**1. Introduction**

Despite a prolonged period of urbanisation and industrialisation globally, 45% of the world's people still live in rural areas [1,2]. In order to promote the reasonable development of rural areas, countries around the world try to take different measures to guide the spatial layout of rural settlements [3]. Some European countries have introduced policies such as "multi-functional agriculture" to adjust the rural layout structure so as to achieve balanced development between urban and rural areas; some Asian countries have implemented the "New Village Movement" to alleviate social conflicts and promote the development of rural areas [4–6]; However, whether in a developed or developing country, policy implementation may not necessarily achieve the expected goals [7], and the loopholes in the policy may

**Citation:** Cao, C.; Song, W. Discerning Spatiotemporal Patterns and Policy Drivers of Rural Settlement Changes from 1962 to 2020. *Land* **2022**, *11*, 1317. https:// doi.org/10.3390/land11081317

Academic Editors: Bangbang Zhang, Yongsheng Wang, Qi Wen and Dazhuan Ge

Received: 8 June 2022 Accepted: 9 August 2022 Published: 15 August 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

also lead to unbalanced development in rural areas [8,9]. For instance, some countries in South America have experienced "false urbanisation" (referring to the phenomenon of the rural population's excessive migration to cities, urbanisation that exceeds the national economic development capacity), resulting in an excessive influx of the rural population into cities, and a large number of rural settlements have been abandoned. The formation of huge "slums" has caused a series of social problems [10]. This experience shows that grasping the changing laws of rural settlements is not only conducive to the rational use of rural land in a region, but also helps to promote the coordinated development of urban and rural areas [11].

How do we correctly grasp the law of changes in rural settlements and guide their rational layout and development? First, we use remote sensing images to understand the distribution of rural settlements. For example, we found through imagery that rural settlements in the Mayo region of Yukon, Canada, are distributed along the river; the farmland in the Wilson area of Kansas in the western United States is distributed in rectangular blocks, and rural settlements are scattered around the farmland; the farmland in China's Guanzhong region surrounds rural settlements. Secondly, governments should formulate corresponding policies on rural settlements according to their own national conditions and regulate the layout of rural settlements with policies. Belgium realizes rural revitalization through organic integration of land planning and rural improvement; and the British government has focused on building central villages, thereby promoting the agglomeration of rural population to central villages. Israel has explored the hierarchical service centre model, which allows the size of rural settlements to adjust to changes in agricultural production methods. In China, there is an urgent need to adjust the layout of rural settlements.

In 2020, there were still 510 million people living in rural areas in China, accounting for 36.11% of the country's total population [12]. According to China's third land survey, the rural residential land area was 21.9356 million hectares, accounting for 62.13% of urban villages and industrial and mining land [13]. China's rural population accounts for 36.11% of the country's total population, but it occupies 62.13% of the country's construction land. In the context of new urbanisation, adjusting the layout of rural settlements is therefore still the top priority. In recent years, with the rapid development of the social economy and the promotion of related policies, the barriers to mobility among the rural population have been gradually broken down, leading to major changes in the pattern of rural settlements [14,15]. In order to cope with these changes, it is necessary to analyse the historical evolution law of rural settlements in depth and then guide their rational layout and development according to the law [16].

Accurate spatial data is the basis for studying the evolution of patterns in rural settlements. Compared with urban land use, rural settlements are smaller and relatively scattered. They are therefore less described in most existing land use maps, and there is no rural settlement type [17] in many global land use maps. For example, the land use survey classification system proposed by the US Geological Survey is divided into nine categories: urban or construction land, agricultural land, grazing land, woodland, waters, wetlands, wasteland, permafrost, and tundra. Some early studies used Landsat to extract rural settlements. The researchers combined terrestrial satellite data with public auxiliary geospatial data and used geospatial data fusion to map rural residential sites in remote areas [18]. Some researchers also used the global urban footprint (GUF) to obtain the rural training samples and used the spectral–texture–time information from the Landsat and Sentinel time series to map the rural residential population [17]. In recent years, many scholars have used SPOT, QuickBird, and other high-resolution images to extract clearer rural settlement data [3,19,20]. In order to extend the time scale of rural residential data acquisition, some scholars have used topographic maps to obtain long-term rural settlement data, but these maps provide less rural settlement information, and the shooting range is limited, so it is difficult to achieve full regional coverage [21]. There are thus still great

challenges facing research into the pattern evolution of rural settlements on the medium and long time scales.

