International Journal of Geo-Information

Article Spatiotemporal Pattern Analysis of China’s Cities Based on High-Resolution Imagery from 2000 to 2015

Hanchao Zhang 1 , Xiaogang Ning 2, Zhenfeng Shao 3,* and Hao Wang 2,4

1 School of Remote Sensing and Information Engineering, Wuhan University, 129 Luoyu Road, Wuhan 430079, China; [email protected] 2 Institute of Photogrametry and Remote Sensing, Chinese Academy of Surveying and Mapping, Beijing 100830, China; [email protected] (X.N.); [email protected] (H.W.) 3 State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, 129 Luoyu Road, Wuhan 430079, China 4 Beijing Key Laboratory of Urban Spatial Information Engineering, Beijing 100038, China * Correspondence: [email protected]; Tel.: +86-27-6877-9859



Received: 28 March 2019; Accepted: 14 May 2019; Published: 22 May 2019


Abstract: The urbanization level in China has increased rapidly since beginning of the 21st century, and the monitoring and analysis of urban expansion has become a popular topic in geoscience applications. However, problems, such as inconsistent concepts and extraction standards, low precision, and poor comparability, existing in urban monitoring may lead to wrong conclusions. This study selects 337 cities at the prefecture level and above in China as research subjects and uses high-resolution images and geographic information data in a semi-automatic extraction method to identify urban areas in 2000, 2005, 2010, and 2015. City size distribution patterns, urban expansion regional characteristics, and expansion types are analyzed. Results show that Chinese cities maintained a high-speed growth trend from 2000 to 2015, with the total area increasing by 115.79%. The overall scale of a city continues to expand, and the system becomes increasingly complex. The urban system is more balanced than the ideal Zipf distribution, but it also exhibited different characteristics in 2005. Urban areas are mostly concentrated in the eastern and central regions, and the difference between the east and the west is considerable. However, cities in the western region continuously expand. Beijing, Shanghai, Tianjin, and Guangzhou are the four largest cities in China. Approximately 73.30% of the cities are expanding in an extended manner; the urban form tends to be scattered, and land use efficiency is low. The new urban areas mainly come from cultivated land and ecological land.

Keywords: urban area; urban expansion; spatiotemporal pattern analysis; high-resolution imagery; China

1. Introduction The urban population is increasing with the development of urbanization, along with the rapid expansion of urban areas [1]. In particular, China’s urbanization entered a new stage from 2000 onwards. The 19th National Congress of the Communist Party of China, the 13th Five-Year Plan, the National New Urbanization Plan (2014–2020), and the Central City Work Conference (2015) proposed to rationally control urban development boundaries, protect basic farmlands, and optimize cities [2,3]. The internal space structure promotes compact and intensive urban development and efficient green development to solve the blind expansion of cities, faster land urbanization than population urbanization, extensive and inefficient land construction, urbanization and industrialization mismatch, urban spatial distribution, and unreasonable scale structures [4]. Many problems, such as insufficient resources and environmental carrying capacity, exist in addition to outstanding ecological and environmental problems [5,6]. The reasonable control of urban growth boundaries, optimization of

ISPRS Int. J. Geo-Inf. 2019, 8, 241; doi:10.3390/ijgi8050241 www.mdpi.com/journal/ijgi ISPRS Int. J. Geo-Inf. 2019, 8, 241 2 of 23 urban internal spatial structures, and promotion of compact urban development have become inherent requirements for sustainable urban development [7,8]. The monitoring and analysis of urbanization play extremely important roles in accurately capturing the urbanization process, scientifically implementing urban planning and smart city construction [9], and promoting urban sustainable development. Urbanization can be measured in terms of population urbanization and land urbanization [10]. Population urbanization can be expressed by the urbanization rate (the ratio of the urban population to the total population) [11]. Land urbanization can be seen in terms of non-agricultural urban areas [12]. Compared with the intangible indicator of population transfer caused by population urbanization, land urbanization, i.e., urban spatial expansion, is the direct result of urbanization acting on geospatial space [13]. Hence, it is easier to measure objectively. Remote sensing technology exhibits the advantages of providing real, objective, and current information and requires a low cost; it has become an important technical means for monitoring land urbanization [14–17]. The primary task of urbanization monitoring is to determine urban areas or boundaries. However, urban areas and boundaries have many related concepts. These definitions have different starting points and characteristics [18,19]. Many studies have regarded construction land or impervious surfaces in land use classification results as urban areas [20,21]. Some researchers have defined an urban area by considering the functions of urban land [22]. Numerous sources of image data are used in monitoring urbanization [23]. During the early days, Moderate Resolution Imaging Spectroradiometer (MODIS) images were the primary source of data for monitoring urbanization [24,25]. Subsequently, images from medium-resolution satellites, such as the Landsat series [26,27], were widely used [28–33]. As the number of launched high-resolution satellites increases, many researchers have attempted to extract urban areas from high-resolution remote sensing images [34–40]. However, these images are mostly used on a single city scale due to their ability to automate extraction. Many scholars have analyzed urban expansion patterns and characteristics by obtaining urban areas [41]. The urban spatial pattern is the overall spatial pattern of a city, the spatial distribution pattern of various elements within a city, and the comprehensive performance of the dynamic evolution process. The overall spatial pattern of a city is based on the urban scope. The internal elements of a city are based on land use [42] and land cover [43], including the spatial characteristics of scale, shape, density, distribution, location, and relationship. Many scholars have also used multisource data, such as remote sensing images. Research on urban overall spatial patterns from static expressions and dynamic analyses are in multiple scales, such as land parcels, blocks, functional areas, cities, urban agglomerations, regions, countries, and the world. Li uses the rank-size distribution and the Gini coefficient method to analyze the urban spatial pattern of the Yangtze River Economic Belt based on nighttime stable light data, multi-temporal urban land products, and multiple sources of geographic data [44]. Zheng analyzes the urban spatial pattern of Hubei province by ’point-axis system’ theory, light index model, gravity model and social network analysis. Liu makes a high-resolution multi-temporal mapping of global urban land using Landsat images based on the Google Earth Engine Platform and Normalized Urban Areas Composite Index [45]. On the basis of urban area extraction, research on urban spatial form introduces fractal dimension, compactness, aggregation index, urban center of gravity transfer index, spatial autocorrelation index, maximum patch index, expansion speed, expansion intensity, extended contribution rate, and landscape [46–48]. Expansion indices and equal measure indicators have been used, and superposition analysis, convex hull analysis, fan analysis, quadrant quantile analysis, buffer analysis, circle analysis, center of gravity migration analysis, and other methods have been conducted to analyze the extent, pattern, and direction of urban boundaries and expansions [49–51]. Subsequently, studies on urban overall spatial morphological characteristics and spatiotemporal evolution have been conducted. The research of various scholars provides numerous methods for understanding the spatiotemporal expansion process of China’s urbanization. However, several shortcomings remain. First, each study presents a different understanding of urban concepts, inconsistent extraction standards, and considerable differences in monitoring results, thereby causing difficulty in performing accuracy ISPRS Int. J. Geo-Inf. 2019, 8, 241 3 of 23 verification and comparative analysis. Second, the results are mostly based on MODIS and Landsat series satellite imagery [52]. The low resolution of these images leads to the poor accuracy of urban extraction results. The beginning of the 21st century was the period of rapid development in Chinese cities; however, high-resolution remote sensing images were still scarce at that time [48]. No set of high-precision monitoring products was available nationwide. To solve the aforementioned problems, a semi-automatic method of urban boundary extraction was proposed by using high-resolution image and geographic information data. Urban landscape and form characteristics, geographical knowledge, and a series of standardized rules were combined to generate a high-precision and consistent urban boundary. The current study uses a semi-automatic method to extract urban area using high-resolution remote sensing imagery and geographic information data. Urban landscape and form characteristics, geographical knowledge, and a series of standardized rules were combined to generate high-precision and consistent urban areas of 337 cities in 2000, 2005, 2010, and 2015. Then, a city size distribution model was verified, urban expansion regional characteristics were analyzed at three main levels, i.e., the city, province and region levels, and expansion types were also analyzed.

2. Study Areas and Data

2.1. Study Areas A total of 337 cities (including capital cities, prefecture-level cities, autonomous prefectures, regions, and alliances, but excluding Hong Kong, Macao, and Taiwan) are selected as research areas (Figure1). A municipal district is selected as the urban extraction scope, and a city with administrative division adjustment is subjected to the 2018 municipal jurisdiction. There are 4 megacities (population of 10 million people or more), 13 supercities (population above 5 million but below 10 million), 36 large cities (population above 1 million but below 5 million) and hundreds of small-medium cities (population below 1 million). The research areas are divided into four regions based on their economic characteristics: the eastern, central, western, and northeastern regions. Cities at the prefecture level and above include national economic, political, and population centers, and their urbanization level can represent the highest in the region. During the early 21st century, cities at the prefecture level and above progressed rapidly in terms of economic and social development, and urban areas expanded rapidly. These cities are representatives of the monitoring and analysis of urbanization spatial expansion in China. ISPRS Int. J. Geo-Inf. 2019, 8, 241 4 of 23 ISPRS Int. J. Geo-Inf. 2019, 8, x FOR PEER REVIEW 4 of 24

FigureFigure 1.1. Distribution of 337 337 citi citieses and and four four regions regions across across the the country. country.

