EN中文

Journal of Geographical Sciences >

2022 , Vol. 32 >Issue 5: 873 - 891

DOI: https://doi.org/10.1007/s11442-022-1976-7

Research Articles

Classification and detection of dominant factors in geospatial patterns of traditional settlements in China

  • WU Shaolin , 1 ,
  • DI Baofeng , 2, * ,
  • Susan L. USTIN 3 ,
  • Constantine A. STAMATOPOULOS 4 ,
  • LI Jierui 2 ,
  • ZUO Qi 2 ,
  • WU Xiao 1 ,
  • AI Nanshan 1
Expand
  • 1. College of Architecture and Environment, Sichuan University, Chengdu 610065, China
  • 2. Institute for Disaster Management and Reconstruction, Sichuan University-The Hong Kong Polytechnic University, Chengdu 610207, China
  • 3. John Muir Institute of the Environment, University of California Davis, CA 95616, USA
  • 4. Stamatopoulos and Associates Co. & Hellenic Open University, Athens 11471, Greece
* Di Baofeng (1977-), PhD and Professor, specialized in disaster management and remote sensing of environment. E-mail:

Wu Shaolin (1997-), PhD Candidate, specialized in settlement environment. E-mail:

Received date: 2020-12-05

  Accepted date: 2021-08-12

  Online published: 2022-07-25

Supported by

National Key Research and Development Program of China(2020YFD1100701)

Social Science Research “14th Five-Year Plan” 2021 Project of Sichuan Province(SC21ST001)

Strategic Priority Research Program of the Chinese Academy of Sciences(XDA23090502)

Fold

Abstract

The geospatial distribution pattern in traditional Chinese settlements (TCSs) reflects the traditional harmony between humans and nature, which has been learned over centuries. However, TCSs have experienced serious disturbances by urbanization and migration. It is crucial to explore the local wisdom of geospatial patterns and dominant factors for TCSs at the national scale in China. This study sought to determine the geospatial wisdom of traditional settlements to enrich our future settlement development with the aim of establishing Chinese settlement values for modern living. Herein, a dataset of 4000 TCSs were analyzed and clustered for environmental factors that affect their geospatial patterns by machine learning algorithms. We concluded that (1) five geospatial patterns of TCSs were clustered on a national scale, and the threshold of environmental factors of TCS groups was detected. (2) Environmental conditions and settlement concepts interacted and determined the similarities and differences among TCS groups. (3) The key boundary for TCSs and the dominant factors for each zone were determined, and topographical conditions and hydrologic resources played significant roles in all five TCS zones. This study provides a better understanding of the adaptability of the environment in relation to the TCSs and aids in planning TCS conservation and rural revitalization.

Key words： rural settlements; geographical pattern; rural revitalization; traditional wisdom

Cite this article

WU Shaolin , DI Baofeng , Susan L. USTIN , Constantine A. STAMATOPOULOS , LI Jierui , ZUO Qi , WU Xiao , AI Nanshan . Classification and detection of dominant factors in geospatial patterns of traditional settlements in China[J]. Journal of Geographical Sciences, 2022 , 32(5) : 873 -891 . DOI: 10.1007/s11442-022-1976-7

1 Introduction

Rural settlements, as the basis of production and life for rural residents, are facing the challenge of transformation and revitalization under contemporary urbanization. Rapid urbanization has changed the rural lifestyle, including society, economy, environment, and culture, and has become an important a driving force for rural decline. In China, although the urbanization rate reached 60.06% in 2019 (National Bureau of Statistics of China, http://www.stats.gov.cn/), rural settlements still support more than 500 million people. Therefore, more attention is needed on rural issues, such as labor force drain (Liu et al., 2010; Siciliano, 2012; Xu et al., 2019), eco-environmental degradation (Liu et al., 2010; Tian et al., 2018), constructed land sprawl (Xi et al., 2012), and cultural decline (Agnoletti, 2014; Ye, 2020). In response, the Chinese government implemented the National Rural Revitalization Strategy. According to the strategy, rural settlements were divided into four types and were transformed in different ways: relocation for unsuited settlements, combination for suburban settlements, improvement for dense settlements, and conservation for characteristic settlements.
Traditional Chinese settlements (TCSs), which are characteristic settlements, have a long history and retain unique information about national and regional cultures requiring protection (Liu et al., 2014). One type of TCS is a community comprised of completely traditional buildings or architectural relics, and the other contains a compatible relationship with the environment. Some TCSs have both architectural and environmental advantages. For example, Sideng village, which was formed before the Yuan Dynasty (AD 1271-1368), is the only surviving ancient market on the Tea-Horse Ancient Road. Sideng village is not only the site of the World’s Most Endangered Architectural Heritage and Intangible Cultural Heritage (wood carving) but also reflects the man-land harmony concept. Sideng village is located in a sheltered area surrounded by mountains and bordered by a river, with more than 80% of vegetation coverage. The geospatial pattern of Sideng village allows the settlement self-sufficiency, which ensures that its culture can persist without outside disturbances. Similarly, the various values of other TCSs should be protected and learned, including objective values (such as architecture) and abstract values (such as concepts of location). Furthermore, the wisdom of geospatial patterns from TCSs promotes the understanding of human-land relationships and will be a good model for relocation and reconstruction in regions (Wang et al., 2021).
The objective human-land relationship is as follows. First, humans are dependent on the earth (environment), and the geographical and physical environment often reflects the regional characteristics of human activity, which restricts the scale and speed of societal development (Wu, 1991; Liu, 1998; Gu, 2000; Moghadam, 2015). Thus, most studies on settlement formation and sustainable development involve the selection and change of environmental conditions, including studies on historic sites (Jones, 2010; Guo et al., 2013; Chen et al., 2015; Xie et al., 2017; Liu et al., 2018; Cui et al., 2019) and existing settlements (Xi et al., 2018; Tian et al., 2020; Fang et al., 2020). On the other side of the human-land relationship, humans occupy the dominant position, and the environment is an object that can be recognized, used, changed, and protected (Wu 1991; Liu 1998). In the process of adapting to the environment, human physiology and psychology, national characteristics, and the social system mode also play a role in the generation and development of settlements (Liu, 1998; Seymour, 2016; Fang, 2004), and human culture and society have become a component of settlement research (Linard et al., 2012; Adger et al., 2013; Fang et al., 2014; Li et al., 2015; Wu et al., 2017).
Following the theory of the human-land relationship mentioned above, to describe the human-land relationship of TCSs and recover the wisdom provided by TCSs, the first step is to understand the environmental patterns of TCSs. Then, we discuss the mechanism of mutual adaptation between humans and the environment, which is reflected in the way of life and production, as well as ethnic and social characteristics. From a macroperspective, the environmental or geographical patterns of TCSs are symbols of profound interrelationships between humans and nature; therefore, detecting the geospatial distribution is also an indispensable part of TCS research and classification. Previous studies have generally been based on the province scale (Tong, 2014; Feng et al., 2017; Meng et al., 2018), basin scale (Yang et al., 2015; Li et al., 2020), or mountain scale (Zhou et al., 2017). From a cultural and social perspective, architecture and cultural heritage, especially ethnic culture, are the most attractive aspects of TCSs; thus, the majority of studies have focused on architectural styles or minorities and classify TCSs according to them (Liu et al., 2010; Li et al., 2018; Zhang et al., 2018; Jia et al., 2021). These studies have contributed details about building structure, ethnic totem, and other cultural genes and are critical focus fields for settlement continuation and TCS protection. However, national-scale research on the geospatial distribution and classification of TCSs has not received enough scholarly attention and needs further discourse.
The classification of TCSs based on environmental data will provide a threshold for managers to protect TCSs and to relocate unsuited settlements. However, what are the dominant factors for a rural settlement? Detecting dominant factors seems to be essential to improve the availability of environmental thresholds. While parametric statistical models, such as the analytic hierarchy process (Zhou et al., 2016) and logistic regression (Yang et al., 2019), are commonly employed for the detection of dominant factors, they are inadequate for capturing complex relationships that are difficult to specify (Liu et al., 2006). Machine learning is a sophisticated statistical approach to modeling complex relationships between various factors and samples (Di et al., 2019). Compared with the aforementioned parametric statistical models, machine learning generally showed better objectivity and accuracy.
Based on the discussion above, the geospatial patterns of TCSs, which reflect the harmonious human-land relationship of settlement and will support TCS management, have not been scholarly quantified. Therefore, the purpose of this study is to explore the geospatial distribution patterns of TCSs nationwide and further discuss the corresponding humanistic characteristics. In particular, (1) TCSs are classified by environmental factors, and the similarities and differences of TCS groups are illuminated with threshold values and provide cultural and social evidence to describe the characteristics of each TCS group. (2) The dominant factors of each TCS zone are identified by machine learning analysis. The results of this study are expected to provide support for TCS conservation and rural revitalization, especially in the field of settlement relocation.