The spatial evolution pattern of rural settlements in different areas often suggests different laws. In short time scales, the size of a rural residential area usually shows a linear trend. For example, from 2009 to 2014, the area of rural settlements in Changchun City showed a decreasing trend [11]; from 2006 to 2015, the Kangbashi New Area in Inner Mongolia had significant spatial expansion characteristics [22]; from 2000 to 2018, the scale of residential settlements in Pudong, Shanghai decreased significantly, showing a decreasing trend from the urban-rural fringe to the outer suburbs [3]; and from 1990 to 2015, the kernel density of rural settlements in Hubei Province decreased, and there were obvious regional differences [13]. The evolution of rural settlement size is also different in the medium and long time scales due to differences in the development scenarios in different regions. For example, some scholars have used historical data left by social anthropologists to analyse the evolution of rural settlements in Xin He Village, China, from 1949 to the present and found that their changes involved a process from stagnation to disorderly expansion to orderly construction [23]. Some scholars have studied the changes in rural settlements in Belarus from 1959 to 2009. Their analysis found that changes in population during different periods affected changes in the number of rural settlements. For example, intensive migration outflow was accompanied by the disappearance of a large number of rural settlements. Some scholars have studied changes to rural settlements in different areas at the same time node. Their results showed that the number of rural settlements in some areas has declined continuously in the past 50 years, while the trend of change in rural settlements in other areas is to decrease first and then increase [24,25]. The above studies show that changes in the spatial pattern of rural settlements are usually relatively simple on short time scales, while the change trends on medium- and long-term scales are often diverse. It is therefore of profound significance to study the evolution of spatial patterns in rural settlements on a long time scale to grasp the rural development in this area.

There is a close connection between the evolution pattern of rural settlements and policy reform [26], and rural policy profoundly affects changes to rural settlements [27,28]. In the second half of the 20th century, with the acceleration of globalisation and urbanisation, many countries issued policies to plan the development of rural settlements [7]. The policies of developed countries mainly focused on the centralised layout of rural settlements, the construction of infrastructure, and other aspects of the rationalisation arrangement [29]. For example, the Japanese village-building movement is characterised by excavating local resources, respecting local characteristics, and using rural resources to develop and promote rural construction. In view of the lack of rural infrastructure and other problems, the UK proposed a village revitalisation pattern focusing on the construction of central villages, and the government formulated a series of policies to promote the concentration of rural settlements in the key development areas designated by the government. In rural France and rural Brazil, agricultural modernisation policies have also caused changes in the local settlement pattern [10]. In the 1980s, residential concentration policies were implemented almost throughout Central and Eastern Europe (such as Hungary and Poland) to promote the centralised development of rural settlements [30]. Developing countries have also promulgated various policies to guide the development of rural settlements. For example, Egypt has issued policies since 1996 to encourage people to settle in the arid regions of the eastern and western desert plateaus and to avoid building new buildings in the floodplain of the Nile River [31]; China implements macropolicies such as "new rural construction" and "new urbanisation" to coordinate urban and rural development and solve problems caused by the layout of some rural settlements [32,33].

In our research, we found that some scholars used topographic maps, text data, and so on to study the changes in the scale of early rural settlements, and the scale of rural settlements showed two trends: expansion and shrinkage. For example, from the 1960s to the 1980s, the number of rural settlements in the Tongzhou District of Beijing decreased from 417 to 365, and the number of rural settlements in the Jizhou District decreased from 660 to 497 [24,25]. Ownership has greatly hindered the production of farmers, and the construction of rural settlements in Xinhe Village has stalled [23]. Rural settlements in the Jinzhong Plain of Shanxi Province have been expanding since 1979 [34].

We found that land institutional change will affect land use change. Identifying policy as one of the main drivers of land-use change and agricultural development, Teka et al. assessed land-use change in northern Ethiopia since the 1960s and found that the land policies of imperial and communist regimes largely promoted arable land. The increase in vegetative land decreases, while in the EPRDF regime, the situation is reversed [35]. Spalding et al. describe the evolution of land tenure in Panama in terms of development process and land policy in Latin America, arguing that land use policy affects land use change at the local level [36]. Munteanu et al. integrated historical maps and satellite imagery of the Carpathians region to assess the impact of nineteenth century agricultural land choices on agricultural development today. They concluded that changes in political systems can affect future land use choices [37]. Wang Juan et al. analysed the dynamics of land policy and land use change in China based on land use data. They found that land use change in China is closely related to changes in government land policy and socioeconomic development [38].

