Environ. Res. Lett. 10 (2015) 034002 doi:10.1088/1748-9326/10/3/034002 102535 LETTER A new urban landscape in East–Southeast Asia, 2000–2010 OPEN ACCESS A Schneider1, C M Mertes1, A J Tatem2,3, B Tan4, D Sulla-Menashe5, S J Graves6, N N Patel7, J A Horton1, RECEIVED 16 May 2014 A E Gaughan8, J T Rollo1, I H Schelly1, F R Stevens8 and A Dastur9 1 Center for Sustainability and the Global Environment, Nelson Institute for Environmental Studies, and Department of Geography, REVISED 22 January 2015 University of Wisconsin-Madison, USA 2 Department of Geography and Environment, University of Southampton, UK ACCEPTED FOR PUBLICATION 3 27 January 2015 Fogarty International Center, National Institutes of Health, Bethesda, Maryland, USA 4 NASA Goddard Space Flight Center/Science Systems and Applications Inc., Lanham, Maryland, USA PUBLISHED 5 Department of Earth and Environment, Boston University, Boston, Massachusetts USA 3 March 2015 6 School of Forest Resources and Conservation, University of Florida-Gainesville, USA 7 Department of Geography and Geoinformation Science, George Mason University, Fairfax, Virginia, USA Content from this work 8 Department of Geography and Geosciences, University of Louisville, Kentucky, USA may be used under the 9 World Bank, Washington DC, USA terms of the Creative Commons Attribution 3.0 E-mail: aschneider4@wisc.edu licence. Any further distribution of Keywords: urbanization, urban sprawl, land cover change, remote sensing, change detection, urban density, population density this work must maintain attribution to the author (s) and the title of the work, journal citation and DOI. Abstract East–Southeast Asia is currently one of the fastest urbanizing regions in the world, with countries such as China climbing from 20 to 50% urbanized in just a few decades. By 2050, these countries are pro- jected to add 1 billion people, with 90% of that growth occurring in cities. This population shift paral- lels an equally astounding amount of built-up land expansion. However, spatially-and temporally- detailed information on regional-scale changes in urban land or population distribution do not exist; previous efforts have been either sample-based, focused on one country, or drawn conclusions from datasets with substantial temporal/spatial mismatch and variability in urban definitions. Using con- sistent methodology, satellite imagery and census data for >1000 agglomerations in the East–South- east Asian region, we show that urban land increased >22% between 2000 and 2010 (from 155 000 to 189 000 km2), an amount equivalent to the area of Taiwan, while urban populations climbed >31% (from 738 to 969 million). Although urban land expanded at unprecedented rates, urban populations grew more rapidly, resulting in increasing densities for the majority of urban agglomerations, includ- ing those in both more developed (Japan, South Korea) and industrializing nations (China, Vietnam, Indonesia). This result contrasts previous sample-based studies, which conclude that cities are uni- versally declining in density. The patterns and rates of change uncovered by these datasets provide a unique record of the massive urban transition currently underway in East–Southeast Asia that is impacting local-regional climate, pollution levels, water quality/availability, arable land, as well as the livelihoods and vulnerability of populations in the region. 1. Introduction hydrology, and biogeochemical cycles that extend beyond municipal boundaries (Seto et al 2010). While We have entered the urban era: cities now form the remote sensing has proven especially useful for basis of the human experience for the majority of the characterizing broad-scale land changes, detailed Earth’s population (UN 2012). Cities today must meet monitoring of urban land use change remains costly the needs of growing populations and expanding and challenging due to the highly heterogeneous economies, while at the same time minimizing their nature of cities, the spectral similarity between new environmental impacts (Grimm et al 2008, Montgom- urban land and other land cover types, and the lack of ery 2008). Expansion of built-up land is often the most cloud-free data in locations where estimates are most direct environmental impact associated with urban needed (e.g. tropics, Mertes et al 2015). As a result, growth, with far-reaching implications for climate, there has been little information on the building boom © 2015 IOP Publishing Ltd Environ. Res. Lett. 10 (2015) 034002 A Schneider et al 113°0'0"E 113°30'0"E 114°0'0"E 139°30'0"E 140°0'0"E 121°0'0"E 121°30'0"E Agglomerations 36°30'0"N Urban land 2000 31°30'0"N Urban expansion, 2000-2010 23°0'0"N Non-urban land 36°0'0"N Water bodies Agglomeration 31°0'0"N boundary 22°30'0"N 0 50 Kilometers 35°30'0"N Study region (right) 30°30'0"N • Agglomerations 1 Sample city location 116°0'0"E 116°30'0"E 100°0'0"E 100°30'0"E 101°0'0"E 101°30'0"E 102°0'0"E 117°0'0"E 117°30'0"E 106°30'0"E 107°0'0"E 14°30'0"N 40°30'0"N 3°30'0"N 39°30'0"N 6°0'0"S 14°0'0"N 40°0'0"N 39°0'0"N 3°0'0"N 6°30'0"S 13°30'0"N 39°30'0"N 38°30'0"N 2°30'0"N 7°0'0"S 120°0'0"E 120°30'0"E 121°0'0"E 121°30'0"E 126°30'0"E 127°0'0"E 127°30'0"E 103°30'0"E 104°0'0"E 104°30'0"E 105°30'0"E 106°0'0"E 21°30'0"N 38°0'0"N 15°0'0"N 30°30'0"N 31°0'0"N 37°30'0"N 21°0'0"N 14°30'0"N 30°30'0"N 30°0'0"N 37°0'0"N 20°30'0"N 29°30'0"N 14°0'0"N 30°0'0"N Figure 1. Maps of urban land extent and urban expansion for 13 of the 30 largest urban agglomerations in East–Southeast Asia. Agglomerations are labeled by largest city (see table A5 for a list