2.2.2.2. Data Data ThisThis study study uses uses high-resolution high-resolution remote remote sensing sensing images images as theas the primary primary source source of data, of data, which which cover thecover entire the study entire area. study The area. data The types data listed types in listed Table in1 areTable arranged 1 are arranged in accordance in accordance with the with adopted the orderadopted of priority. order of Amongpriority. these Among data, these the data, 2015 remotethe 2015 sensing remote imagessensing of images the study of the area study is primarily area is fromprimarily the “Geographical from the “Geographical Conditions MonitoringConditions (GCM)”Monitoring national (GCM)” project national [53], whichproject is[53], obtained which from is theobtained National from Basic the GeographicNational Basic Information Geographic Center. Information Remote Center. sensing Remote images sensing of 2010 images were of mostly2010 providedwere mostly by the provided Ministry by ofthe Natural Ministry Resources. of Natura Insul Resources.fficient imagesInsufficient were images supplemented were supplemented by collecting theby firstcollecting national the censusfirst national image census data of image the country. data of Satellitethe country. imagery Satellite was imagery obtained was from obtained September from to December,September and to December, the resolution and isthe generally resolution higher is gene thanrally 1m. higher In 2000 than and 1m. 2005, In 2000 remote and sensing 2005, remote images, includingsensing images, high-resolution including remote high-resolution sensing images remote and sensing aerial photographs, images and wereaerial mostly photographs, from surveying were andmostly mapping from surveying geographic and information mapping geographic departments. information Insufficient departments. images were Insufficient supplemented images through were networksupplemented downloading through and network purchasing, downloading and images and purchasing, from September and toimages December from wereSeptember preferred. to ADecember small number were ofpreferred. areas, below A small 1%, number were supplemented of areas, below by 1%, images were supplemented from the years by before images and from after duethe toyears the widebefore coverage and after of due the to study the wide area andcovera thege long-time of the study span. area To ensureand the the long-time consistency span. of To the extractionensure the results, consistency the image of the data extraction were uniformly results, resampled the image to data a resolution were uniformly of 2 m. Theresampled auxiliary to data a primarilyresolution included of 2 m. The the resultsauxiliary of thedata GCM primarily and basic included geographic the results information, of the GCM which and were basic obtained geographic from theinformation, Ministry of which Natural were Resources. obtained Thesefrom the data Ministry included of landNatural cover, Resources. urban road These elements, data included geographical land names,cover, townships,urban road andelements, county geographical administrative names, divisions townships, and can and provide county goodadministrative references divisions for urban areaand extraction.can provide good references for urban area extraction.

ISPRS Int. J. Geo-Inf. 2019, 8, x FOR PEER REVIEW 5 of 24

ISPRS Int. J. Geo-Inf. 2019, 8Table, 241 1. Primary remote sensing image data used in this study. 5 of 23

Year Month Data Type/Source (Resolution) Coverage Ratio April, July, August,Table 1. PrimaryAerial remote Images sensing (1m), image IKONOS data used in this study. 2000 September, October, (1m), small number of Landsat Year Month Data Type/Source (Resolution) Coverage Ratio November, December data (30 m) April, July, August, June, July, August, QuickBirdAerial Images (0.61m), (1 m),aerial IKONOS images (1 m), 2000 September, October, small number of Landsat data (30 m) 2005 September,November, October December, (1m), IKONOS (1m), SPOT 5 November, December (2.5m), minor Landsat data (30m) WorldView-1/2 (22.37%), June, July, August, QuickBird (0.61 m), aerial images (1 m), 2005 September, October, WorldView-1/2IKONOS (1 (0.5m), m), SPOT QuickBird 5 (2.5 m), aerialWorldView-1 images (19.76%),/2 (22.37%), September,November, October December, (0.61m),minor aerial Landsat images data (0.5m), (30 m) IKONOSaerial (14.62%), images (19.76%), others 2010 IKONOS (14.62%), November, December WorldView-1SPOT 5 (2.5m),/2 (0.5 ALOS m), QuickBird (2.5m), (0.61 m), (43.25%) September, October, others (43.25%) 2010 aerial imagesCIRS-P5 (0.5 (2.2m) m), SPOT 5 (2.5 m), November, December WorldView-1/2ALOS (2.5 m),(0.5m), CIRS-P5 Pleiades (2.2 m) September, October, (0.5m),WorldView-1 aerial images/2 (0.5 m), (0.5m), Pleiades SPOT (0.5 m), September, October, 20152015 aerial images (0.5 m), SPOT 6/7 (1.5 m), NovemberNovember,, December December 6/7 (1.5m), Mapping Satellite I (2 Mapping Satellite I (2 m), ZY-3 (2.1 m) m), ZY-3 (2.1m)

3.3. Methods Methods TheThe method method adopted adopted in in this this stud studyy includes includes three three major steps: (1) (1) data data preprocessing, preprocessing, (2) (2) urban urban areaarea extraction, and and (3) urban spatiotemporal pattern analysis. Figure Figure 2 showsshows thethe overalloverall analysisanalysis processprocess of of the spatiotemporal pattern of Chinese ci cities.ties. First, First, high-resolution high-resolution remote remote sensing sensing images images obtainedobtained inin 2000, 2000, 2005, 2005, 2010, 2010, and 2015and were2015 screened.were screened. Second, imageSecond, preprocessing, image preprocessing, fusion, mosaicking, fusion, mosaicking,and cropping and were cropping performed were onperformed the data on to satisfythe data the to requirements.satisfy the requirements. Third, the Third, method the proposed method proposedin [35] was in used[35] was to extractused to urbanextract areas, urban which areas, werewhich then were manually then manually modified modified in accordance in accordance with withcity characteristicscity characteristics and theand extractionthe extraction principle princi adoptedple adopted to obtain to obtain the finalthe final urban urban area. area. Lastly, Lastly, the thespatiotemporal spatiotemporal pattern pattern of of Chinese Chinese cities cities was was analyzed analyzed from from three three aspects: aspects: urbanurban size distribution pattern,pattern, urban expansion area char characteristics,acteristics, and extended form.

FigureFigure 2. 2. FlowchartFlowchart of of the the analysis analysis of of China’ China’ss urban urban expansion expansion characteristics. characteristics.

ISPRS Int. J. Geo-Inf. 2019, 8, 241 6 of 23

3.1. Data Preprocessing Data preprocessing was accomplished using professional software, such as ArcGis10.3, ERDAS 2016, and ENVI 5.3. (1) Data screening Data screening requires selecting images with a resolution higher than 2 m within a city’s jurisdiction. The image phase is preferably summer, and the cloud snow coverage is preferably less than 10%. The 2000 national coordinate system is used, and other coordinate system images are required after correcting the registration. In addition, some data from the Ministry of Natural Resources require extensive inspection and screening due to problems including the inability to open data copying errors, incomplete data, and redundant data. (2) Orthorectification correction This method primarily includes the orthorectification of a single image and the registration of images from different coordinate systems. Orthorectification corrects imaging deviations caused by the imaging method used by the sensor during image capture, the influence of a change in the external orientation element, terrain fluctuation, Earth’s curvature, atmospheric refraction, and the influence of Earth’s rotation. Image registration completes the geometric transformation of images from other coordinate systems to the 2000 national coordinate system by selecting the same name point. (3) Image mosaicking This process includes image stitching and acquiring uniformity. After image registration is completed, all the images are uniformly resampled to a resolution of 2 m in order to mosaic. A large number of remote sensing images are spliced to obtain a geometrically uniform color image by searching for the optimal mosaic edges and applying image tonal adjustment. (4) Image cropping This process is primarily used in urban administrative division to appropriately cut an entire remote sensing image obtained by splicing, thereby removing unnecessary parts and reducing the amount of data. (5) Administrative division boundary treatment Administrative division line vector data are obtained from the Ministry of Natural Resources, and the administrative division line vector of a city jurisdiction is selected by comparing the city jurisdiction code in the National Administrative Division Manual. The vector data of the administrative divisions of each municipal district are merged to generate the administrative boundaries of a city’s jurisdiction. The demarcation line between the sub-city jurisdiction and the administrative division of the city is reserved.

3.2. Urban Area Extraction Urban area extraction is a combination of automatic computer extraction and manual interpretation correction. The automatic extraction of residential polygons from high-resolution remote sensing images is performed using fused right-angled and right-angled features [38]. The polygons of the settlements are superimposed with the location of the district government, whereas the polygons of the concentrated contiguous settlements, where the district government is located, are used as the initial urban area. The classification system and standards for geographic country monitoring indicate that the original urban area should be manually interpreted and modified in accordance with the urban area extraction principles and rules to obtain the final urban area. The automatic method uses feature-level-based fusion of right-angle-corners and right-angle-sides to extract the human settlements. It is composed of five steps, namely (1) detection of line segments, (2) detection of Harris corners, (3) verification of corners by line segments, (4) construction of built-up area index, (5) thresholding of human-settlement index. After the automatic extraction of human settlements, manual interpretation is used for urban extraction. Urban area extraction should start from actual city construction completion under full consideration of urban landscape. Buildings are the core; natural landscapes are the auxiliary and roads ISPRS Int. J. Geo-Inf. 2019, 8, 241 7 of 23 are the bond of the urban landscape. Using top-down process, rough initial boundary is extracted first, and then refine it under the principles based on the GCM data, fundamental infrastructure of geographical information. The process should be in reverse chronological order and must ensure the accurate topology between multi-phase boundaries. The final urban area includes the central city and the enclave-type urban area. The revised principles and rules for an initial urban area are as follows: (1) Principle for defining the boundaries of administrative divisions. An urban area must be within the scope of the administrative division of the city and must not exceed the boundaries of the administrative division. (2) Principle for drawing outlines of urban area. An urban area outline is preferentially drawn along the boundary of linear objects, such as roads and rivers, and cannot cross buildings (districts) and structures. The inner side of an urban boundary line should have no cultivated land. When an urban boundary falls on a river bank and the surrounding area of the river bank is a green landscape for urban construction land, the urban boundary line is drawn along the high-water line of the river, and the green part is classified as an urban area; otherwise, the road near the river bank is drawn. Urban boundary is preferentially sketched along the border of complete block, which has obvious boundary, such as courtyards or fences. If the block boundary is not obvious or the proportion of unconstructed area in the whole block is high, urban boundary is sketched along the actual boundary of the constructed land. (3) Principle for determining the central city. The central centralized and contiguous region, which is named as central urban areas, is connected to district government through the road, and the border of the region extends towards outside until the distance of constructed land to central urban areas is more than 50 m. (4) Principle for determining urban landscapes. An urban landscape mostly includes urban housing construction areas, urban roads, urban green spaces, city squares, parking lots, stadiums, and other landscapes. The block surrounded by urban roads is the most important urban landscape feature. (5) Principle for judging enclave urban areas. An enclave-type urban area, i.e., an enclave-type concentrated contiguous area, is an important part of an urban area. It exhibits the following characteristics: close contact with the central city through a road; urban landscape features; an area larger than 60 hectares; and contains at least one of the following geographic units (urban-level street offices, administrative departments, industrial and mining areas, development districts, research institutes, universities, and colleges). (6) Principle for the reverse extraction of urban boundaries from new time to old time. First, the urban area extraction of the latest phase is completed. Second, the inverse incremental updating technique is used to obtain the urban area change element in the order of the current situation from new to old. Lastly, the post-phase area is merged with the urban area change to obtain the previous time. The urban area of the phase ensures accurate topology between multi-period boundaries. The combination of automatic computer extraction and manual interpretation correction considerably saves computational efficiency compared with performing manual visual inspection and manual sketching to obtain the initial urban area. After testing, the efficiency of automatic identification is determined to be nearly twice higher than that of manual visual interpretation. Combined with the initial urban area interpretation correction, the comprehensive improvement efficiency reaches 30%. Examples of typical features within urban areas which can help us to identify the urban area in this study are presented in Figure3. ISPRS Int. J. Geo-Inf. 2019, 8, 241 8 of 23 ISPRS Int. J. Geo-Inf. 2019, 8, x FOR PEER REVIEW 8 of 24