From the current point of view, China's successively implemented rural settlement policies have changed significantly over the past 60 years, and homesteads have undergone a transition from private ownership to public ownership. This paper aims to solve the following two questions: (1) Changes in the scale of rural settlements are not clear in the period before remote sensing data, so is the scale of rural settlements expanding or shrinking? (2) The homestead has undergone a transition from private ownership to public ownership, and this change is decisive, so when did changes in the scale of rural settlements become more drastic?

Changes in the pattern of rural settlements have obvious period characteristics. The analysis and study of the evolutionary characteristics of rural settlements on a medium and long time scales can provide an effective basis for the scientific and reasonable planning of rural settlements. Previous researchers have mostly analysed changes in rural settlement patterns under the influence of driving factors such as terrain, water sources, traffic, altitude, and human activities [39]. There is currently less work on systematically assessing the impact of rural settlement policies in the medium and long-term scales. We have studied changes to rural settlements in some developed and developing countries and found that there is indeed a close connection between the evolutionary pattern of rural settlements and policy reform. Based on the decryption of military satellite images, this study reveals the spatial evolution characteristics of rural settlements in Dingzhou, China from 1962 to 2020 and explores the impact of policies on rural settlements and changes in different periods. The specific purpose of this study was as follows: (1) obtain medium- and long-term historical data for rural settlements in Dingzhou City, China by decrypting military satellite remote sensing images; (2) uncover the spatial evolution characteristics of rural settlements in Dingzhou City from 1962 to 2020; (3) analyse the effect of rural settlement policies on changes in rural settlements patterns in different periods and summarise the evolutionary characteristics of the different stages of rural settlement spatial patterns.

#### **2. Overview of Study Area and Data Sources**

#### *2.1. Overview of Study Area*

Dingzhou is a county-level city (A county-level city is one of the administrative divisions in China, with the same administrative status as municipal districts, counties, and autonomous counties.) directly under the Central Government of Hebei Province, China. It is located between 38◦14 N–38◦40 N and 114◦48 E–115◦15 E (Figure 1). In 2018, Dingzhou City had jurisdiction over 25 towns (streets) and 542 villages (communities), covering an area of 1283 square kilometres. The terrain of Dingzhou is flat and slightly inclined from northwest to southeast. It has a temperate–warm temperate, semi-humid,

and semi-arid continental monsoon climate. The average annual temperature is 12.4 °C, and the interannual temperature difference is not large.

**Figure 1.** Location of Dingzhou City. Note: NCQ: Nancheng Qu Street, BCQ: Beicheng Qu Street, XCQ: Xicheng Qu Street, CAL: Chang'an Lu Street, LZ: Liuzao Town, QFD: Qingfeng Dian Town, PC: Pangcun Town, ZL:Zhuan Lu Town, MYD: Mingyue Dian Town, DND: Dingning Dian Town, DT: Dongting Town, DXZ: Daxin Zhuang Town, DW: Dongwang Town, GP: Gaopeng Town, XY: Xingyi Town, LQG: Liqin Gu Town, ZW: Ziwei Town, KY: Kaiyuan Town, DLC: Dongliu Chun Town, HTZ: Haotou Zhuang Hui Township, DLZ: Dalu Zhuang Town, XC: Xicheng Town, XZ: Xizhong Town, ZC: Zhou Cun Town, YJZ: Yangjia Zhuang Town. (Figure created in Arc GIS 10.5 ESRI, https://www.esri.com (accessed on 11 December 2021)).

Dingzhou City is an important node city in the Beijing–Tianjin–Hebei Economic Zone in Hebei Province. In 2020, the GDP of Dingzhou reached RMB 3.419 billion, an increase of 3.4% over the previous year. As of 2020, the resident population of Dingzhou is 1,095,900. Its urban population is 577,400, accounting for 52.69%, and the rural population is 518,500, accounting for 47.31%. According to the sixth national census in 2010, the urban population has increased by 102,800, and the rural population has decreased by 171,500, meaning that the proportion of urban population increased by 11.96%.