of cities within each agglomeration). Note that the scale is held constant across all urban agglomerations. that is accompanying population growth in many buildings) with >50% coverage of a landscape unit (here, a 250 m pixel). We synthesize this information developing countries (China, India, etc) other than with population density maps developed using demo- case-study analysis of individual cities (Schneider and graphic data at the finest administrative unit available Woodcock 2008, Angel et al 2011), or country-level and empirically-tested population-land cover rela- assessments (Liu et al 2005, Wang et al 2012). Compar- tionship-based methods (Tatem et al 2007). To ing urban populations has also been notoriously address the issue of comparability, we conduct our difficult due to differences in census timing, data analysis of regional urbanization trends using the availability/quality, and most critically, the consider- urban agglomeration as the unit of analysis. We per- able variability in how cities are defined, whether by form a comparative analysis to understand within- population threshold, functional area, or administra- nation and across-nation trends in East–Southeast tive boundaries (Cohen 2004). One of the few studies Asia (figure 1, A1) recognizing that such a regional reporting transnational urban land and population approach cannot account for each city’s circumstances trends concluded that cities are universally spreading or individual drivers/impacts. Our results likely pro- out and declining in density (Angel et al 2005). While duce a conservative estimate of urban change in the there is evidence to contradict this in East Asia region, and may differ from ‘official’ statistics (World (Murakami et al 2005, Bagan and Yamagata 2012), Bank 2015) as a result of necessary choices regarding there has been no systematic way to compare trends definitions, spatial scale, and data sources. Our aim is across cities, nations, or regions. not to replace national estimates, but to offer a con- To describe urban trajectories across East and sistent approach for regional comparability of all cities Southeast Asia10 systematically, we characterize urban >100 000 in the region. extent and urban expansion 2000–2010 using Moder- ate Resolution Imaging Spectroradiometer (MODIS) satellite observations (Mertes et al 2015). In these 2. Background maps, urban land refers to places dominated by the ‘built environment’, which includes all non-vegeta- Great strides have been made to map population tive, human-constructed elements (e.g. roads, distribution using consistent data and methods (Balk et al 2006, Tatem et al 2007), but they depict 10 East Asia includes China (including Hong Kong SAR, Macao population as measured at one point in time, and at SAR), Taiwan, Japan, Mongolia, North Korea, South Korea; South- best adjust only for changing population growth rates east Asia includes Brunei Darussalam, Cambodia, Indonesia, Laos, Malaysia, Myanmar, Philippines, Singapore, Thailand, Timor-Leste, at the country level. Similarly, urban maps from Vietnam. remote sensing data have been limited to either static 2 Environ. Res. Lett. 10 (2015) 034002 A Schneider et al global maps that sacrifice detail to provide areal potential areas of urban and peri-urban growth coverage (Potere et al 2009), or local maps of (Webster 2002). metropolitan growth that forego coverage in favor of Urban expansion 2000–2010 was mapped in two spatial detail (Seto et al 2010). Moreover, of the many steps, beginning with delineation of the c. 2010 urban global urban maps now available (Elvidge et al 2007, extent. A probability surface of urban land was devel- Schneider et al 2009), none characterize changes in oped from three years of 500 m MODIS imagery (table urban land. A2) and training samples for urban and non-urban Recent comparative work on urbanization has areas photo-interpreted using very high resolution found a middle ground by focusing on local maps (VHR) Google Earth imagery (1–4 m resolution). A (typically 30 m resolution) for a sample of cities (Seto separate probability surface based on vegetation char- et al 2011, Angel et al 2012, Taubenbock et al 2012, acteristics of urban and non-urban areas was pro- Schneider and Mertes 2014). These studies point to duced from 250 m MODIS enhanced vegetation index several key findings: (1) cities in developing countries (EVI) data (Tan et al 2011) and integrated with the are consistently smaller or more compact than those in 500 m probabilities according to Bayes’ Rule (Mertes developed nations (Huang et al 2007, Schneider and et al 2015). Woodcock 2008); (2) developing country cities typi- To detect change, we assume all urban expansion cally undergo some decrease in population density in 2000–2010 to be unidirectional and occur within the the core during their development trajectory (Mur- mapped 2010 urban extent. We again exploit EVI data akami et al 2005); and (3) cities are declining in density in a multi-date composite technique (annual max- (Angel et al 2011, 2012). imum for each year, 2001–2010) by stacking all images Although they represent advances in our empirical for classification with a boosted decision tree (Quinlan understanding of urbanization, these studies (and 1993) to map (a) stable urban areas; and (b) areas that their conclusions) suffer from several limitations. were developed 2000–2010. This approach relies on Many rely on a limited very city sample (Murakami the assumption that any conversion from a non-urban et al 2005, Taubenbock et al 2012); most