2 Data and methods

2.1 Data

TCS data were from batches one to four of the TCS information published by the Ministry of Housing and Urban-rural Development of China, and a total of 4000 TCSs (96.31% of batches one to four of the TCS information) were identified nationwide due to their available location (Traditional Chinese Settlements Digital Museum, http://www.dmctv.cn/prevue.aspx). There are more TCSs (population) in the southeast than in the northwest (Figure 1). A total of 95.00% of the TCSs emerged before the beginning of the 20th century, and approximately one-third of the TCSs were formed during the Yuan Dynasty (AD 1271-1368). A total of 3.83% of the TCSs did not have a known formation time, while 47% of the TCSs were formed in the last 70 years.
Figure 1 Study area and distribution of TCS
Geospatial patterns of the TCSs reflect environmental changes, cultural succession, and the relationships among human needs, activities, and natural resources. Clustering TCSs with quantifiable natural factors and continuous values is the first step to fully describe the man-land relationship in TCSs. Indeed, current studies suggest that topography, meteorology, and water are the primary environmental factors affecting settlements (Hill, 2003; Kirch et al., 2004; Tian et al., 2012; Guo et al., 2013; Fang and Jawitz, 2019; Xu et al., 2019; Fang et al., 2020). Thus, three groups of factors were considered: topography, meteorology, and natural resources. Topographical factors include elevation (ELE) and slope (SLOPE). Topographical factors are most easily noticed in the settlement distribution. The separation and connection between settlements often depends on topographical conditions, and the degree of information transmission and integration of settlements affects the degree of similarity of the humanistic characteristics of settlements (Song and Zhang, 2017; Tambassi, 2018). ELE and SLOPE not only determine the lifestyle and production mode (such as farming or grazing) but also affect construction and geoenvironmental security. In mountainous areas, topographical conditions become one of the controlling factors in settlement locations (Potosyan, 2017). Meteorological factors include temperature (TEM), precipitation (PRE), air pressure (APR), pan evaporation (EVP), sunshine duration (SSD), and relative humidity (RHU). Meteorological factors are important conditions in the growth of crops and are related to farm animal health, human health, and residential suitability (Hu et al., 2019; Seymour, 2016). The primary consideration for inhabitants is food satisfaction, and the TEM, PRE, EVP, and SSD groups determined the type and growth season of crops, which selected, to some extent, the style of production and social characteristics (McLeman and Smit, 2006; Hemming et al., 2008; Hu et al., 2011; Che et al., 2014; Cui et al., 2019; Fletcher et al., 2020). In addition, other climatic factors have an impact on comfort level and human heath, such as TEM, APR, and RHU (Tsutsumi et al., 2007; Davis et al., 2016; Yang and Matzarakis, 2019; Raymond et al., 2020). Natural resources include the normalized difference vegetation index (NDVI), horizontal distance to the closest river (HDR), soil bulk density (SBD), soil organic carbon content (SOC), and soil electrical conductivity (SEC). Sustainable materials from the surroundings ensure settlement development, and the selection of settlement location reflects resource dependence (Zhou et al., 2013). Among the natural resources, water resources are fundamental to sustain life. Vegetation contributes to improving soil quality, water conservation, ecosystem support (including the prevention of erosion from winds and water), and disaster prevention (e.g., landslides, debris flows) (Jones, 2010), and vegetation resources are measured by NDVI. Rivers provide conditions (source for transportation, communication and trade, fish and other food resources, water for human and livestock consumption, and for growing crops) for the emergence of settlements, but potential flood risk should also be considered (Alemu, 2016; González-Arqueros et al., 2018; Peng et al., 2020). Soil-related factors (SBD, SOC, and SEC) are used to indicate soil fertility, and fertile land can efficiently support communities with high-quality food and energy. Table 1 shows the description and source of each factor for the study units.
Table 1 Data description and data source
Types Factors Unit Description Data resource
Topographical factors ELE m Digital Elevation Model (DEM), 30m resolution basic topographical map of China SRTM (shuttle radar topography mission)
SLOPE ° Raster data of slope Slope conversion from DEM data (by
ArcGIS platform)
Meteorological factors TEM ℃ Average annual temperature,
5-year average of meteorological station data (2013-2017)		 National Meteorological Information Center (http://data.cma.cn)
EVP mm Average annual evaporation,
5-year average of meteorological station data (2013-2017)		 National Meteorological Information Center (http://data.cma.cn)
PRE mm Average annual precipitation,
5-year average of meteorological station data (2013-2017)		 National Meteorological Information Center (http://data.cma.cn)
APR 100Pa Average annual atmospheric pressure, 5-year average of meteorological station data (2013-2017) National Meteorological Information Center (http://data.cma.cn)
RHU % Average annual air humidity,
5-year average of meteorological station data (2013-2017)		 National Meteorological Information Center (http://data.cma.cn)
SSD h Average annual sunshine duration,
5-year average of meteorological station data (2013-2017)		 National Meteorological Information Center (http://data.cma.cn)
Natural
resources		 NDVI - Average annual Normalized Vegetation Index (2018) Resource and Environment Data Center (Xu, 2018)
HDR m Vector data of river in the study area
(distance from river in meters)		 Resource and Environment Data Center
S_BD g·cm‒3 Soil Bulk Density Cold and Dry Area Scientific Data Center (http://westdc.westgis.ac.cn/)
S_OC %(weight) Soil Organic Carbon Content Cold and Dry Area Scientific Data Center (http://westdc.westgis.ac.cn/)
S_EC dS·m‒1 Soil Conductivity Cold and Dry Area Scientific Data Center (http://westdc.westgis.ac.cn/)