(a) High-rise building area (b) Low-rise building area

(c) Urban roads (d) Urban green space

(e) Urban water (f) A structure

(g) Urban village (h) Enclave

FigureFigure 3. 3.Examples Examples ofof urbanurban featurefeature interpretationinterpretation symbol: ( (aa)) an an example example of of high-rise high-rise buildings buildings in in urbanurban areas; areas; ( b(b)) an an example example ofof low-riselow-rise buildings in urban area; area; ( (cc)) an an example example of of roads roads in in urban urban area; area; (d()d an) an example example of of green green space space nearnear thethe edgeedge of urban area; area; ( (ee)) an an example example of of urban urban water water, ,namely namely an an artificialartificial lake, lake, in in urban urban area; area; (f) an(f) examplean example of structure, of structure, namely name a stadium,ly a stadium, in urban in urban area; ( garea;) an example(g) an ofexample urban village; of urban (h )village; an example (h) an of example enclave, of namely enclave, an namely industrial an industrial and mining and area, mining of urbanarea, of area. urban area.

ISPRS Int. J. Geo-Inf. 2019, 8, 241 9 of 23

3.3. City Size Distribution City size distribution refers to the hierarchical distribution of cities in a country or region. Urban population, urban area, economic indicators, and composite indicators are frequently used as indicators of urban scale [54,55]. An urban area is an urban scale indicator [56]. The characteristics of urban distribution within a region are determined by analyzing the relationship between the sequence of large and small cities in the region and its scale [57]. The order-scale rule examines the size distribution of an urban system from the relationship between city size and city size order. Its formula is as follows:

q P = P R− (1) i 1· i where Ri is the city sorted by size from large to small i with the order P, i is the city size with the order Ri, P1 is the theoretical value of the first city size, and parameter q is typically called the Zipf index. In the study of urban systems, the order-scale rule is frequently combined with fractal theory to better understand the characteristics of city size distribution. The fractal dimension (D) of city size distribution and q in Equation (1) exhibit the following relationship in accordance with fractal theory:

D q = R2 (2) × where R2 represents the coefficient of determination, D is the fractal dimension and q is the Zipf index which can reflect the equilibrium of an urban system. (1) When q is large and D is small, the city size distribution in a region is concentrated, large cities with high ranks are prominent, and small and medium-sized cities with moderate and low ranks are insufficiently developed. The scale difference between cities is considerable, and the urban scale system is unbalanced. (2) When D is large and q are small, the city size distribution in the region is relatively scattered, the scale of high-level cities is not prominent, and small and medium-sized cities are relatively developed. The city size distribution only differs slightly, and the urban scale system is more balanced. (3) When q and D are close to 1, the ratio of the size of the first city to that of the smallest city is close to the total number of cities in the region. The city size distribution is close to the ideal state of Zipf, and the proportion of cities in each scale is reasonable. The theory of city size distribution indicates that a regular sequence distribution is obtained when q = 1, namely, the Czech Republic model. In the first distribution, i.e., when q > 1, the urban population is concentrated, the urban system is dominated by large cities, and small and medium-sized cities are insufficiently developed. In a sequence distribution, i.e., when q < 1, the urban population is scattered, large cities in the urban system are not prominent, and small and medium-sized cities are developed. When q = 0, all urban populations are equal and belong to the average distribution. However, when q = , the region has only one city. ∞ lgP1 can stand for structural capacity (Sc). For urban systems, when the structural capacity is large, the urban system is complex and the overall scale is large. When the structural capacity is small, the urban system is simple and the overall size is small.

3.4. Urban Expansion Analysis Indicator After obtaining the urban area, urban time–space expansion is analyzed using urban expansion speed and intensity. (1) Expansion speed Vi is the annual growth rate of an urban area in a certain city within a particular period. It indicates the absolute (area) difference of the expansion speed of different urban areas per unit time.   V = ∆U / ∆t 100% (3) i ij j × where Vi is the urban expansion speed, ∆Uij is the i-th research unit urban expansion area in the j period, and ∆t is the time span of the j period. ISPRS Int. J. Geo-Inf. 2019, 8, x FOR PEER REVIEW 10 of 24 ISPRS Int. J. Geo-Inf. 2019, 8, 241 10 of 23 where 𝑉 is the urban expansion speed, ∆𝑈 is the 𝑖-th research unit urban expansion area in the 𝑗period, and ∆𝑡 is the time span of the j period. (2) Expansion intensity Ni is the annual expansion ratio of an urban area in a certain city relative (2) Expansion intensity 𝑁 is the annual expansion ratio of an urban area in a certain city relative toto the the base base period period within within a a particular particular period. period. It It in indicatesdicates the the relative relative (proportional) (proportional) difference difference in in the the speedspeed of of expansion expansion of of different different urban areas per unit time.   𝑁N =∆𝑈= ∆U⁄/∆𝑡∆t ×𝑀M × 100% (4) (4) i ij j × i × where 𝑁 is the urban expansion intensity, ∆𝑈 is the 𝑖-th urban expansion study area in the j where Ni is the urban expansion intensity, ∆Uij is the i-th urban expansion study area in the j period, ∆𝑡 M 𝑖 period,∆tj is the time is the span time of thespanj period, of the j and period, Mj is and the total isi -ththe unit total urban -th unit area urban during area the during initial stage the initial of the stagej period. of the j period.

3.5.3.5. Urban Urban Expansion Expansion Type Type TheThe type type of of urban urban space space expansion expansion typically typically includes includes infilling infilling and and edge-expansion edge-expansion [58], [58], which which cancan be be determined determined via via convex convex hull hull anal analysisysis [59]. [59 The]. The convex convex hull hull of a of city’s a city’s land land outline outline refers refers to the to smallestthe smallest convex convex polygon polygon that that contains contains the theoutline outline of ofthe the city’s city’s periphery. periphery. The The convex convex hull hull can can be be consideredconsidered an an urban urban area area or or the the potential potential control control area of a city. TheThe convex convex shell shell analysis analysis method method can can be be used used to to identify identify the the spatial spatial expansion expansion pattern pattern of of urban urban land.land. If If urban urban land land expansion expansion belongs belongs to to the the filling filling pattern, pattern, then then the the extended extended land land is is mostly mostly inside inside thethe convex convex hull hull of of the the urban urban area area.. If If urban urban land land expansion expansion belongs belongs to to the the expansion expansion pattern, pattern, then then the the extendedextended land land is is mostly mostly outside outside the the convex convex hull hull of of the the urban urban area. area. That That is, is, in in urban urban land land expansion, expansion, ifif the the areaarea inside inside the the convex convex shell shell is is larger larger than than the the area area outside outside the the convex convex shell, shell, then then the the urban urban land land expansionexpansion belongs belongs to to the the filling filling pattern pattern and and vice vice vers versa.a. Figure Figure 4 shows us the examples of two types.

(a) Infilling Type (b) Edge-expansion Type

FigureFigure 4. 4. ExamplesExamples for for urban urban expansion expansion types. types.

