Research fields and data sources
place of learning
Hubei Province is located in the central part of the Yangtze River and borders Anhui Province to the east, Jiangxi and Hunan Provinces to the south, Chongqing to the west, Shaanxi Province to the northwest, and Henan Province to the north. As of the end of 2020, the permanent population of Hubei Province reached 57.75 million people, of which 36.32 million were urban residents and 21.43 million were rural residents. In addition, the number of rural residents ranks 11th out of 31 provincial-level administrative regions in mainland China. In addition, Hubei's total agricultural output reached 730.36 billion yuan, making Hubei the seventh largest agricultural province in China. Total area is 185,900km2There are 13 prefecture-level administrative regions in Hubei Province: Wuhan, Huangshi, Shiyan, Yichang, Xiangyang, Ezhou, Jingmen, Xiaogan, Jingzhou, Huanggang, Xianning, Suizhou, and Enshi, including 103 prefecture-level cities. (Figure 2)). In this study, urban areas involving county-level administrative districts were excluded, and the remaining 70 county-level units away from cities were selected.
Location and evaluation unit of Hubei Province, China.
data source
The data in this article includes spatial vector data and statistical data. Spatial vector data is rural construction land approved by the government from 2009 to 2018. Such land consists of patches containing the number of registrations of approval, date of modification, area, and spatial location of rural construction land. These data are included in the Land Use Change Survey Database and provided by the Hubei Province Department of Natural Resources. Statistical data on socio-economic development, resource and environmental protection and utilization can be found in Hubei Provincial Statistical Yearbook (2010-2020), China Statistical Yearbook (Prefecture Level) (2010-2020), and Hubei Provincial Statistical Bulletin (2009-2020). Retrieved from 2020). 2020).
indicator system
Characteristics of government support for rural development based on the LRCLE patch
Support strength: area change
The dynamic degree of single land use (DDSLU) was chosen to describe the area change of legal rural construction land characterizing the strength of government support (Liu et al. 2014). To characterize the intensity of LRCLE over a period of time in different units, the ratio of the area of statutory rural construction land increment to the area of the reference period was calculated. The formula for calculating the statutory rural construction land increment DDSLU is as follows:
$$S=\frac{{S}_{a}-{S}_{b}}{{S}_{b}}\,* \,\frac{1}{\Delta t}$$
(1)
Here, S is the ratio of the area of statutory rural construction land increment to the area of the reference period in different periods. \({S}_{b}\) and \({S}_{a}\) The area of legal rural construction land at the beginning and end of the monitoring period, respectively. Δt is the monitoring period. He divided the changes in the area of legal rural construction land over the past 10 years into three periods (2009-2011, 2012-2014, and 2015-2018).
Spatial direction of support: directional distribution
The analysis of changes in the spatial distribution direction refers to the outline and dominant direction of regional economic attributes and geographical factors in the spatial distribution, and the standard deviation ellipse (SDE) is a classical method for analyzing the characteristics of the spatial distribution direction. This is one of the approaches. Therefore, SDE is selected to quantitatively explain the centrality, distribution, orientation, and spatial form of the spatial distribution of legal rural construction land increments from a global and spatial perspective (Du et al. 2019 ). The format of SDE is:
$$\,{{SDE}}_{x}=\sqrt{\frac{{\sum }_{i=1}^{n}{({x}_{i}-\bar{X}) }^{2}}{n}}$$
(2)
$${{SDE}}_{y}=\sqrt{\frac{{\sum }_{i=1}^{n}{({y}_{i}-\bar{Y})}^ {2}}{n}}$$
(3)
where \({x}_{i}\) and \({y}_{i}\) are the coordinates of the element I, \(\left\{\bar{X,}\bar{Y}\right\}\) represents the mean center of the element, n Equal to the total number of elements. Here's how to calculate the rotation angle:
$$\tan \theta =\frac{A+B}{C}$$
(Four)
$$A=\left(\mathop{\sum }\limits_{i=1}^{n}{\widetilde{x}}_{i}^{2}-\mathop{\sum }\limits_{i =1}^{n}{\widetilde{y}}_{i}^{2}\right)$$
(Five)
$$B=\sqrt{{\left(\mathop{\sum }\limits_{i=1}^{n}{\widetilde{x}}_{i}^{2}-\mathop{\sum } \limits_{i=1}^{n}{\widetilde{y}}_{i}^{2}\right)}^{2}+4{\left(\mathop{\sum }\limits_{i =1}^{n}{\widetilde{x}}_{i}{\widetilde{y}}_{i}\right)}^{2}}$$
(6)
$$C=2\mathop{\sum }\limits_{i=1}^{n}{\widetilde{x}}_{i}{\widetilde{y}}_{i}$$
(7)
where \({x}_{i}\) and \({y}_{i}\) is the coordinate of the mean center of. \({x}_{i}\) and \({y}_{i}\), Each. The standard deviations on the x and y axes are:
$${\sigma }_{x}=\sqrt{2}\sqrt{\frac{{\sum }_{i=1}^{n}{({\widetilde{x}}_{i}\ cos \theta -{\widetilde{y}}_{i}\sin \theta )}^{2}}{n}}$$
(8)
$${\sigma }_{y}=\sqrt{2}\sqrt{\frac{{\sum }_{i=1}^{n}{({\widetilde{x}}_{i}\ sin \theta +{\widetilde{y}}_{i}\cos \theta )}^{2}}{n}}$$
(9)
The geometric centroid of each added legal rural construction land patch was taken as the coordinate location. In addition, the percentage of the area of each added rural construction land patch to the total area of the added rural construction land patch was used as a weight to calculate the SDE of legal rural construction land expansion.