2.2 Methods

Since the geospatial pattern of TCSs has not been identified at the national scale, clustering algorithms were applied to classify TCSs through natural factors. Clustering algorithms divide samples without labels, and such unsupervised classifiers allow us to detect the similarities and differences of the data itself better than artificial classifications. Then, humanistic factors, including production style, architectural features, and ethnic differences, were discussed for each group. The importance of factors was obtained by the machine learning algorithm random forest (RF) due to its excellent performance in multidimensional datasets (Breiman, 2001; Conrad et al., 2015; Zhang et al., 2018; Xu et al., 2019).

2.2.1 Clustering methods

Cluster analysis is a technique for grouping similar observations, data types, or feature vectors in an unsupervised manner according to their similar characteristics (Jain et al., 1999). Cluster analysis aims to identify and classify groups of similar objects (Govender and Sivakumar, 2020). Clustering is one of the most useful tasks in data mining for identifying interesting distributions and patterns in the underlying data and for data compression (Halkidi, 2001).
There is no clear guideline for choosing the best clustering algorithms in various datasets. Following previous studies, two clustering algorithms were selected: K-Means and hierarchical clustering. The K-Means clustering algorithm is one of the most widely used clustering methods and was proposed more than 60 years ago by Steinhaus (1956). Hierarchical clustering attempts to form a branching, tree-like clustering structure based on datasets at different levels (Gauch and Whittaker, 1981) and has applications in a variety of fields, such as biology, environment, and sociology. However, how is the optimal clustering chosen? A common way is to apply clustering methods for different numbers of groups and then use a quality index to evaluate the clustering performance (Brentan et al., 2018). The silhouette coefficient (SC) (Rousseeuw, 1987) is a common evaluation index of clustering quality. SC values range from -1 to 1, indicating not well-clustered and well-clustered observations, respectively.

2.2.2 Random forest

RF is a machine learning algorithm based on decision trees that generates multiple classification trees through a random extraction of variables, summarizes the results, and outputs the importance of each factor. Since the RF is a supervised classification algorithm, it obtains both positive and negative samples to learn the sample characteristics and finally achieves classification prediction. In this study, positive samples were TCSs and negative samples were the other settlements (non-TCSs). To ensure the feasibility of the results, non-TCSs were from the same Thiessen polygon corresponding to one TCS because the environmental conditions within one Thiessen polygon are roughly similar (Li et al., 2019). ROC (receiver operating characteristic) is often used to evaluate a classifier. In this study, the greater the ROC is, the better the classification of traditional and non-TCSs is. The absence of dominant factors led to a dramatic decrease in ROC; thus, dominant factors were shifted by a factor reduction experiment (the factors with high importance were removed one by one in order of factor importance and re-entered the classifier).

3 Results and discussion

3.1 Classification

3.1.1 Spatial distribution of TCS groups

A total of 4000 TCSs were classified by the K-Means and hierarchy algorithms with a maximum number of 500 iterations and were compared in terms of clustering quality using the SC value (Table 2). Overall, the K-Means method performed better than the hierarchy. The maximum SC value appeared in the K-Means method and corresponded to classification number 5. Therefore, the K-Means algorithm with 5 clusters was chosen to classify TCSs.
Table 2 SC value of clustering algorithm
Clustering number 2 3 4 5 6 7 8
K-Means 0.3113 0.2805 0.2987 0.3119 0.2771 0.2619 0.2751
Hierarchy 0.2935 0.2653 0.2877 0.3002 0.3105 0.2575 0.2335
TCSs were clustered into five groups, and each group had an aggregation center (Figure 2). There are two obvious normal boundaries among the TCS groups: the boundary of the Tibetan Plateau and the Qinling-Huaihe Line. The Tibetan Plateau is a specific geographical unit with unique environmental conditions. The Qinling-Huaihe Line is the dividing line of northern China and southern China. GROUP 5 is located on the inner Tibetan Plateau, and GROUP 1 is separated by the two boundaries. Most GROUP 1 settlements are sparsely located in northern China. GROUP 1 accounts for less than 20% (756, 18.90%) of the total TCSs but covers nearly one-third of China’s land area, thereby indicating its scattered distribution overall and the similar geospatial pattern of northern China. The most concentrated area of GROUP 1 is the Yellow River Basin, especially in Shanxi Province. GROUP 2 has the largest number of TCSs (1420, 35.50%) among the five groups and is mainly distributed in southeastern China, especially in the Zhejiang and Fujian provinces. GROUP 3 settlements are mainly located in Guizhou, Chongqing, Hunan, and Sichuan and are the second largest group of TCSs (1063, 26.58%). Many settlements of GROUP 3 are located in the border area between Guizhou and Hunan. Most of the TCSs in Yunnan Province are classified as GROUP 4. In this area, the TCS distribution is dense, with 518 settlements accounting for 12.95% of the total. GROUP 5 accounts for the smallest proportion (243, 6.08%) among the five groups and is sparsely distributed. The geospatial patterns of the same group of TCSs are similar, but different groups have different geospatial patterns, which can be intuitively perceived from the example photographs (Figure 2). The similarities and differences among the TCS groups will be discussed in the following section in combination with the statistical data.
Figure 2 Five groups of TCS and sample photos of each group. (The legend of sample photos includes the name, main ethnic residents and province of sample TCS. Photo source: Traditional Chinese Settlements Digital Museum, http://www.dmctv.cn/prevue.aspx)