4.4. Results Results and and Discussion Discussion 4.1. Accuracy Analysis of Urban Area Extraction Results 4.1. Accuracy Analysis of Urban Area Extraction Results This study extracts the urban areas of 337 cities at prefecture level and above. Reliability and high This study extracts the urban areas of 337 cities at prefecture level and above. Reliability and precision are primarily reflected in the following three aspects. high precision are primarily reflected in the following three aspects. (1) The technical method guarantees high precision of the extraction result. The urban area is first (1) The technical method guarantees high precision of the extraction result. The urban area is extracted by adopting the automatic extraction method of residential points with right-angled features first extracted by adopting the automatic extraction method of residential points with right-angled to ensure the consistency of the extraction results. In addition, the adopted extraction principles and features to ensure the consistency of the extraction results. In addition, the adopted extraction standards are strict and standardized, and operability is strong. Interpreters must strictly follow the six principles and standards are strict and standardized, and operability is strong. Interpreters must standardized extraction principles and the specific technical regulations so as to ensure the reliability strictly follow the six standardized extraction principles and the specific technical regulations so as of the urban area extraction results. to ensure the reliability of the urban area extraction results. (2) As the primary data, high-resolution remote sensing images guarantee the high precision of the extraction result. The adopted remote sensing image exhibits high resolution and can effectively

ISPRS Int. J. Geo-Inf. 2019, 8, x FOR PEER REVIEW 11 of 24

ISPRS Int. J. Geo-Inf. 2019, 8, 241 11 of 23 (2) As the primary data, high-resolution remote sensing images guarantee the high precision of the extraction result. The adopted remote sensing image exhibits high resolution and can effectively avoid misclassification misclassification and leakage. The The results results of of some studies show that resolution considerably influencesinfluences the the accuracy accuracy of of urban urban area area extraction. extraction. Compared Compared with with MODIS MODIS and Landsat and Landsat images, images, high- resolutionhigh-resolution images images can effectively can effectively avoid avoid the misclassification the misclassification and leakage and leakage of urban of urban boundaries boundaries and canand combine can combine urban urban and and urban–rural urban–rural areas areas and and townships. townships. Thus, Thus, the the construction construction land land can can be separated usingusing high-resolution high-resolution remote remote sensing sensing images, images, which which is di ffiiscult difficult to be separatedto be separated using MODIS using MODISor Landsat or Landsat images. images. (3) High-quality High-quality auxiliary auxiliary data data guarantee guarantee the the high high precision precision of extraction of extraction results. results. Difficult Diffi areascult haveareas real have and real reliable and reliable auxiliary auxiliary data support data support to ensure to ensure the accuracy the accuracy of the of extracted the extracted results. results. The largestThe largest difficultydifficulty in extracting in extracting urban urban boundaries boundaries is isthe the determination determination of of concentrated concentrated contiguous urban landscapes and enclave-type urban areas. The current study solves the three aforementioned urban-related problems by considering the the results results of of GCM and the data of urban functional units, units, development zones,zones, bonded bonded areas, areas, and and urban urban roads ro inads the 2015in the basic 2015 geographical basic geographical condition monitoringcondition monitoringresults. The results. large features The large are features identified. are identified. The capital cities are selected to be compared with other results, such as the result of MODIS in the report “East Asia's Asia’s Changing Changing Urban Urban Landscape” Landscape” (product (product A) and the result of manual extraction by Landsat images (product B) in [60]. [60]. Two Two figures figures from [60] [60] may give give us us a a more more succinct succinct comparison comparison of different different results (Figures(Figure 5 and 6Figure). 6).

3,000 Urban area for 2000 of product A Expansion urban area of 2000-2010 of product A 2,750 Urban area for 2000 of product B Expansion urban area of 2000-2010 of product B 2,500 Urban area for 2000 of product C Expansion urban area of 2000-2010 of product C 2,250 2,000 1,750 1,500 1,250 Urban Area 1,000 750 500 250 0 Beijing Shanghai Tianjin Guangzhou Wuhan Chongqing Xi'an Chengdu Shenyang Jinan Hangzhou Nanjing Changchun Shijiazhuang Zhengzhou Urumqi Harbin Taiyuan Kunming Changsha Lanzhou Hefei Fuzhou Nanning Hohhot Haikou Nanchang Guiyang Yinchuan Lhasa Xining

Capital Cities

Figure 5.5. TheThe comparisoncomparison of of extraction extraction results results of urban of urban areas. areas. Product Product C is the C resultis the of result proposed of proposed method. method. Figure5 shows us that the area of product A is apparently larger than PRODUCT B and product C. AndFigure product 5 shows B is slightlyus that the larger area than of product product A A. is The apparently t-test tells larger us that than product PRODUCT B and B product and product C are C.basically And product consistent. B is slightly However, larger product than A product is quite A. di ffTheerent t-test from tells product us that B product and product B and C. product Then, we C arechoose basically the large consistent. different However, cities to compare.product A is quite different from product B and product C. Then, we chooseThe comparison the large different of Beijing cities (Figure to compare.6) shows us that the urban product A is much larger than productThe Bcomparison and product of C. Beijing From the(Figure high-resolution 6) shows us images, that the we urban can find product that there A is aremuch many larger mistakes than productin product B and A. Many product farmlands, C. From foreststhe high-resolution even some mountains images, we are can mistaken find that for there urban are areas. many The mistakes results inof product product A. B andMany the farmlands, results of forests this paper even are some similar. mountains The figures are mistaken from Figure for urban6b to areas. Figure The6i results show ofthat product the urban B and areas the with results some of high-risethis paper buildings are simila andr. The obvious figures aggregation from Figure in Figure 6b to6 Figureb are excluded 6i show thatfrom the product urban B. areas However, with some the low-rise high-rise buildings buildings in and the urban–ruralobvious aggregation integration in areaFigure (Figure 6b are6c), excluded part of fromthe rural product area B. in However, Figure6d, the and low-rise the large buildings farmland in areasthe urban–rural in Figure6 eintegration were incorrectly area (Figure classified 6c), part as ofbuilt-up the rural areas area in in the Figure results 6d, for and B, the and large the resultsfarmland were areas compared in Figure with 6e were the corresponding incorrectly classified Landsat5 as

ISPRSISPRS Int. Int. J. J. Geo-Inf. Geo-Inf.2019 2019, ,8 8,, 241 x FOR PEER REVIEW 12 12 of of 24 23

built-up areas in the results for B, and the results were compared with the corresponding Landsat5 TM images Figure6f,g. It can be seen from Figure6h,i that in these areas, except for Figure6f, it is TM images Figure 6f,g. It can be seen from Figure 6h,i that in these areas, except for Figure 6f, it is obvious that there is a missing phenomenon in product B, and that it is difficult to accurately identify obvious that there is a missing phenomenon in product B, and that it is difficult to accurately identify thethe boundary boundary inin FigureFigure6 6g–i.g–i. Nearby features features can can ea easilysily lead lead to tomisunderstan misunderstandings.dings. According According to the to theextraction extraction results results of Beijing, of Beijing, the accuracy the accuracy of result of result A is the A is worst, the worst, and the and precision the precision of result of B result is high. B is high.The Theaccuracy accuracy of our of ourresult result is the is thehighest. highest. Prod Productuct A has A has the the worst worst accura accuracy,cy, mainly mainly because because the the resolutionresolution of of MODIS MODIS is is much much lowerlower thanthan thatthat of the others; it it is is difficult difficult to to achieve achieve the the high-resolution high-resolution extractionextraction result result and and it it no no longer longer joinsjoins thethe comparisoncomparison of Chongqing city. city.

FigureFigure 6. 6.Comparison Comparison examplesexamples forfor urban boundaries of A, A, B B and and C C in in 2010—Beijing, 2010—Beijing, Chongqing. Chongqing.

InIn Chongqing, Chongqing, Figure Figure6 6jj shows shows the the overalloverall situationsituation of the results of of the the two two major major urban urban areas. areas. ItIt can canbe be seen seen that thatthe the extractionextraction resultsresults are genera generallylly consistent, consistent, and and the the details details of of specific specific boundaries boundaries areare slightly slightly di different.fferent. Figure Figures6k–n 6k–n show show the the boundary bounda ofrythe of the more more specific specific di ff differences,erences, from from Figure Figure6k,n showed6k,n showed that there that there is some is some leakage leakage in productin product B; B; Figure Figure6l 6l shows shows that that there there are are somesome mistakes of of farmlandfarmland and and bare bare land land in in product product B.B. FigureFigure6 6mmshows shows thatthat thethe results in this paper paper are are more more accurate accurate thanthan those those of of product product B. B. ByBy comparing comparing thethe resultsresults ofof thethe threethree didifferentfferent datadata sources extraction, extraction, it it can can be be seen seen that that the the resolutionresolution of of the the data data sourcesource forfor urbanurban area extracti extractionon has has a a remarkable remarkable effect effect on on the the result result accuracy. accuracy.

ISPRS Int. J. Geo-Inf. 2019, 8, x FOR PEER REVIEW 13 of 24

Using high-resolution data to extract urban areas can effectively help in avoiding mistakes, omission and ambiguous boundaries; it can also accurately distinguish the city, the integration of urban and rural areas, rural, and accurately obtain the city internal elements distribution, morphology and structure information; these extraction results can play an important role in urban development.

4.2. Analysis of City Size Distribution ISPRS Int. J. Geo-Inf. 2019, 8, 241 13 of 23 The total area of Chinese cities was 16012.84 km2 in 2000, 21200.41 km2 in 2005, 27778.11 km2 in 2010, and 34554.31 km2 in 2015, thereby indicating an increase of 32.40%, 31.04%, and 24.39%. The Using high-resolution data to extract urban areas can effectively help in avoiding mistakes, omission area in 2015 is 2.16 times that which existed in 2000. This result shows that China’s urban growth has and ambiguous boundaries; it can also accurately distinguish the city, the integration of urban and rural maintained a significant growth trend. The urban sequence-scale analysis of the extraction results of areas, rural, and accurately obtain the city internal elements distribution, morphology and structure the four phases (2000, 2005, 2010, and 2015) is performed by regarding the urban area of China’s information; these extraction results can play an important role in urban development. prefecture level as the city scale index. 4.2. AnalysisAfter abandoning of City Size the Distribution unreliable tail of the ranks [61], the results in Figure 7 show that the model fitting coefficient R2 of each period is above 0.97, which indicates that the model fitting degree is 2 2 2 better.The Moreover, total area the of order-scale Chinese cities method was 16,012.84can better km describein 2000, the 21,200.41distribution km ofin Chinese 2005, 27,778.11 city size. kmThe 2 scalein 2010, of the and first 34,554.31 cities in km eachin period 2015, is thereby smaller indicating than the theoretical an increase value. of 32.40%, The scale 31.04%, of the andtop cities 24.39%. is relativelyThe area in close, 2015 isthereby 2.16 times indicating that which that existedthe firs int degree 2000. This of China’s result shows urban that system China’s is urbaninsufficiently growth prominent,has maintained and athus, significant not a typical growth first trend. distribution. The urban The sequence-scale values of the analysis Zipf index of the in extraction2000, 2005, results 2010, andof the 2015 four are phases 0.82, 0.86, (2000, 0.85, 2005, and 2010, 0.82, andrespectively. 2015) is performed The fractal bydimension regarding D thevalues urban are area1.19, of1.13, China’s 1.15, andprefecture 1.18, respectively, level as the thereby city scale indicating index. that the scale distribution of Chinese cities is better than the ideal AfterZipf abandoningdistribution. the In unreliableterms of decentralization, tail of the ranks [ 61the], thescale results of high-level in Figure 7cities show is that insufficiently the model 2 prominent,fitting coeffi andcient small R of and each medium-sized period is above cities 0.97, whichare relatively indicates developed. that the model The distribution fitting degree of is urban better. scalesMoreover, is relatively the order-scale small, and method the scale can bettersystem describe of cities the is distributionmore balanced. of Chinese city size. The scale of the firstIn 2000–2015, cities in each the period gradual is smaller upward than trend the of theoretical the structural value. capacity The scale over of the time top indicates cities is relatively that the overallclose, thereby scale of indicating Chinese cities that the continues first degree to expand of China’s and urban the urban system system is insu isffi becomingciently prominent, increasingly and complex.thus, not In a typicalthe existing first distribution.research on the The evolution values of of theurban Zipf land index use in scale 2000, distribution 2005, 2010, in and China 2015 from are 19900.82, to 0.86, 2000, 0.85, the andscale 0.82, distribution respectively. of urban The land fractal use dimension in the country D values exhibits are a 1.19,decline 1.13, in the 1.15, Zipf and index 1.18, andrespectively, a trend of thereby increasing indicating equilibrium. that the The scale research distribution findings of Chinese(Figure 8) cities from is 2000 better to than2015 theindicate ideal Zipfthat thedistribution. value of the In Zipf terms index of decentralization, of China’s urban the system scaleof presented high-level a rising cities istrend insu andfficiently then decreased prominent, with and time,small from and medium-sized0.82 in 2000 to a cities maximum are relatively of 0.86 developed.in 2005, andThe then distribution dropped to of0.82 urban in 2015. scales Additionally, is relatively thesmall, evolution and the of scale the city system size of distribution cities is more exhibited balanced. different phase characteristics in 2005.