#### *2.2. Data Sources*

The data used in this study are mainly remote sensing image data, land use maps, and social and economic data for Dingzhou city. The rural settlements in 1962 and 1972 were identified from KeyHole remote sensing images [24]. KeyHole is a series of American reconnaissance satellites. They are military reconnaissance satellites with a spatial resolution of 1.5–3 m. Most of the KeyHole satellite images are concentrated between 1960 and 1980. So far, the first-generation images captured by KeyHole have been decrypted. We use the decrypted images of Dingzhou City in 1962 and 1972 to extract rural settlements. The spatial resolution of this image is 2 m. The data for rural settlements in 1990, 2000, 2010 and 2020 came from the Data Centre for Resources and Environmental Sciences, Chinese Academy of Sciences. These data are based on Landsat TM/ETM remote sensing images and the China–Brazil Earth Resources Satellite (CBERS-1), which are 30 m spatial resolution images generated by human–computer interaction [40]. The social and economic data for the per capita net income of farmers, per capita housing area of farmers, and the population of Dingzhou City were obtained from "*New Hebei 60 Years of 1949–2009*" [41].

#### **3. Research Methods**

#### *3.1. Theoretical Framework*

Maslow divided human needs from low to high into five levels: physiological needs, safety needs, social needs, esteem needs, and self-actualisation [42]. Maslow's theory of psychological needs is also applicable to a farmer's need for housing, which drives farmers to make decisions corresponding to the level of housing demand based on maximising their own interests. This kind of housing demand shows the characteristics of stages: in the first stage, farmers are at the lowest level of survival needs, and housing is needed only to meet the simplest living functions such as rest. In the second stage, with the deepening of reforms and opening up and the growth of the rural economy in China, farmers began to pursue luxurious and extravagant housing forms, far exceeding the needs and functions of normal housing. In the third stage, with the improvement of the education levels among farmers, they gave up houses that reflected a certain status and began to pursue rural houses that were comfortable. In the fourth stage, with the further improvement of educational levels, they began to demand a better quality of life. In the fifth stage, farmers change from rational needs to ideological needs, pursuing an ideal state of housing and hoping to realise self-worth.

Game theory mainly studies the interaction between incentive structures. The changing demands for farmers' housing reflects the changes in rural settlements and is the result of farmers gaming based on their own needs and external conditions. Policy is an external condition that has a strong guiding and restricting effect on land management [43], which is an important factor in farmers' decision-making. We divided the rural homestead policy into five periods according to its characteristics and trends: ownership transition, "unified planning", "paid use", "connection between increase and decrease", and "separation of three rights". The housing needs of farmers are different at different times, and the degree of—and their sensitivity to—policy feedback also varies. The gaming between farmers' needs and policy implementation directly causes changes in rural settlements. The continual development of the economy means that changes in the housing needs of farmers and in macro policy are inevitable. This has become the basic driving force, following the laws of Maslow's psychological needs theory and game theory, thereby affecting the scale of rural settlements, and forming the relationship curve of the "policy-scale of rural settlements" in different periods (Figure 2).

(I) Changes in the scale of rural settlements were relatively stable during the period of ownership transition. In theory, ownership change is a strong policy stimulus for rural settlements. However, at this time, farmers at the level of subsistence needs had low living standards and poor economic conditions, and their requirements for living space were relatively simple. There were no significant changes in rural settlements. (II) During the period of "unified planning", the scale of rural settlements changed in an inverted "U" shape. The rural economy developed rapidly after the reforms and opening up, and the basic survival needs of farmers (food, clothing, housing, and transportation) were met. With the relaxation of policies on the management of rural settlements, farmers achieved the conditions necessary to pursue superior housing (luxurious and extravagant forms of housing), leading directly to the continual expansion of the scale of rural settlements. After the housing boom in rural areas, the state and local governments issued policies in a timely manner in order to control the scale of rural settlements and made strict regulations regarding the area of homesteads. Farmers changed the form of their housing according to the requirements of the policy, reducing the scale of their housing, and effectively restrained the disorderly expansion of rural settlements. (III) The scale of rural settlements shrank during the "paid use" period, and their spatial patterns were optimised. Decision-makers took into account the fact that over-occupancy and random construction by people building multiple houses seriously affected the appearance of villages, and the multiple houses owned by a single family meant that a large amount of rural land was concentrated in the hands of a few people, which damaged the interests of other farmers. In order to solve these problems, the "one household, one house" policy was implemented. During this period, the