3.1.2 Similarities and differences of TCS groups

To detect the similarities and differences among the TCS groups, a K-W nonparametric test (Figure 3a) was applied to compare the TCS groups. A box plot consists of a box body (1/4 to 3/4 positions), a median line, a mean value, whiskers, and endpoints and illuminates competing statistical data; thus, a box plot was applied to show specific information of environmental factors. In general, P values of the K-W test less than 0.05 were considered to indicate a confidence level greater than 95% with a significant difference. Figure 3a illustrates that the environmental factors in both pairs, GROUP 1 - GROUP 3 and GROUP 1 - GROUP 5 do not have significant differences. In particular, GROUP 1 and GROUP 3 have no significant differences in topographical conditions (ELE, PRE and HDR). GROUP 1 and GROUP 5 show great similarities in the resource factors NDVI, SBD, and SEC. Compared with other factors, NDVI was similar in all categories except for GROUP 5. The box plot shows the comprehensive comparison among (Figure 3b) TCS groups, as well as the threshold of environmental factors on TCS groups. Figure 3b shows the statistics for each environmental factor for the TCS groups. The synthesis of multiple factors identifies the unique characteristics of types of TCS, which are different not only in their environmental characteristics but also in their socioeconomic structures under the combined action of environmental conditions and history.
Figure 3 Similarities and differences of the environmental factors on TCS groups
(a) Significance test results of environmental factors for group pairs
(b) Value of environment factors per group (Variable names are identified in Table 1. SEC of GROUP 1, GROUP 2, and GROUP 5 are concentrated at 0.1)
Given the topographical, meteorological, and natural resource implications, there are representative cultural features and local lifestyles in every TCS group. Therefore, it is necessary to integrate natural and social sciences to fully understand the sustainable human-land relationship reflected by TCSs.
GROUP 1 is sparsely distributed in a large region of northern China, but the environmental conditions of GROUP 1 fluctuate less than in other groups, as indicated in Figure 3b, where the box diagrams are generally short. There are various structural styles among GROUP 1, such as typical adobe houses in the northwest region, the yurts of the nomadic people in Inner Mongolia, and small windowed dwellings in the northeastern region (Figure 2); however, all of them adapt to the dry and cold environmental conditions in northern China. The most concentrated distribution area of GROUP 1 is Shanxi Province. Archaeological research shows that southern Shaanxi was the origin of Lantian man, a subspecies of Homo erectus and a hominid precursor of modern humans, and is one of the birthplaces of the Chinese nation. Historically, many dynasties have set their capital in the Yellow River Basin; thus, China’s major economic and cultural centers are also located there relative to the historical period (Zou, 2006). The powerful political and economic status of the area allowed settlements to develop over a long time. The regional economy of Jiangsu and Zhejiang has gradually developed since the Tang Dynasty (AD 816-907) (Ye and Ma, 1990), and the economic center has moved to the lower reaches of the Yangtze River. To date, the lower reach of the Yangtze River is still one of the most internationally influential economic zones in China. Moreover, abundant water and heat resources, as well as flat land, have attracted a large number of people to engage in agriculture (Li, 1997). A dense population and strong economy made the buildings of GROUP 2 smaller but with more elaborate craftsmanship, such as garden-style buildings linking inside and outside activities (Figure 2), which reflect generally mild and warm weather (Xiong et al., 2020).
Through the ethnic identification of TCSs (ethnic data source: Traditional Chinese Settlements Digital Museum, http://www.dmctv.cn/prevue.aspx), it was found that the proportion of Han settlements in GROUP 1 and GROUP 2 was greater than that in the other three groups. GROUP 3 is clumped in the border area between Guizhou and Hunan, and this region is a concentrated zone of the Kam-Tai and Miao-Yao languages and the Han language. A number of GROUP 3 settlements are minority settlements such as Tujia, Miao, Zhuang, and Dong. In fact, these minorities have jointly developed a unique culture of farming by taking advantage of local soil and climatic conditions. As shown in Figure 3b, the humid climate and mountainous terrain have promoted a special farming culture with paddy-field-fish-culture and terraces for hillslope agriculture, which is usually rotated twice or three times a year. Rice and maize were their traditional crops. In particular, Zhuang was among the first to cultivate rice, and their rice culture is well developed. Chinese Baijiu, which is made from agricultural products such as glutinous rice and sorghum, became an important aspect of their farming culture and national traditions (Bao, 2006). Chinese Baijiu was used for sacrifice and was used for daily drinking to eliminate fatigue. This Chinese Baijiu culture is not represented in other TCS groups.
GROUP 4 is located in Yunnan Province with high ELE (but lower than that of GROUP 5) with less extreme climatic conditions than those of GROUP 5 and a wider distribution range of resources (Figure 3). Compared with GROUP3, the ELE and SLOPE of GROUP 4 have larger spans, which means that the vertical band of the environment in GROUP 4 is more varied. In addition, the Hengduan Mountains running north to south caused little possibilities for connectivity and communication in Yunnan (Chow, 2005; Wang et al., 2014). Physical isolation due to difficulty traveling from one settlement to another across mountains and the rapidly changing environment in the vertical direction may be the reasons why multiethnic cultures came into being in this region, such as Yi, Bai, Hani, Wa, Dai, and many other ethnic groups that have lived in Yunnan for a long time. In contrast to GROUP 3, there are various cultures among the minorities of GROUP 4. The ancient east-west and north-south minority migrations integrated into Yunnan, who are a people with a wide range of agricultural heritage systems such as forestry and animal husbandry (Shiro et al., 2007). Intermediate environmental conditions and migration resulted not only in the diversity of ethnic cultures but also in diverse agricultural production. Thus, a variety of production modes existed simultaneously in GROUP 4, such as rice terraces and rubber forestry. To avoid soil erosion caused by steep slopes (Figure 3), the Yi and Hani peoples cultivated their fields like stairs, forming a unique terraced landscape. The high ELE and PRE (Figure 3) allowed the healthy growth of rubber, and some Dai TCSs make a living from rubber. Although the TCSs of Yunnan contain minority settlements, the environmental conditions of the dwelling district (usually valleys between two mountains) are relatively similar; thus, the clustering algorithm classified most of the TCSs of Yunnan as GROUP 4. One example is the inclined roofs of houses in almost all of GROUP 4. Depending on the PRE and the surrounding plants, people use green tiles (usually Bai and Yi TCSs), thatches (usually Wa TCSs), and bamboo (usually Dai TCSs) to build inclined roofs (Figure 2).
GROUP 5’s climatic conditions are distinctly different from those of the other groups, as shown in Figure 3. This region is home to Tibetans in China and to the Tibeto-Buran language. GROUP 5 lives in areas with extreme topographical, climatic, and natural resource conditions. High ELE and low TEM limit the growth of crops, which means that each GROUP 5 settlement requires resources that cover a larger area. Therefore, GROUP 5 settlements are scattered and built from windproof stone in order to have sufficient resources and escape cold winds. Some of them are seminomadic, i.e., traveling between grazing sites (at different elevations or aspects) for animals and their own food, medical, and shelter resources. Influenced by Tibetan Buddhism, GROUP 5 has a strong sense of protecting biological resources, especially animals (Huang et al., 2020). For example, yaks provide meat, milk, fur, and fuel (usually the excrement) and have also become an indispensable means of transportation for nomads. Additionally, yaks also became an important totem and spiritual carrier in Tibetan culture. A yak horn totem is used to pray for health and good fortune, and yaks are considered to play a key role in some sacrificial activities.