4 4 y = -0.8207x + 3.2254 y = -0.8595x + 3.4063 3.5 R² = 0.9761 3.5 R² = 0.9752 q=0.8207 q=0.8595 3 3 D=1.1894 D=1.13 2.5 2.5

2 2

1.5 1.5

1 1 00.511.522.5 00.511.522.5 ISPRS Int. J. Geo-Inf. 2019, 8(,a x) FOR2000 PEER REVIEW (b) 2000 14 of 24

4 4 y = -0.847x + 3.507 y = -0.8219x + 3.5644 3.5 R² = 0.9736 3.5 R² = 0.9732 q=0.8470 q=0.8219 3 3 D=1.14 D=1.18 2.5 2.5

2 2

1.5 1.5

1 1 00.511.522.5 00.511.522.5 (c) 2000 (d) 2000

FigureFigure 7. 7. Cities'Cities’ rank-size rank-size plots plots of of China China for for 2000, 2000, 2005, 2005, 2010, 2010, 2015. 2015. The The abscissa abscissa (x) (x) is is lg(R), lg(R), where where R R standsstands for for the the city city ranks ranks of of top200. top200. The The abscissa abscissa (x) (x) is is lg(R), lg(R), where where R R st standsands for for the the city city rank rank of of top200. top200. TheThe vertical vertical coordinates coordinates (y) is lg(S), where S stands for ci cityty size, namely urban area.

3.70 0.87 0.86 3.60 3.56 0.86 3.51 3.50 0.85 3.41 0.85 3.40 0.84

3.30 0.83 3.23 3.20 0.82 0.82 0.82 Structural Capacity Zipf Index 3.10 0.81 2000 2005 2010 2015

Figure 8. Change of Zipf Index and structural capacity in China’s cities system from 2000–2015.

China implemented the urban development policy of “strictly controlling the scale of large cities and highlighting the development of small towns,” combined with its urban development policy from 1996 to 2000, and implemented the diversified urban development policy of “coordinated development of large, medium, and small cities and towns” in 2000–2005 (Table 2).

Table 2. The policies and characteristics of urban development during different periods.

Period Policy Characteristics Rapid and extensive 2000– Strictly controlling the scale of large cities and development of large 2005 highlighting the development of small towns cities 2005– Development of large Intensive sustainable development of urbanization 2010 cities is constrained Taking the big cities as the basis, focusing on small and medium-sized cities, gradually forming a group of cities Coordinated development 2010– with large radiation effects, and promoting the of large, medium, and 2015 coordinated development of large, medium, and small small cities and towns cities and towns Policy changes have led to the rapid and extensive development of large cities with natural advantages. The results of this study for 2000–2005 indicate that the expansion area of the top 10 large cities and megacities accounted for 31.50% of the total expansion area, and the expansion area of three

ISPRS Int. J. Geo-Inf. 2019, 8, x FOR PEER REVIEW 14 of 24

4 4 y = -0.847x + 3.507 y = -0.8219x + 3.5644 3.5 R² = 0.9736 3.5 R² = 0.9732 q=0.8470 q=0.8219 3 3 D=1.14 D=1.18 ISPRS Int. J. Geo-Inf. 2019, 8, 241 14 of 23 2.5 2.5

2 2

1.5 In 2000–2015, the gradual upward trend of the1.5 structural capacity over time indicates that the overall scale of Chinese cities continues to expand and the urban system is becoming increasingly 1 1 complex. In the existing research on the evolution of urban land use scale distribution in China from 00.511.522.5 00.511.522.5 1990 to 2000, the scale distribution of urban land use in the country exhibits a decline in the Zipf index and a trend of increasing(c) 2000 equilibrium. The research findings (Figure8) from (d) 2000 2000 to 2015 indicate that the value of the Zipf index of China’s urban system presented a rising trend and then decreased with Figure 7. Cities' rank-size plots of China for 2000, 2005, 2010, 2015. The abscissa (x) is lg(R), where R time, from 0.82 in 2000 to a maximum of 0.86 in 2005, and then dropped to 0.82 in 2015. Additionally, stands for the city ranks of top200. The abscissa (x) is lg(R), where R stands for the city rank of top200. the evolution of the city size distribution exhibited different phase characteristics in 2005. The vertical coordinates (y) is lg(S), where S stands for city size, namely urban area.

3.70 0.87 0.86 3.60 3.56 0.86 3.51 3.50 0.85 3.41 0.85 3.40 0.84

3.30 0.83 3.23 3.20 0.82 0.82 0.82 Structural Capacity Zipf Index 3.10 0.81 2000 2005 2010 2015

Figure 8. ChangeChange of of Zipf Zipf Index Index and structural capa capacitycity in China’s citi citieses system from 2000–2015.

China implemented the urban development policy of “strictly controlling the scale of large cities and highlightinghighlighting thethe development development of of small small towns,” towns, combined” combined with with its urbanits urban development development policy policy from 1996from to1996 2000, to and 2000, implemented and implemented the diversified the diversif urbanied development urban development policy of “coordinated policy of “coordinated development developmentof large, medium, of large, and medium, small cities and and small towns” cities in and 2000–2005 towns” (Tablein 2000–20052). (Table 2).

TableTable 2. The policies and characteristics of urban development during didifferentfferent periods.

PeriodPeriod Policy Policy Characteristics Characteristics Strictly controlling the scale of large cities and highlighting the RapidRapid and and extensive extensive 2000–2000–2005 Strictly controlling the scale of large cities and development of small towns development of of large large cities 2005 highlighting the development of small towns Developmentcities of large cities 2005–2010 Intensive sustainable development of urbanization 2005– Developmentis constrained of large Intensive sustainable development of urbanization 2010 Taking the big cities as the basis, focusing on small and cities is constrained Coordinated development medium-sized cities, gradually forming a group of cities with 2010–2015 Taking the big cities as the basis, focusing on small and of large, medium, and large radiation effects, and promoting the coordinated medium-sized cities, gradually forming a group of cities Coordinatedsmall cities development and towns 2010– development of large, medium, and small cities and towns with large radiation effects, and promoting the of large, medium, and 2015 coordinated development of large, medium, and small small cities and towns Policy changes have led tocities the rapidand towns and extensive development of large cities with natural advantages.Policy changes The results have of thisled to study the forrapid 2000–2005 and extensive indicate development that the expansion of large area cities of the with top 10natural large advantages.cities and megacities The results accounted of this study for 31.50% for 2000–2005 of the total indi expansioncate that the area, expansion and the expansionarea of the areatop 10 of large three 2 citiesmegacities and megacities (area greater accounted than 500 for km 31.50%) accounted of the total for expansion a total expansion area, and area the ofexpansion 12.48%. area During of three this period, the top cities in China achieved remarkable development that considerably exceeded those of small and medium-sized cities. The first place rose, equilibrium declined, and the city size distribution was closer to the Czech Republic model. In 2005, China proposed to follow the urbanization strategy of intensive development, which constrained the development of large cities to a certain extent. In the next 5 years, the proportion of ISPRS Int. J. Geo-Inf. 2019, 8, 241 15 of 23 the top 10 cities expanded to 20.06%, and the top 3 cities were ranked. The proportion of urban areas decreased to 8.46%, the first degree of urban systems declined, equilibrium increased, and the Zipf index decreased. After 2011, China adhered to the policy of “taking the big cities as the basis, focusing on small and medium-sized cities, gradually forming a group of cities with large radiation effects, and promoting the coordinated development of large, medium, and small cities and towns.” The top 10 cities in 2000–2015 accounted for an expanded area. The ratio of the urban expansion area to the top 3 areas was further reduced to 5.80%. The first degree of the urban system and the Zipf index further declined, and the Chinese urban system became more balanced.