Spatial accumulation of supports: volume distribution and landscape patterns
It is difficult to clarify regional agglomeration based on administrative divisions. The number of patches determined by grid analysis is often used to analyze the characteristics of local aggregation and the distribution of hot spots. Therefore, considering the grid cells included in the related rural construction land survey (Yang et al. 2015), this study uses 10 km to count the number of added legal rural construction land patches. × 10 km grid was used. Additionally, the discreteness of legal rural construction land expansion (DDLRCLE) was defined as the number of legal rural construction land patches added to each grid cell. Changes in DDLRCLE explain local spatiotemporal processes of rural construction land expansion.
Landscape pattern indexes are widely used to describe the spatiotemporal characteristics of landscapes. Specifically refers to natural or man-made geological formations. The index can be divided into his three scales: patch, class and landscape scales (Yohannes et al. 2021). Considering that this paper focuses on a single type of land use change, class-level indicators were selected to calculate the morphological characteristics of added legal rural construction land patches on a microscopic scale. it was done. This index includes patch density (PD), edge density (ED), maximum patch index (LPI), landscape shape index (LSI), fractal index distribution (FRAC_AM), and Euclidean nearest neighbor distance (ENN_AM) . LPI, LSI, and FRAC_AM represent morphological features, and PD, ED, and ENN_AM represent structural features. Relationships between different grouped indexes are verified against each other.
Direction of rural development
Based on a theoretical analytical framework that takes into account the research on indicators used to evaluate rural development (Peng and Wang, 2020; Long et al. 2022; Li et al. 2021), and based on farmers' aspects An indicator system has been established. It reflects the direction of rural development, including livelihood and social security, rural public services, peasant production, and industrial development. The indicator system is shown in Table 1.
panel data regression
Based on panel data from prefecture-level administrative districts in Hubei province, a panel data regression model was constructed after unit root test and variance inflation factor test. This model can identify the direction of government support through legal rural construction land allocation for rural development.
Unit root test: Augmented Dickey-Fuller test (ADF)
When using time series models, the time series must be smooth. Therefore, the first step should be a smoothness test, and the commonly used rigorous statistical test is the ADF test, which is a unit root test. If a unit root exists, this result indicates that the time series is unbalanced. Panel data regression is usually not possible when there is a unit root, that is, when the time series data is not stationary. However, it is possible to differentiate the data, and usually two differentiations are performed. The second-order difference is a second-order difference based on the first-order difference. If the second-order differencing is still non-stationary, the data is poor and no longer really significant, so no further differencing is usually performed.
Variance Inflation Factor (VIF) Test
Multicollinearity refers to the high correlation between explanatory variables in a linear regression model, which distorts the model estimates or makes accurate estimation difficult. VIF was chosen to test whether there is multicollinearity in indicators of rural development orientation.
The variance of the parameter estimates is:
$$V{ar}({\hat{\beta }}_{i})=\frac{{\delta }^{2}}{{\sum }_{t=1}^{n}{\ left({x}_{{it}}-{\bar{x}}_{i}\right)}^{2}}\frac{1}{1-{R}_{i}^{2 }}$$
(Ten)
where \({R}_{i}^{2}\) is an explanatory variable \({I}\) as a dependent variable. The goodness of fit after regression for other explanatory variables is:
$${x}_{{ij}}={\alpha }_{0}+{\alpha }_{1}{x}_{1j}+{\alpha }_{2}{x}_{ 2j}+\cdots +{\upsilon }_{j}$$
(11)
The second half is removed separately to obtain the variance inflation factor (VIF).
$${VIF}=\frac{1}{1-{R}_{k}^{2}}$$
(12)
if \({x}_{i}\) Other explanatory variables are highly multicollinear and have large values of: \({R}_{i}^{2}\) In other words, the value of VIF increases.if \({VIF}\, >\, 10\)is determined to be an explanatory variable. \(I\) Other explanatory variables may have significant cointegration problems.
panel data regression
Panel data regression typically falls into three categories: pooled regression (POOL) models, fixed effects (FE) models, and random effects (RE) models. Among these, FE models are further classified into time fixed effects (one-way FE) models, individual fixed effects (one-way FE) models, and time-individual fixed effects (two-way FE) models. The basic equations for the pooled time fixed-effects model, individual fixed-effects model, and time-individual fixed-effects model are:
$${y}_{{it}}=\alpha +{x}_{{it}}\beta +{\mu }_{{it}}\,i=1,2,\cdots N{\ rm{;}}j=1,2,\cdots T$$
(13)
$${y}_{{it}}={\lambda }_{t}+\mathop{\sum }\limits_{k=2}^{K}{{\beta }_{k}x}_ {{kit}}+{\mu }_{{it}}$$
(14)
$${y}_{{it}}={\gamma }_{t}+\mathop{\sum }\limits_{k=2}^{K}{{\beta }_{k}x}_ {{kit}}+{\mu }_{{it}}$$
(15)
$${y}_{{it}}={\lambda }_{i}+{\gamma }_{t}+\mathop{\sum }\limits_{k=2}^{K}{{\ Beta }_{k}x}_{{kit}}+{\mu }_{{it}}$$
(16)
The best model is selected based on the F test, Breusch-Pagan and Hausman tests. The F-test is used to compare the FE model and the pooled model selection, with a P value less than 0.05 indicating that the FE model is better. In the opposite case, the pooled model is used. The Breusch-Pagan test was used to compare the selection of the RE model and the pooled model, with a P value less than 0.05 indicating a superiority of the RE model. In the opposite case, the pooled model is used. The Hausman test is used to compare the FE and RE model selections.a P A value less than 0.05 means the FE model is better. In the opposite case, the RE model is used.