3.2 TCS zones and dominant environmental factors

3.2.1 TCS zones

The most important purpose of classification is to develop targeted protection measures for different TCS groups. The protection and development of TCSs often requires the support of detailed planning from local areas at county or even village scale. According to the sampling area demarcated by Thiessen polygon, the relative similarity of environmental factors can be proven. Thus, the Thiessen polygons of similar TCS groups were assembled as one TCS zone, and there were five TCS zones in total (Figure 4), corresponding to the five TCS groups. Furthermore, among the 13 environmental factors (Table 1), the importance of environmental factors in each TCS zone was identified by the RF model. Finally, the factors with the highest performance were chosen as dominant factors by the ROC validation method in each zone. It was found that SLOPE and HDR are both important for geospatial distribution in the five zones. Besides SLOPE and HDR, the remaining dominant factors were used to name the five zones: EES, POR, TP, PN, and NE (Figure 4).
Figure 4 Map showing dominant factors per TCS zone. See Table 1 for list of factors.
Traditional zoning schemes provide support for classification management and human settlement planning. From a physiographical perspective, there are many schemes of physiographical zones, and they have different levels of focus and detail, such as topography and climatic zones (Huang, 1959; Fu et al., 2001; Gao et al., 2010; Peng et al., 2018). However, there are three boundary lines that are common to most of the various zoning schemes: the Tibetan Plateau boundary, the Qinling-Huaihe Line, and the 400-mm isohyet line. Similarly, the Tibetan Plateau boundary and the Qinling-Huaihe Line also became the boundaries between EES, NE, and the two southern zones, namely, POR and TP. The 400-mm isohyet line was not represented in the TCS zoning scheme. A possible reason for this is that GROUP 1 was concentrated in northern China, and the bias represents the environmental characteristics of the whole northern region. TCS zones also allowed finer division in the southern region because of the consideration of resource factors such as rivers and NDVI.
From a humanistic perspective, Liu et al. (2010) proposed that the typical landscape characteristics of traditional settlements were divided into three large-scale landscape regions, 14 landscape regions, and 76 landscape subregions. Liu’s team first divided the settlement landscape into three major zones based on climate characteristics (humid, semihumid, or plateau climate) and then further divided and named the settlement zones based on topographical conditions, architectural structure, cultural genes, and other factors, which are a comprehensive classification system. Due to the strong complement of architectural landscape, Liu’s classification system not only shows the natural conditions but also carries on a more adequate classification of human settlement environments. Starting from the limitation and threshold of environmental conditions, TCS zones offered the dominant environmental factors and paid attention to the human differences caused by the environment to avoid losing the human characteristics of settlements from the reference of environmental conditions. In actual traditional settlement protection and rural development, planners not only need to guide the sustainable development of settlements from the positive side but also to protect and inherit the traditional landscape of settlements. However, they also need to rely on the threshold of dominant factors to comply with the environmental restrictions of settlement development to avoid the long-term disharmonious development of man-land relationships.
The human-land relationship in settlements is complex, which is reflected in the diversity and integration of natural and humanistic factors. This classification and zoning study of TCSs only adopted natural factors for analysis, which was insufficient. Settlement space is 2-D or even 3-D data, including settlement form, architectural structure, and interaction between living and production space. Therefore, in future research, it is necessary to improve the data richness of the settlement itself and expand point data into polygon data. Environmental factors play a restrictive role in settlements and their evolution process.

3.2.2 Dominant factors

SLOPE and HDR are the common dominant factors in all zones. SLOPE is important for settlement construction and farming. Flat terrain ensures the safety of buildings and simplifies construction, agricultural soil fertility, and erosion protection, and in areas with smaller slopes, road extensions lead to an expanded number of settlements, which will be larger and more populated. Mountainous areas, such as terraced fields in PN, use stepped buildings and fields to reduce soil instability on farms. HDR represents surface water resources in every respect, from precipitation, rivers, and groundwater runoff to water storage in lakes, dams, and wetlands. HDR gives the cost distance between mountain settlements and river valleys. River valleys or riparian areas provide habitable and cultivated plains and are associated with milder climates, especially in mountainous and arid areas. Moreover, settlements are more likely to be built in areas sufficiently close to rivers to obtain more resources but far enough away from the river to avoid flooding disasters, and these risks are reflected in the HDR value.
Moreover, the other dominant factors supply the designated emphasis in different zones (Figure 4). ELE was more important in the northwest region (EES and NE) than in the southeast (POR, TP, and PN). Housing security and transportation conditions are controlled by ELE (Bi et al., 2010). The economic production and life of the settlements in the northwest region are affected by poor rain-heat conditions and are sensitive to ELE values, while settlements in the southeast are not. Water is a survival factor for EES. Water comes from surface water and rainfall, but EES are also more sensitive to EVP because of less precipitation compared to milder climates. Intense solar radiation affects the evapotranspiration of water resources, which is represented by EVP and SSD. The long-term adaptive evolution of these settlements has formed a sustainable systematic relationship between settlements and water resources. Therefore, the location of areas for storing and retaining water is key to settlements in EES when the water source is relatively fixed or highly seasonal (Omer et al., 2020).
Another dominant factor in the NE region is NDVI, which is a typical factor reflecting vegetation conditions. Animistic religions by minority inhabitants of the NE region have a positive impact on the environment, such as the indirect protection and display of plants and animals. Vegetation contributes to the microclimate regulation of rural settlements (Liu et al., 2019). The environmentally fragile NE regions are more dependent on natural resources such as water in light of climate change (Fang et al., 2016). At the same time, the Grain for Green Project in the NE region, as the source region of several major rivers in China, has also contributed to NDVI sensitivity. For the same reason, NDVI is also the dominant factor in the PN region. Adequate rainfall, comfortable temperatures, and fertile soils are all requirements for agricultural development. Both POR and PN are regions in southeastern China with developed agricultural cultures (Ye and Ma, 1990). Despite rapid industrialization and urbanization in the POR region, nearly half the population still lives in rural areas (Long et al., 2009). It should be noted that the loss of cultivated land in the POR region has been studied extensively by scholars (Yang and Li, 2000; Liu et al., 2010; Su et al., 2011). This study also verifies the sensitivity of soil fertility and climate in this region. This suggests that the southeast region of China is more dependent on farming and that agriculture-related environmental conditions must be an important consideration there.
The sustainable development of settlements is not the result of a single factor but of the interaction and counteraction with the full environmental system. As described in the classification process, there are similarities between some types of rural settlements, but these similarities are not identical. What this paper hopes to do is to provide a reference with a range of crucial environmental factors for each of the TCS groups, which will assist planners of specific rural settlement groups in making decisions more efficiently and in a more consistent manner based on scientifically determined criteria. Human selection and transformation of the environment is another part of the man-land relationship. Human social psychological cognition, physiological senses, economy, culture, ethnicity, religion, and other humanistic factors are the key factors for settlement planning. Therefore, a comprehensive clustering and analysis model combining humanistic factors and natural factors is urgently needed to support TCS protection work and the construction of future settlements.