4.3. Analysis of Urban Expansion Characteristics In 2015, the total urban area of prefecture-level cities reached 34,554.31 km2, accounting for 2.17% of the total administrative area of the country and exhibiting an increase of 115.79% compared with that in 2000. The spatiotemporal expansion process of urban areas at the prefecture level and above is analyzed in four scales: cities, provinces, urban agglomerations, and economic zones. Figure9 shows the detailed information concerning the spatiotemporal distribution characteristics of China’s cities. The analysis results indicate that among the top five cities in terms of urban area, Beijing, Shanghai, Tianjin, and Guangzhou occupied the top four places from 2000 to 2015. Wuhan ranked fifth in 2000, 2005, and 2010, but was replaced by Shenzhen in 2015. As the capital of China, Beijing ranks third in terms of urban area. It maintained an expansion intensity of 2.04% from 2000 to 2015, with a total expansion of 241.89 km2, which is equivalent to one-third of the urban area in Beijing in 2000. Shanghai’s urban area expanded more rapidly. In the past 15 years, its expansion strength of 6.44% expanded by 760.44 km2, which is nearly double compared with its urban area of 787.31 km2 in 2000. Tianjin has more than doubled in 15 years, and Guangzhou and Wuhan have expanded by 60%. The urban area of Chongqing in 2015 is nearly twice as large as it in 2000. In February 2010, the National Urban System Plan issued by the Ministry of Housing and Urban–Rural Development in Beijing identified Beijing, Tianjin, Guangzhou, and Chongqing as China’s national central cities. At the national level, the aforementioned six cities are guaranteed to lead the cities in China. The collection and distribution functions can obtain the positioning of the national central city and is highly significant for the development of the region. The results of this study show that the central cities of the five major countries have expanded by 2,295.51 km2 in 15 years, i.e., an increase of 82.45%. In addition, the five national central cities are the top five cities in the 2015 urban area rankings. The central cities of Chengdu, Wuhan, Zhengzhou, and Xi’an, which were identified later, ranked in the top 10, except for Zhengzhou. This result confirmed the practical guiding significance of the research results for the urban system planning of the country’s comprehensive understanding of urban expansion. The top three provinces with the largest urban area in 2015 were Jiangsu, Shandong, and Guangdong. The three provinces with the smallest urban areas were Tibet, Qinghai, and Hainan. The urban area of Hainan Province is small, but its urban area is high. In 2000–2015, the top three provinces with the largest urban expansion area and rate were Jiangsu, Shandong, and Guangdong. The three provinces with the smallest expansion area and speed were Tibet, Qinghai, and Hainan. The expansion rates of Jiangsu, which has the largest expansion area, and Tibet, which has the smallest expansion area, differ by more than 50 times. The top three provinces with the highest expansion intensity were Ningxia, Jiangxi, and Jiangsu, and their expansion strengths were 15.88%, 14.48%, and 13.32%, respectively. The latter three provinces and cities with smaller expansion intensity were Beijing, Henan, and Tibet, with expansion strengths of 2.04%, 3.04%, and 3.33%, respectively. ISPRS Int. J. Geo-Inf. 2019, 8, 241 ISPRS Int. J. Geo-Inf. 2019, 8, x FOR PEER REVIEW 17 of 24 16 of 23

FigureFigure 9. 9.Spatiotemporal Spatiotemporal distributiondistribution map map of of urban urban expansion expansion of Chin of China’sa’s cities cities in 2000–2015. in 2000–2015. Urban Urban area and area expansion and expansion area unit: area km unit:2, Expansion km2, Expansion speed unit: speed km2 per unit: km2 per year. year.

ISPRS Int. J. Geo-Inf. 2019, 8, 241 17 of 23

The urban areas of the four regions (Northeast China, East China, Central China, and West China) increased in stages based on the analysis of different economic zones. In 2015, the urban area of the eastern region was the largest, whereas that of the northeast region was the smallest. In terms of urban area, the largest in the eastern region was 3.79% in 2015 and the lowest in the western region was 0.92%, thereby indicating that the land urbanization level in the western region is still the lowest in the four regions of the country. From 2000 to 2015, the expansion area and rate of each region were ranked from high to low as follows: eastern region, western region, central region, and northeastern region. The extended area in the eastern region was 8356.14 km2, and the expanded area in the northeastern region was 3563.97 km2. From 2000 to 2015, the analysis of different economic zones showed that the expansion intensity of each region was ranked from high to low as follows: western region, eastern region, northeastern region, and central region. The expansion intensity in the eastern region reached 8.61%, whereas the expansion intensity in the central region was only 6.67%, which was the slowest growth in the area.

4.4. Analysis of Urban Expansion Type The convex hull analysis method is used to determine and analyze the types of urban expansion in China. Figure 10a shows the expansion type of China’s 31 provincial capitals. Figure 10b presents the specific percent of area in or out of the convex hull for China’s provincial capital cities. From 2000 to 2015, China’s capital cities of infilling type included Lanzhou, Guangzhou, Shanghai, Tianjin, Urumqi, Shijiazhuang, Guiyang, Hangzhou, Taiyuan, Jinan, Lhasa, and Wuhan. From 2000 to 2015, China’s edge-expansion provincial capital cities included Harbin, Xi’an, Hefei, Yinchuan, Nanning, Kunming, Fuzhou, Nanchang, Nanjing, Chongqing, Changsha, Xining, Zhengzhou, Chengdu, Hohhot, Shenyang, Haikou, Beijing, and Changchun. Harbin is the most extensive city and Lanzhou is the most intensive one. Some megacities, such as Guangzhou, Tianjin and Shanghai, also lead the way in terms of intensive development. But, the urban areas of Chongqing are widely dispersed geographically. From 2000 to 2015, the number of infilling type cities accounted for 26.7% and included Taiyuan City, Jiamusi City, Yichang City, Baishan City, Panjin City, Yanbian Korean Autonomous Prefecture, Shiyan City, Liaoyang City, Lanzhou City, Zaozhuang City, Datong City, and other 90 cities, including Jinzhou City, Baoji City, and Luoyang City. The proportion of extended cities accounted for 73.3% and included Turpan, Fuyang, Bortala Mongolian Autonomous Prefecture, Beijing, Tianjin, Huizhou, Fangchenggang, Wuzhong, Lu’an, Hebi, Qingyuan, Xiamen, and 247 cities, including Zhongwei City. From 2000 to 2015, the expansion of urban space at the prefecture level and above was primarily epitaxial, the urban form tended to be scattered, and urban land use efficiency declined. Moreover, all of the cities in edge-expansion type are Fujian, Hainan, Jiangsu, Ningxia and Anhui. The most intensive province is Jilin and Tibet, Shandong, Shanxi and Heilongjiang are also more intensive than extensive. And 61.11% of the cities in Northeast China are infilling type, 31.33% for Central China, 19.39% for East China and 18.80% for West China. ISPRS Int. J. Geo-Inf. 2019, 8, 241 18 of 23 ISPRS Int. J. Geo-Inf. 2019, 8, x FOR PEER REVIEW 19 of 24

(a) (b)

Figure 10. Expansion types of capital cities in ChinaFigure from 10.2000Expansion to 2015. types of capital cities in China from 2000 to 2015.

ISPRS Int. J. Geo-Inf. 2019, 8, 241 19 of 23

4.5.ISPRS Analysis Int. J. Geo-Inf. of Land 2019 Use, 8, x Change FOR PEER in REVIEW UrbanSprawl Areas 20 of 24 4.5.From Analysis 2000 of Land to 2015, Use Change 11,106.45 in Urban km2 Sprawlcultivated Areas land, 1906.27 km2 forest and grassland, and 2 1992.55From km 2000water to area2015, and11106.45 unutilized km2 cultivated land have land, been 1906.27 converted km2 forest into and urban grassland, areas and (Figure 1992.55 11 ). Urbanizationkm2 water area has and destroyed unutilized forest land and have grassland been converted to the extentinto urban double areas the (Figure size of 11). the Urbanization Beijing urban area.has destroyed About 10% forest of the and sprawl grassland area to isthe at extent the cost double of forestthe size and of the grassland. Beijing urban More area. than About half 10% of the sprawlof the area sprawl consists area is of at cultivated the cost of land; forest water and areasgrassl andand. unutilizedMore than landhalf of have the alsosprawl suff areaered consists great losses. of Urbanizationcultivated land; of China water is areas at the and expense unutilized of nature land have and arablealso suffered land. Itgreat seriously losses. threatensUrbanization people’s of livelihoodsChina is at and the biodiversity expense of innature the urbanand arable landscape. land. It seriously threatens people’s livelihoods and biodiversityThe analysis in the was urban carried landscape. out with province as the unit. Urban sprawl in 28 provinces, except for Beijing,The analysis Tibet and was Guangzhou, carried out iswith dominated province byas the the unit. occupation Urban sprawl of cultivated in 28 provinces, land. The except expansion for ofBeijing, Beijing Tibet in this and period Guangzhou, is combined is dominated construction by the land occupation such as of rural cultivated residential land. areasThe expansion and occupied of cultivatedBeijing in land. this Theperiod urban is combined expansion construction of Tibet is mainly land such from as forest rural and residential grass land areas and and Guangdong occupied is a watercultivated area land. and unused The urban land. expansion of Tibet is mainly from forest and grass land and Guangdong is aFrom water 2000 area toand 2015, unused Zhejiang, land. Chongqing, Anhui, Henan, Jilin, Jiangsu, Shanghai, Ningxia, ShandongFrom and 2000 Shanxi to 2015, rank Zhejiang, the top 10Chongqing, provinces Anhui, for the Henan, area of cultivatedJilin, Jiangsu, land Shanghai, occupied Ningxia, by urban expansion.Shandong Tibet, and Shanxi Hunan, rank Hainan, the top Xinjiang, 10 provinces Inner Mongolia,for the area Guizhou, of cultivated Jiangxi, land Yunnan, occupied Guangxi by urban and Fujianexpansion. are the Tibet, top 10 Hunan, provinces Hainan, in terms Xinjiang, of land Inner area Mongolia, occupied Guizhou, by forest Jiangx and grass.i, Yunnan, The top Guangxi 10 provinces and Fujian are the top 10 provinces in terms of land area occupied by forest and grass. The top 10 in terms of occupied construction land area are Beijing, Shaanxi, Sichuan, Guangxi, Hebei, Qinghai, provinces in terms of occupied construction land area are Beijing, Shaanxi, Sichuan, Guangxi, Hebei, Hainan, Liaoning, Fujian And Shanxi. The top 10 provinces with occupied water area and unutilized Qinghai, Hainan, Liaoning, Fujian And Shanxi. The top 10 provinces with occupied water area and land area are: Guangdong, Gansu, Tianjin, Hubei, Guizhou, Xinjiang, Heilongjiang, Fujian, Ningxia unutilized land area are: Guangdong, Gansu, Tianjin, Hubei, Guizhou, Xinjiang, Heilongjiang, and Inner Mongolia. Fujian, Ningxia and Inner Mongolia.