government also planned the layout of rural settlements and improved rural infrastructure construction. Village and town planning improved the living conditions of farmers, and the functional layout of rural areas became more reasonable. Farmers pursued a clean and comfortable living environment, tended to participate in the construction of village and town planning, and gave up scattered and complex residential forms, resulting in a reduction in the area of rural settlements. (IV) The scale of rural settlements reduced further during the period of "connecting increase and decrease". The government further explored the homestead system in order to optimise the economic and social development pattern of urban and rural areas and increased the consolidation of rural residential land through the implementation of policies. At this stage, the residential comfort needs of farmers were met, and their needs for residential scale tended to be rational as their education levels increased. The policy also incentivised the withdrawal of homesteads (the local government gave incentives or subsidies to villagers who voluntarily vacated their homesteads), which directly mobilised the enthusiasm of farmers and made them more willing to withdraw from unnecessary homesteads. (V) The scale of rural settlements gradually stabilised during the "separation of three rights" period. The government actively carried out pilot work for the reforms of the "separation of three rights" system to fully stimulate the power and vitality of the circulation of homesteads, thereby increasing the collective income from circulation, and the standard pay for withdrawing homesteads was also raised. Farmers changed from having rational needs to ideal needs at this time. They gave timely feedback regarding policy, actively cooperated with the pilot work of the reform of the homestead system, and hoped to realise their self-worth in the process of the reform of the homestead system. Due to the long-term policy regulation and the effective development of rural settlements, the scale and pattern of rural settlements was optimised to a considerable extent. At this time, the scale of rural settlements did not change much.

**Figure 2.** Theoretical framework.

#### *3.2. Collection of Rural Settlements in the Historical Period*

We deciphered the KeyHole remote sensing images from 1962 and 1972 and extracted the information about rural settlements (Figure 3). Before interpretation, an image needed to be preprocessed and compared to the land use map in 2000 for geometric correction [44,45]. The geographic coordinate system of the land use map for Dingzhou in 2000 was GCS\_Krasovsky\_1940, and coordinate correction was performed via the polynomial correction method. The specific interpretation process is shown in Figure 3, and these steps were all carried out in Arc GIS software. The spatial resolution of rural settlements was 2 m in 1962 and 1972 and 30 m in 1990, 2000, 2010, and 2020. We resampled the interpretation results so that the resolution of the data would be the same for all years, ensuring that the data processing and analysis of rural settlements were based on uniform spatial coordinates and uniform spatial resolution.

**Figure 3.** Visual interpretation process.

Since the year the images were interpreted cannot be checked in the field and there is no high-precision image data, accuracy was evaluated using expert interpretation and crowdsourcing tests [25]. We identified 300 random points corresponding to the ground class in remote sensing images in 1962 and 1972 (Figure 4) and identified random points as control points, and the final verification passed 287 (1962) and 279 (1972) random points; assessment accuracy was 95.7% and 93%, respectively.

#### *3.3. Kernel Density Estimation*

Kernel density estimation (KDE) can be used to study patch distribution density, spatial extent and intensity, and patch distribution density increases with increases in the kernel density value. This method is often used to detect spatial hotspots and identify location where high- or low-value elements cluster in space, which intuitively represents variability in the spatial density of rural settlements. The kernel density estimation is calculated using the following formula [46]:

$$f(x, y) = \frac{1}{n!\hbar^2} \sum\_{i=1}^n K(\frac{d\_i}{n}) \tag{1}$$

where *f* (*x, y*) represents the kernel density value of the point (*x, y*); *h* is the bandwidth or smoothing parameter; *K* represents the kernel function; and *di* represents the distance between the point (*x, y*) and the *i*-th observed position.

**Figure 4.** Assessment of rural residential accuracy in Dingzhou in 1962 and 1972 (Figure created in Arc GIS 10.5 ESRI, https://www.esri.com (accessed on 20 December 2021)).

#### *3.4. Spatial Change Pattern of Rural Settlements*

According to the changes of rural settlement characteristics in Dingzhou in the past 60 years and to related research, the change process of the spatial distribution of rural settlements in Dingzhou is divided into expansion pattern, merge pattern, retreated pattern, and urbanisation pattern (Figure 5) [25]. The diffusion pattern reflects the expansion of rural settlements on the original basis (Figure 5a); the merger pattern involves the merging of

two or more rural settlements (Figure 5b); the evacuation pattern involves rural settlements being transformed into other land use types (Figure 5c); and the urbanisation pattern refers to the transformation of rural settlements into urban land (Figure 5d).

**Figure 5.** Change pattern of rural settlements. (Note: (**a**): expansion pattern; (**b**): merge pattern; (**c**): retreated pattern; (**d**): urbanization pattern).