4 ConclusionAcknowledgements

The rural development and characteristics of TCSs are diverse and complex. They vary considerably in both their natural environment and human society. In this study, the TCSs were clustered into five groups at the national scale. The boundary of the Tibetan Plateau and the Qinling-Huaihe Line were detected as China’s important geographical boundaries for different TCS groups. The conception of TCSs is manifested in the clustering and distribution of settlements, architectural style, ethnic characteristics, and cultural similarities and differences. The dominant factors of the natural environment for five TCS zones were detected to provide guidance for local TCS protection and new settlement development. All groups were found to be sensitive to topographical conditions and hydrologic resources. This classification analysis and dominant factor exploration will support the management and projection of TCSs and promote rural revitalization in China.
The authors would like to thank Dr. Shuo Shi, Mr. Ke Xiong, and Ms. Ya’nan Duan from Sichuan University, Dr. Guojie Chen and Dr. Yucheng He from Institute of Mountain Hazards and Environment, CAS for professional advice.

References

Publishing order | Descend order by publishing year | Descend order by cited within
[1]
Adger W N, Barnett J, Brown K et al., 2013. Cultural dimensions of climate change impacts and adaptation. Nature Climate Change, 3(2): 112-117.

DOI

[2]
Agnoletti M, 2014. Rural landscape, nature conservation and culture: Some notes on research trends and management approaches from a (southern) European perspective. Landscape and Urban Planning, 126: 66-73.

DOI

[3]
Alemu M M, 2016. Sustainable land management. Journal of Environmental Protection, 7(4): 502-506.

DOI

[4]
Bao Y, 2006. A preliminary study on the diet culture of Miao & Dong nationality in southeast Guizhou Province. Journal of Southeast Guizhou Vocational College of Technology for Nationalities (Comprehensive Edition), 6(2): 36-37, 62. (in Chinese)

[5]
Bi X L, Wang H, Ge J P, 2010. The changes of densities and patterns of roads and rural buildings: A case study on Dongzhi Yuan of the Loess Plateau, China. Environmental Monitoring and Assessment, 164(1-4): 549-560.

DOI

[6]
Breiman L, 2001. Random forests. Machine Learning, 45(1): 5-32.

DOI

[7]
Brentan B, Meirelles G, Luvizotto E et al., 2018. Hybrid SOM+ k-Means clustering to improve planning, operation and management in water distribution systems. Environmental Modelling & Software, 106: 77-88.

[8]
Che M L, Chen B Z, Innes J L et al., 2014. Spatial and temporal variations in the end date of the vegetation growing season throughout the Qinghai-Tibetan Plateau from 1982 to 2011. Agricultural and Forest Meteorology, 189/190: 81-90.

DOI

[9]
Chen F H, Dong G H, Zhang D J et al., 2015. Agriculture facilitated permanent human occupation of the Tibetan Plateau after 3600 B.P. Science, 347(6219): 248-250.

DOI PMID

[10]
Chow C S, 2005. Cultural diversity and tourism development in Yunnan Province, China. Geography, 90: 294-303.

DOI

[11]
Conrad C, Rudloff M, Abdullaev I et al., 2015. Measuring rural settlement expansion in Uzbekistan using remote sensing to support spatial planning. Applied Geography, 62: 29-43.

DOI

[12]
Cui J X, Chang H, Burr G S et al., 2019. Climatic change and the rise of the Manchu from Northeast China during AD 1600-1650. Climatic Change, 156(3): 405-423.

DOI

[13]
Davis R E, McGregor G R, Enfield K B, 2016. Humidity: A review and primer on atmospheric moisture and human health. Environmental Research, 144(Pt A): 106-116.

[14]
Di B F, Zhang H Y, Liu Y Y et al., 2019. Assessing susceptibility of debris flow in Southwest China using gradient boosting machine. Scientific Reports, 9(1): 12532.

DOI

[15]
Fang C L, 2004. Recent progress of studies on man-land relationship and its prospects in China. Acta Geographica Sinica, 59(Suppl.): 21-32. (in Chinese)

[16]
Fang F, Ma L B, Fan H et al., 2020. The spatial differentiation of quality of rural life based on natural controlling factors: A case study of Gansu Province, China. Journal of Environmental Management, 264: 110439.

DOI

[17]
Fang Y, Jawitz J W, 2019. The evolution of human population distance to water in the USA from 1790 to 2010. Nature Communications, 10(1): 430.

DOI PMID

[18]
Fang Y P, Fan J, Shen M Y et al., 2014. Sensitivity of livelihood strategy to livelihood capital in mountain areas: Empirical analysis based on different settlements in the upper reaches of the Minjiang River, China. Ecological Indicators, 38: 225-235.

DOI

[19]
Fang Y P, Zhao C, Rasul G et al., 2016. Rural household vulnerability and strategies for improvement: An empirical analysis based on time series. Habitat International, 53: 254-264.

DOI

[20]
Feng Y F, Yu W Y, Lei R L, 2017. Spatial distribution features and controlling factors of traditional villages in Guangdong Province. Scientia Geographica Sinica, 37(2): 236-243. (in Chinese)

[21]
Fletcher A L, Chen C, Ota N et al., 2020. Has historic climate change affected the spatial distribution of water-limited wheat yield across Western Australia? Climatic Change, 159(3): 347-364.

DOI

[22]
Fu B J, Liu G H, Chen L X et al., 2001. Scheme of ecological regionalization in China. Acta Ecologica Sinica, 21(1): 1-6. (in Chinese)

[23]
Gao H B, Huang J, Li S C et al., 2010. The new progresses and development trends in the research of physio-geographical regionalization in China. Progress in Geography, 29(11): 1400-1407. (in Chinese)

[24]
Gauch H G, Whittaker R H, 1981. Hierarchical classification of community data. The Journal of Ecology, 69(2): 537-557.

DOI

[25]
González-Arqueros M L, Mendoza M E, Bocco G et al., 2018. Flood susceptibility in rural settlements in remote zones: The case of a mountainous basin in the Sierra-Costa region of Michoacán, Mexico. Journal of Environmental Management, 223: 685-693.

DOI PMID

[26]
Govender P, Sivakumar V, 2020. Application of k-means and hierarchical clustering techniques for analysis of air pollution: A review (1980-2019). Atmospheric Pollution Research, 11(1): 40-56.

DOI

[27]
Gu N Z, 2000. Geography environment and culture: Also discusses the methodology of geographical environmental determinism research. Zhejiang Social Sciences, (3): 133-140. (in Chinese)

[28]
Guo Y Y, Mo D W, Mao L J et al., 2013. Settlement distribution and its relationship with environmental changes from the Neolithic to Shang-Zhou dynasties in northern Shandong, China. Journal of Geographical Sciences, 23(4): 679-694.

DOI

[29]
Halkidi M, 2001. On clustering validation techniques. Journal of Intelligent Information Systems, 17(2/3): 107-145.

DOI

[30]
Hemming S, Dueck T, Janse J et al., 2008. The effect of diffuse light on crops. Acta Horticulturae, (801): 1293-1300.

[31]
Hill M, 2003. Rural Settlement and the Urban Impact on the Countryside. London: Hodder ＆Stoughton.

[32]
Hu K J, Guo Y M, Yang X C et al., 2019. Temperature variability and mortality in rural and urban areas in Zhejiang province, China: An application of a spatiotemporal index. Science of the Total Environment, 647: 1044-1051.