Cultivated Land Forest and Grassland Water Area and Unutilized Land Construction Land

100.00% 90.00% 80.00% 70.00% 60.00% 50.00% 40.00% 30.00% 20.00% 10.00% 0.00% Zhejiang Chongqing Anhui Henan Jilin Jiangsu Shanghai Ningxia Shandong Shanxi Hebei Hubei Sichuan Jiangxi Tianjin Liaoning Yunnan Heilongjiang Mongolia Inner Fujian Shaanxi Qinghai Hunan Guizhou Beijing Guangxi Xinjiang Gansu Shandong Hainan Tibet

FigureFigure 11. 11.Proportion Proportion of of land land useuse occupiedoccupied by urba urbann sprawl sprawl areas areas of of China’s China’s cities cities 2000–2015. 2000–2015.

5.5. Conclusions Conclusions ThisThis study study proposes proposes a semi-automatic a semi-automatic method method for urbanfor urban area extractionarea extraction based onbased high-resolution on high- remoteresolution sensing remote imagery sensing and imagery geographic and informationgeographic information data-assisted data-assisted automatic automatic computer computer recognition andrecognition manual interpretation. and manual interpretation. It has obtained It 337has prefecture-levelobtained 337 prefecture-level cities in China cities in 2000, in China 2005, in and 2000, 2010. The2005, conclusions and 2010. based The conclusions on the high-precision based on the urban high-p arearecision of the urban fourth area phase of (2015)the fourth and thephase analyses (2015) of and the analyses of urban spatiotemporal patterns are as follows: urban spatiotemporal patterns are as follows: (1) From 2000 to 2015, China’s cities generally expanded rapidly and maintained an exponential (1) From 2000 to 2015, China’s cities generally expanded rapidly and maintained an exponential growth trend. By 2015, the urban area was equivalent to 2.16 times that in 2000. The urban areas are growth trend. By 2015, the urban area was equivalent to 2.16 times that in 2000. The urban areas mostly concentrated in the eastern and central regions, and the eastern and western regions are are mostly concentrated in the eastern and central regions, and the eastern and western regions are considerably different. However, the western region is continuously expanding. considerably different. However, the western region is continuously expanding.

ISPRS Int. J. Geo-Inf. 2019, 8, 241 20 of 23

(2) The overall scale of Chinese cities continues to expand, the urban system is becoming increasingly complex, the scale distribution is more dispersed than the ideal Zipf distribution, the scale of high-level cities is insufficiently prominent, and small and medium-sized cities are relatively developed. After 2005, the urban system has become increasingly balanced. (3) The urban areas of Beijing, Shanghai, Tianjin, and Guangzhou always occupied the top four places in 2000–2015. In 2015, Shenzhen ranked fourth and Chongqing ranked sixth. Five national central cities and China’s four major lines have been identified. The cities’ positioning remains the same. (4) In the study area, 73.30% of the cities exhibit expansive expansion, the urban form tends to be scattered, and land use efficiency is low. (5) The new urban areas mainly come from cultivated land and ecological land (forest, grassland, water area); the urbanization threatens people’s livelihoods and biodiversity. The proposed method needs high-resolution imagery and high-quality thematic data to ensure accuracy; however, the storage and management of so much data is a significant problem. This study only analyzes three aspects: urban size distribution, extended regional characteristics, and expansion types. The next step will be to combine urban, economic, population, land use, and other thematic data to investigate urban expansion coordination, land use efficiency, spatial form change, and land occupation type. A comprehensive analysis of other aspects will be conducted to explore other urban development rules in China.

Author Contributions: Hanchao Zhang conceived and designed the research; Hao Wang and Xiaogang Ning contributed materials; and Hanchao Zhang and Zhenfeng Shao wrote the paper. Funding: This work was supported in part by the National key R & D plan on strategic international scientific and technological innovation cooperation special project under Grant 2016YFE0202300, the National Natural Science Foundation of China under Grants 61671332, 41771452, 41401513, and 41771454, the Natural Science Fund of Hubei Province in China under Grant 2018CFA007, Fundamental Research Funds for the Central Public-interest Scientific Institution under Grant 7771803, Beijing Key Laboratory of Urban Spatial Information Engineering under Grant 2018202. Acknowledgments: This work was supported by National 973 Basic Research and Development Program project (No. 2012CB719904), The National Key Research and Development Program of China (No. 2016YFE0205300), National Natural Science Foundation of China (No. 41401513), National Science and Technology Specific Projects (No. 2012YQ1601850 and No. 2013BAH42F03), Program for New Century Excellent Talents in University (No. NCET-12-0426), Beijing Key Laboratory of Urban Spatial Information Engineering (No. 2018202). Conflicts of Interest: The authors declare no conflict of interest.

References

1. Li, D.; Zhao, X.; Li, X. Remote sensing of human beings—A perspective from nighttime light. Geo-Spat. Inf. Sci. 2016, 19, 69–79. [CrossRef] 2. Huang, D.Q.; Huang, J.; Liu, T. Delimiting urban growth boundaries using the clue-s model with village administrative boundaries. Land Use Policy 2019, 82, 422–435. [CrossRef] 3. Song, W.; Wu, K.N.; Zhao, H.F.; Zhao, R.; Li, T. Arrangement of high-standard basic farmland construction based on village-region cultivated land quality uniformity. Chin. Geogr. Sci. 2019, 29, 325–340. [CrossRef] 4. Xi, B.D.; Li, X.G.; Gao, J.X.; Zhao, Y.; Liu, H.L.; Xia, X.F.; Yang, T.X.; Zhang, L.Y.; Jia, X. Review of challenges and strategies for balanced urban-rural environmental protection in china. Front. Environ. Sci. Eng. 2015, 9, 371–384. [CrossRef] 5. Chen, M.; Liu, W.; Lu, D. Challenges and the way forward in china’s new-type urbanization. Land Use Policy 2016, 55, 334–339. [CrossRef] 6. Li, D.; Ma, J.; Cheng, T.; Genderen, J.L.V.; Shao, Z. Challenges and opportunities for the development of megacities. Int. J. Digit. Earth 2018, 1–14. [CrossRef] 7. Li, J.; Zhang, Z.; Ma, L.; Gu, Q.; Wang, K.; Xu, Z. Assessment on the impact of arable land protection policies in a rapidly developing region. ISPRS Int. J. Geo-Inf. 2016, 5, 69. [CrossRef] 8. Liu, S.; Wang, D.; Li, H.; Li, W.; Wu, W.; Zhu, Y. The ecological security pattern and its constraint on urban expansion of a black soil farming area in northeast china. ISPRS Int. J. Geo-Inf. 2017, 6, 263. 9. Li, D.; Cao, J.; Yao, Y. Big data in smart cities. Sci. Chin. Inf. Sci. 2015, 58, 179–190. [CrossRef] ISPRS Int. J. Geo-Inf. 2019, 8, 241 21 of 23

