and economic development of Anhui Province.
• From the perspective of the driving factors, the combined influence of any two driving factors is greater than that of a single factor. However, from the analysis of single and double factors, the urban built-up area and the degree of opening to the outside world are the key controlling factors affecting the ULUE value; however, these two main controlling factors have obvious duality and complexity.

In general, this study enriches research on urban land use efficiency by comprehensively analyzing the prevailing situation, spatial and temporal evolution laws, and driving mechanisms of urban land use efficiency in Anhui Province. The research results have the following policy implications: First, the government needs to recognize the trend of hierarchical differentiation of ULUE in Anhui Province and formulate corresponding land use policies for each city to ensure the coordinated use of urban land systems throughout the province. Second, the government should strengthen its comprehensive consideration of the complexity of different driving factors, appropriately increase urban built-up areas, and adopt a more active urban opening-up policy. Third, the government needs to actively integrate and optimize science and technology education resources, improve the technical level and quality of the labor force, and enhance cities’ ability to innovate and compete. Moreover, we emphasize the importance of further research and the need to conduct targeted research on the land-use efficiency of different types of cities and different functional areas. Finally, the study postulates that, due to the new urban development model brought about by high-quality urban development and social change, research on urban land use efficiency will also show trends of multidisciplinary, multi-source data, and multi-research perspectives.