DOI

[33]
Hu M Q, Mao F, Sun H et al., 2011. Study of normalized difference vegetation index variation and its correlation with climate factors in the Three-River-Source region. International Journal of Applied Earth Observation and Geoinformation, 13(1): 24-33.

DOI

[34]
Huang B W, 1959. Draft of comprehensive physical regionalization of China. Chinese Science Bulletin, (18): 594-602. (in Chinese)

[35]
Huang L, Tian L J, Zhou L H et al., 2020. Local cultural beliefs and practices promote conservation of large old trees in an ethnic minority region in southwestern China. Urban Forestry & Urban Greening, 49: 126584.

[36]
Jain A K, Murty M N, Flynn P J, 1999. Data clustering: A review. ACM Computing Surveys, 31(3): 264-323.

DOI

[37]
Jia Z Y, Meng C Y, Zhou Z X, 2021. A 3-D morphological approach on spatial form and cultural identity of ethnic mountain settlements: Case from Guizhou, China. Journal of Mountain Science, 18(5): 1144-1158.

DOI

[38]
Jones E E, 2010. An analysis of factors influencing sixteenth and seventeenth century Haudenosaunee (Iroquois) settlement locations. Journal of Anthropological Archaeology, 29(1): 1-14.

DOI

[39]
Kirch P V, Hartshorn A S, Chadwick O A et al., 2004. Environment, agriculture, and settlement patterns in a marginal Polynesian landscape. Proceedings of the National Academy of Sciences, 101(26): 9936-9941.

DOI

[40]
Li B Q, 1997. Civilization process of Changjiang valley. Archaeology and Cultural Relics, (4): 12-18+84. (in Chinese)

[41]
Li G Y, Jiang C H, Du J et al., 2020. Spatial differentiation characteristics of internal ecological land structure in rural settlements and its response to natural and socio-economic conditions in the Central Plains, China. Science of the Total Environment, 709: 135932.

DOI

[42]
Li G Y, Jiang G H, Jiang C H et al., 2018. Differentiation of spatial morphology of rural settlements from an ethnic cultural perspective on the Northeast Tibetan Plateau, China. Habitat International, 79: 1-9.

DOI

[43]
Li J R, Zhang H H, Luo Y Z et al., 2019. Stepwise genetic algorithm for adaptive management: Application to air quality monitoring network optimization. Atmospheric Environment, 215: 116894.

DOI

[44]
Li Y R, Long H L, Liu Y S, 2015. Spatio-temporal pattern of China’s rural development: A rurality index perspective. Journal of Rural Studies, 38: 12-26.

DOI

[45]
Linard C, Gilbert M, Snow R W et al., 2012. Population distribution, settlement patterns and accessibility across Africa in 2010. PLOS ONE, 7(2): e31743.

DOI

[46]
Liu D J, Hu J, Chen J Z et al., 2014. The study of spatial distribution pattern of traditional villages in China. China Population, Resources and Environment, 24(4): 157-162. (in Chinese)

[47]
Liu P L, Liu C L, Deng Y Y et al., 2010. Landscape division of traditional settlement and effect elements of landscape gene in China. Acta Geographica Sinica, 65(12): 1496-1506. (in Chinese)

[48]
Liu Q, Li G, Kong D Y et al., 2018. Climate, disasters, wars and the collapse of the Ming Dynasty. Environmental Earth Sciences, 77(2): 44.

DOI

[49]
Liu Q, Peng P H, Wang Y K et al., 2019. Microclimate regulation efficiency of the rural homegarden agroforestry system in the Western Sichuan Plain, China. Journal of Mountain Science, 16(3): 516-528.

DOI

[50]
Liu S J, 1998. Academician Wu Chuanjun’s human geographical thoughts and man-nature relationship system theory. Progress in Geography, 17(1): 12-18. (in Chinese)

[51]
Liu Y, Guo H C, Zou R et al., 2006. Neural network modeling for regional hazard assessment of debris flow in Lake Qionghai Watershed, China. Environmental Geology, 49(7): 968-976.

DOI

[52]
Liu Y S, Wang J Y, Long H L, 2010. Analysis of arable land loss and its impact on rural sustainability in Southern Jiangsu Province of China. Journal of Environmental Management, 91(3): 646-653.

DOI PMID

[53]
Liu Y S, Liu Y, Chen Y F et al., 2010. The process and driving forces of rural hollowing in China under rapid urbanization. Journal of Geographical Sciences, 20(6): 876-888.

DOI

[54]
Liu Y T, He S J, Wu F B et al., 2010. Urban villages under China’s rapid urbanization: Unregulated assets and transitional neighbourhoods. Habitat International, 34(2): 135-144.

DOI

[55]
Long H L, Zou J, Liu Y S, 2009. Differentiation of rural development driven by industrialization and urbanization in eastern coastal China. Habitat International, 33(4): 454-462.

DOI

[56]
McLeman R, Smit B, 2006. Migration as an adaptation to climate change. Climatic Change, 76(1): 31-53.

DOI

[57]
Meng L R, Huang J, Dong J H, 2018. Assessment of rural ecosystem health and type classification in Jiangsu province, China. Science of the Total Environment, 615: 1218-1228.

DOI

[58]
Moghadam D M, 2015. A brief discussion on human/nature relationship. International Journal of Humanities and Social Science, 5(6): 90-93.

[59]
Omer A, Elagib N A, Zhuguo M et al., 2020. Water scarcity in the Yellow River Basin under future climate change and human activities. Science of The Total Environment, 749: 141446.

DOI

[60]
Peng J, Mao Q, Du Y Y et al., 2018. Research frontier and challenges of terrestrial system regionalization in China. Progress in Geography, 37(1): 121-129. (in Chinese)

[61]
Peng L, Xia J, Li Z H et al., 2020. Spatio-temporal dynamics of water-related disaster risk in the Yangtze River Economic Belt from 2000 to 2015. Resources, Conservation and Recycling, 161: 104851.

DOI

[62]
Potosyan A H, 2017. Geographical features and development regularities of rural areas and settlements distribution in mountain countries. Annals of Agrarian Science. https://doi.org/10.1016/j.aasci.2017.02.012.

[63]
Raymond C, Matthews T, Horton R M, 2020. The emergence of heat and humidity too severe for human tolerance. Science Advances, 6(19): eaaw1838.

DOI PMID

[64]
Rousseeuw P J, 1987. Silhouettes: A graphical aid to the interpretation and validation of cluster analysis. Journal of Computational and Applied Mathematics, 20: 53-65.

DOI

[65]
Seymour V, 2016. The human-nature relationship and its impact on health: A critical review. Frontiers in Public Health, 4: 260.

[66]
Shiro C, Furtad J I, Shen L X et al., 2007. Coping with pressures of modernization by traditional farmers: A strategy for sustainable rural development in Yunnan, China. Journal of Mountain Science, 4(1): 57-70.