10. Shyamantha, S.; Ronald C., E.; Murayama, Y. Spatiotemporal analysis of urban growth using gis and remote sensing: A case study of the colombo metropolitan area, sri lanka. ISPRS Int. J. Geo-Inf. 2016, 5, 197. 11. Peng, J.; Tian, L.; Liu, Y.; Zhao, M.; Hu, Y.N.; Wu, J. Ecosystem services response to urbanization in metropolitan areas: Thresholds identification. Sci. Total Environ. 2017, 607–608, 706–714. [CrossRef] [PubMed] 12. Rimal, B.; Zhang, L.F.; Keshtkar, H.; Haack, B.N.; Rijal, S.; Zhang, P. Land use/land cover dynamics and modeling of urban land expansion by the integration of cellular automata and markov chain. ISPRS Int. J. Geo-Inf. 2018, 7, 21. [CrossRef] 13. Liu, Y.; Zeng, C.; Cui, H.T.; Song, Y.H. Sustainable land urbanization and ecological carrying capacity: A spatially explicit perspective. Sustainability 2018, 10, 16. [CrossRef] 14. Gong, J.; Peng, F.; Le, W.; Wu, H. Built-up area detection from a stereo pair using disparity information. Geomat. Inf. Sci. Wuhan Univ. 2016, 41, 427–433. 15. Zou, J.; Chen, Y.; Ding, G.; Xuan, W. A clustered threshold method for extracting urban built-up area using the dmsp/ols nighttime light images. Geomat. Inf. Sci. Wuhan Univ. 2016, 41, 196–201. 16. Anderson, K.; Ryan, B.; Sonntag, W.; Kavvada, A.; Friedl, L. Earth observation in service of the 2030 agenda for sustainable development. Geo-Spat. Inf. Sci. 2017, 20, 77–96. [CrossRef] 17. Li, D.; Yao, Y.; Shao, Z.; Wang, L. From digital earth to smart earth. Chin. Sci. Bull. 2014, 59, 722–733. [CrossRef] 18. Schneider, A.; Friedl, M.A.; Potere, D. Mapping global urban areas using modis 500-m data: New methods and datasets based on ‘urban ecoregions’. Remote Sens. Environ. 2010, 114, 1733–1746. [CrossRef] 19. Bagan H, Y.Y. Landsat analysis of urban growth: How tokyo became the world’s largest megacity during the last 40 years. Remote Sens. Environ. 2012, 127, 210–222. [CrossRef] 20. Li, X.; Peng, G.; Lu, L. A 30-year (1984–2013) record of annual urban dynamics of beijing city derived from landsat data. Remote Sens. Environ. 2015, 166, 78–90. [CrossRef] 21. Li, M.; Zang, S.; Wu, C.; Tian, Y. Spatial and temporal variation and its driving forces of urban impervious surface in urban-rural continuum of harbin. Acta Geogr. Sin. 2017, 72, 105–115. 22. Woodcock, C.E. Compact, dispersed, fragmented, extensive? A comparison of urban growth in twenty-five global cities using remotely sensed data, pattern metrics and census information. Urban Stud. 2014, 45, 659–692. 23. Griffiths, P.; Hostert, P. Mapping megacity growth with multi-sensor data. Remote sens environ. Remote Sens. Environ. 2010, 114, 426–439. [CrossRef] 24. Bo, W.; Guo, Q.; Fang, F.; Su, Y.; Wang, R. Mapping us urban extents from modis data using one-class classification method. Remote Sens. 2015, 7, 10143–10163. 25. Corresponding, E.B.; Belward, A.S. Glc2000: A new approach to global land cover mapping from earth observation data. Int. J. Remote Sens. 2005, 26, 1959–1977. 26. Rahman, M. Detection of land use/land cover changes and urban sprawl in al-khobar, saudi arabia: An analysis of multi-temporal remote sensing data. ISPRS Int. J. Geo-Inf. 2016, 5, 15. [CrossRef] 27. Di Palma, F.; Amato, F.; Nolè, G.; Martellozzo, F.; Murgante, B. A smap supervised classification of landsat images for urban sprawl evaluation. ISPRS Int. J. Geo-Inf. 2016, 5, 109. [CrossRef] 28. Guindon, B.; Ying, Z.; Dillabaugh, C. Landsat urban mapping based on a combined spectral–spatial methodology. Remote Sens. Environ. 2004, 92, 218–232. [CrossRef] 29. Gao, F.; Colstoun, E.B.D.; Ma, R.; Weng, Q.; Masek, J.G.; Chen, J.; Pan, Y.; Song, C. Mapping impervious surface expansion using medium-resolution satellite image time series: A case study in the yangtze river delta, china. Int. J. Remote Sens. 2012, 33, 7609–7628. [CrossRef] 30. Sun, Z.; Wang, C.; Guo, H.; Shang, R. A modified normalized difference impervious surface index (mndisi) for automatic urban mapping from landsat imagery. Remote Sens. 2017, 9, 942. [CrossRef] 31. Son, N.T.; Chen, C.F.; Chang, L.Y.; Chen, C.R.; Thanh, B.X. Urban growth mapping from landsat data using linear mixture model in ho chi minh city, vietnam. J. Appl. Remote Sens. 2012, 6, 100–106. 32. Angiuli, E.; Trianni, G. Urban mapping in landsat images based on normalized difference spectral vector. IEEE Geosci. Remote Sens. Lett. 2013, 11, 661–665. [CrossRef] 33. Wang, L.; LI, C.C.; Ying, Q.; Cheng, X.; Wang, X. China’s urban expansion from 1990 to 2010 determined with satellite remote sensing. Sci. Bull. 2012, 57, 2802–2812. [CrossRef] 34. Chen, H. Automatic urban area extraction using a gabor filter and high-resolution remote sensing imagery. Geomat. Inf. Sci. Wuhan Univ. 2013, 38, 1063–1067. ISPRS Int. J. Geo-Inf. 2019, 8, 241 22 of 23

35. Tao, C.; Tan, Y.; Cai, H.; Bo, D.U.; Tian, J. Object-oriented method of hierarchical urban building extraction from high-resolution remote-sensing imagery. Acta Geod. Cartogr. Sin. 2016, 39, 39–45. 36. Huan, N.I.; Lin, X.; Ning, X. A method for extracting human settlements from remote sensing image using right angle corners features. Geomat. Inf. Sci. Wuhan Univ. 2017, 42, 648–655. 37. Ning, X.; Lin, X. An index based on joint density of corners and line segments for built-up area detection from high resolution satellite imagery. ISPRS Int. J. Geo-Inf. 2017, 6, 338. [CrossRef] 38. Lin, X.; Ning, X. Extraction of human settlements from high resolution remote sensing imagery by fusing features of right angle corners and right angle sides. Acta Geod. Cartogr. Sin. 2017, 46, 83–89. 39. Guan, X.; Shuai, L.; Jie, B.; Fei, W.; Li, Z.; Qiang, W.; He, J.; Chen, T. Urban land-use classification by combining high-resolution optical and long-wave infrared images. Geo-Spat. Inf. Sci. 2017, 20, 299–308. [CrossRef] 40. Xin, H.; Li, J.; Liao, W.; Chanussot, J. Information extraction from remote sensing imagery. Geo-Spat. Inf. Sci. 2017, 20, 297–298. 41. Jiao, L.; Wu, S. Analyzing the characteristics of the expansion of the metropolises in china from 1990 to 2010 using self-organizing neural network. Geomat. Inf. Sci. Wuhan Univ. 2014, 39, 1435–1440. 42. Calegari, G.; Carlino, E.; Peroni, D.; Celino, I. Extracting urban land use from linked open geospatial data. ISPRS Int. J. Geo-Inf. 2015, 4, 2109–2130. [CrossRef] 43. Song, X.P.; Huang, C.; Townshend, J.R. Improving global land cover characterization through data fusion. Geo-Spat. Inf. Sci. 2017, 20, 1–10. [CrossRef] 44. Li, Y.; Shao, H.; Jiang, N.; Shi, G.; Cheng, X. The evolution of the urban spatial pattern in the yangtze river economic belt: Based on multi-source remote sensing data. Sustainability 2018, 10, 21. [CrossRef] 45. Liu, X.; Hu, G.; Chen, Y.; Li, X.; Xu, X.; Li, S.; Pei, F.; Wang, S. High-resolution multi-temporal mapping of global urban land using landsat images based on the google earth engine platform. Remote Sens. Environ. 2018, 209, 227–239. [CrossRef] 46. Keeratikasikorn, C. A comparative study on four major cities in northeastern thailand using urban land density function. Geo-Spat. Inf. Sci. 2018, 21, 93–101. [CrossRef] 47. Hao, G.; Simwanda, M.; Murayama, Y. An internet-based gis platform providing data for visualization and spatial analysis of urbanization in major asian and african cities. ISPRS Int. J. Geo-Inf. 2017, 6, 257. 48. Song, X.P.; Sexton, J.O.; Huang, C.; Channan, S.; Townshend, J.R. Characterizing the magnitude, timing and duration of urban growth from time series of landsat-based estimates of impervious cover. Remote Sens. Environ. 2016, 175, 1–13. [CrossRef] 49. Lu, H.; Zhang, M.; Sun, W.; Li, W. Expansion analysis of yangtze river delta urban agglomeration using dmsp/ols nighttime light imagery for 1993 to 2012. ISPRS Int. J. Geo-Inf. 2018, 7, 52. [CrossRef] 50. Chang, S.; Jiang, Q.; Wang, Z.; Xu, S.; Jia, M. Extraction and spatial–temporal evolution of urban fringes: A case study of changchun in jilin province, china. ISPRS Int. J. Geo-Inf. 2018, 7, 241. [CrossRef] 51. Cai, J.; Bo, H.; Song, Y. Using multi-source geospatial big data to identify the structure of polycentric cities. Remote Sens. Environ. 2017, 202, 210–221. [CrossRef] 52. Alsaaideh, B.; Tateishi, R.; Dong, X.P.; Hoan, N.T.; Al-Hanbali, A.; Bai, X. New urban map of eurasia using modis and multi-source geospatial data. Geo-Spat. Inf. Sci. 2017, 20, 29–38. [CrossRef] 53. Zhang, J.; Liu, J.; Zhai, L.; Hou, W. Implementation of geographical conditions monitoring in beijing-tianjin-hebei, china. ISPRS Int. J. Geo-Inf. 2016, 5, 89. [CrossRef] 54. Luckstead, J.; Devadoss, S. A comparison of city size distributions for china and india from 1950 to 2010. Econ. Lett. 2014, 124, 290–295. [CrossRef] 55. Fang, C.L.; Wang, Z.B. Quantitative diagnoses and comprehensive evaluations of the rationality of chinese urban development patterns. Sustainability 2015, 7, 3859–3884. [CrossRef] 56. Kinoshita, T.; Kato, E.; Iwao, K.; Yamagata, Y. Investigating the rank-size relationship of urban areas using land cover maps. Geophys. Res. Lett. 2008, 35, 4. [CrossRef] 57. Chen, Z.H.; Fu, S.H.; Zhang, D.Y. Searching for the parallel growth of cities in china. Urban Stud. 2013, 50, 2118–2135. [CrossRef] 58. Fang, L.; Zhang, Z.; Xiao, W. Forms of urban expansion of chinese municipalities and provincial capitals, 1970s–2013. Remote Sens. 2016, 8, 930. 59. Zhou, Y.; Tu, M.; Wang, S.; Liu, W. A novel approach for identifying urban built-up area boundaries using high-resolution remote-sensing data based on the scale effect. ISPRS Int. J. Geo-Inf. 2016, 7, 135. [CrossRef] ISPRS Int. J. Geo-Inf. 2019, 8, 241 23 of 23

60. Zhang, H.; Ning, X.; Wang, H.; Shao, Z. High accuracy urban expansion monitoring and analysis of china’s provincial capitals from 2000 to 2015 based on high-resolution remote sensing imagery. Acta Geogr. Sin. 2018, 73, 81–99. 61. Li, D.; Yu, H.; Li, X. The spatial-temporal pattern analysis of city development in countries along the belt and road initiative based on nighttime light data. Geomat. Inf. Sci. Wuhan Univ. 2017, 42, 711–720.

© 2019 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 (http://creativecommons.org/licenses/by/4.0/).