DOI

[67]
Siciliano G, 2012. Urbanization strategies, rural development and land use changes in China: A multiple-level integrated assessment. Land Use Policy, 29(1): 165-178.

DOI

[68]
Song M Y, Zhang Y, 2017. Research on the relationship between geographical factors, sports and culture. Advances in Physical Education, 8(1): 66-70.

DOI

[69]
Steinhaus H, 1956. Sur la division des corps mat´eriels en parties. Bull. Acad. Polon, Sci. IV(C1.III):801-804.

[70]
Su S L, Zhang Q, Zhang Z H et al., 2011. Rural settlement expansion and paddy soil loss across an ex-urbanizing watershed in eastern coastal China during market transition. Regional Environmental Change, 11(3): 651-662.

DOI

[71]
Tambassi T, 2018. From geographical lines to cultural boundaries: Mapping the ontological debate. Rivista Di Estetica, (67): 150-164.

[72]
Tian G J, Kong L Q, Liu X J et al., 2018. The spatio-temporal dynamic pattern of rural domestic solid waste discharge of China and its challenges. Environmental Science and Pollution Research, 25(10): 10115-10125.

DOI

[73]
Tian G J, Qiao Z, Zhang Y Q, 2012. The investigation of relationship between rural settlement density, size, spatial distribution and its geophysical parameters of China using Landsat TM images. Ecological Modelling, 231: 25-36.

DOI

[74]
Tong Y Q, 2014. Research on spatial differentiation of Chinese traditional villages based on GIS. Human Geography, 29(4): 44-51. (in Chinese)

[75]
Tsutsumi H, Tanabe S, Harigaya J et al., 2007. Effect of humidity on human comfort and productivity after step changes from warm and humid environment. Building and Environment, 42(12): 4034-4042.

DOI

[76]
Wang F H, Zhang L, Zhang G P et al., 2014. Mapping and spatial analysis of multiethnic toponyms in Yunnan, China. Cartography and Geographic Information Science, 41: 86-99.

DOI

[77]
Wang N, Fang M, Beauchamp M et al., 2021. An indigenous knowledge-based sustainable landscape for mountain villages: The Jiabang rice terraces of Guizhou, China. Habitat International, 111: 102360.

DOI

[78]
Wu C J, 1991. Man-earth areal system: The core of geographical study. Economic Geography, 11(3): 1-4. (in Chinese)

DOI

[79]
Wu Z L, Li B, Hou Y, 2017. Adaptive choice of livelihood patterns in rural households in a farm-pastoral zone: A case study in Jungar, Inner Mongolia. Land Use Policy, 62: 361-375.

DOI

[80]
Xi C B, Qian T L, Chi Y et al., 2018. Relationship between settlements and topographical factors: An example from Sichuan Province, China. Journal of Mountain Science, 15(9): 2043-2054.

DOI

[81]
Xi F M, He H S, Clarke K C et al., 2012. The potential impacts of sprawl on farmland in Northeast China: Evaluating a new strategy for rural development. Landscape and Urban Planning, 104: 34-46.

[82]
Xie Y W, Wang G S, Wang X Q et al., 2017. Assessing the evolution of oases in arid regions by reconstructing their historic spatio-temporal distribution: A case study of the Heihe River Basin, China. Frontiers of Earth Science, 11(4): 629-642.

DOI

[83]
Xiong Y, Zhang J P, Xu X Y et al., 2020. Strategies for improving the microclimate and thermal comfort of a classical Chinese garden in the hot-summer and cold-winter zone. Energy and Buildings, 215: 109914.

DOI

[84]
Xu D D, Deng X, Guo S L et al., 2019. Labor migration and farmland abandonment in rural China: Empirical results and policy implications. Journal of Environmental Management, 232: 738-750.

DOI

[85]
Xu F, Ho H C, Chi G Q et al., 2019. Abandoned rural residential land: Using machine learning techniques to identify rural residential land vulnerable to be abandoned in mountainous areas. Habitat International, 84: 43-56.

DOI

[86]
Xu J, Ma H T, Luo J et al., 2019. Spatial optimization mode of China’s rural settlements based on quality-of-life theory. Environmental Science and Pollution Research, 26(14): 13854-13866.

DOI

[87]
Xu X L, 2018. Annual vegetation index (NDVI) spatial distribution data set (China, 2018). Data Registration and Publishing System of Resource and Environmental Science Data Center, Chinese Academy of Sciences

[88]
Yang H, Li X B, 2000. Cultivated land and food supply in China. Land Use Policy, 17(2): 73-88.

DOI

[89]
Yang R, Liu Y S, Long H L et al., 2015. Spatio-temporal characteristics of rural settlements and land use in the Bohai Rim of China. Journal of Geographical Sciences, 25(5): 559-572.

DOI

[90]
Yang R, Xu Q, Xu X F et al., 2019. Rural settlement spatial patterns and effects: Road traffic accessibility and geographic factors in Guangdong Province, China. Journal of Geographical Sciences, 29(2): 213-230.

DOI

[91]
Yang S Q, Matzarakis A, 2019. Implementation of human thermal comfort and air humidity in Köppen-Geiger climate classification and importance towards the achievement of Sustainable Development Goals. Theoretical and Applied Climatology, 138(1): 981-998.

DOI

[92]
Ye C, 2020. The lost countryside: Spatial production of rural culture in Tangwan village in Shanghai. Habitat International, 98: 102-137.

[93]
Ye S Z, Ma Q Y, 1990. The areal conditions, bases and differences of the rural urbanization in China’s coastal development region. GeoJournal, 21(1/2): 39-48.

DOI

[94]
Zhang R X, Di B F, Luo Y Z et al., 2018. A nonparametric approach to filling gaps in satellite-retrieved aerosol optical depth for estimating ambient PM2.5 levels. Environmental Pollution, 243: 998-1007.

DOI

[95]
Zhang Y Z, Baimu S L, Tong J et al., 2018. Geometric spatial structure of traditional Tibetan settlements of Degger County, China: A case study of four villages. Frontiers of Architectural Research, 7(3): 304-316.

DOI

[96]
Zhou G H, He Y H, Tang C L et al., 2013. Dynamic mechanism and present situation of rural settlement evolution in China. Journal of Geographical Sciences, 23(3): 513-524.

DOI

[97]
Zhou W, Gao X S, Ling J et al., 2016. Research and application of TR-SE mode about rural settlements layout in Qinba Mountain areas in Sichuan. Scientific and Technological Management of Land and Resources, 33(6): 87-93. (in Chinese)

[98]
Zhou Z X, Wang X D, Qian Y, 2017. A GIS based research on spatial distribution of the mountainous settlements in karst valley: A case study of Baishui River Valley, Guizhou. Design Community, (1): 14-22. (in Chinese)

[99]
Zou Y L, 2006. Environmental change and urban vicissitude of the Yellow River Basin in historical period. Jianghan Tribune, (5): 98-105. (in Chinese)

Options
Outlines

/

〈 〉