exclude land cover to developed land is detectable through small-and medium-sized cities (100 000-1 million) changes in vegetation content (Schneider et al 2010, where the majority of new growth is taking place Mertes et al 2015). (UN 2012). Definitional issues also jeopardize com- The final maps were assessed for accuracy using a parability: nearly all utilize municipal boundaries to two-tiered approach. The 2010 urban map was first clip the city extent, making it nearly impossible to assessed using a stratified random sample of 6528 sites compare trends across places. In addition, adminis- 0.132 km2 in size, and the maps of urban expansion trative divisions often under-bound the built-up were assessed using a separate random sample of 2086 extent, so new growth in fringe areas is not captured. sites (0.06 km2, to align with the 250 m resolution). Finally, many studies rely only on population or Test sites were assessed within Google Earth against remote sensing, failing to connect the two to provide a VHR data in a double‐blind assessment by a team of complete picture of urban trends meaningful for photo-interpretation analysts, and labeled as urban/ environmental assessment, land use planning, and non-urban land (tier one), or urban land/urban regional policy implementation. expansion 2000–2010 (tier two) according to the >50% built-up threshold (note that the 50% threshold is used throughout to maintain consistency with pre- 3. Methods vious urban remote sensing efforts). Overall accuracy measures for the maps were calculated by comparing 3.1. Satellite-based maps of urban expansion the maps against the test sites. The results indicate that To establish potential locations of urban land, the map accuracies for urban extent (tier one) range study extent was first established by synthesizing all between a maximum of 93% to a minimum of 79% for contemporary city point data (table A1) with a c. 2000 each country, and for urban expansion (tier two), map of urban extent developed from MODIS 500 m between 91% and 70%, confirming their suitability for data (Schneider et al 2009, 2010). The MODIS 500 m this analysis (Mertes et al 2015). urban extent map has been shown to have the highest locational accuracy of available maps and a zero 3.2. Population density maps omission rate for cities globally (Potere et al 2009). Human population census data and corresponding Where city points did not align with the MODIS map administrative boundaries at the finest level available or vice versa, the locations were manually checked were obtained from multiple recent censuses in each against Google Earth data and adjusted. The final study nation (table A3). If they did not align with the c. 2000 extent was created by categorizing the identified urban and 2010 time points, the population data were patches into small, medium, and large classes accord- adjusted forward or backward using inter-censal ing to their spatial extent and population, and buffer- growth rates and linear estimation methods. High ing by 5, 25, and 100 km respectively to include resolution census data were then used to establish 3 Environ. Res. Lett. 10 (2015) 034002 A Schneider et al population densities for each time point (2000, 2010) convention in urban geography, and standardized the on a biome-by-biome basis for each land cover type in study extent for each small city using adaptive radial the region, following previous WorldPop (www. zones (5, 10, 15 km) based on 2010 population size worldpop.org.uk) mapping approaches (Tatem (Dietzel et al 2005, Seto and Fragkias 2005). et al 2007, Gaughan et al 2013). These population densities were then used as weights to distribute the 4. Results population across the raster cells, an approach that has been shown to produce more accurate disaggregations 4.1. Regional and country-level results than previous approaches that rely on disaggregation Across the region, the total net increase in urban land to very coarse data (nighttime lights data) or areal area was >34 000 km2 from 2000 to 2010, expanding weighting alone (Linard et al 2010, 2013, Gaughan from 155 000 to 189 000 km2. While urban land area et al 2013). After synthesizing all population data with increased >22%, urban populations climbed >31%, land cover information and built-up extent to map adding 231 million persons in just ten years (from 738 population density, we count only the population cells to 969 million). The rapid pace of population change is fully contained within the built-up area. With this clear in the average rates of change for each country approach, we avoid the problems common to urban (table A4): cities in the region grew annually at 2.8%, population data, including the lack of data at disag- with Malaysia, Vietnam, Cambodia, and Laos all gregated scales, country-dependent definitions and observing rates well above this average (4.0–7.8%). In delineations of urban versus rural that lead to drasti- contrast, the rates of change for urban land average cally different population estimates, and changes in 2.0% annually, with only China, the Philippines, census geographies that require adjustment so mea- Cambodia, and Laos having rates above this sures reflect true population growth/decline rather level (2.2–3.2%). than differences due to changes in administrative The results suggest that urban population growth boundaries (Cohen 2004). has outpaced land expansion, a trend we measure expli- citly using urban density. Here we estimate persons per 3.3. Analysis square kilometer of built-up land since the conven- We defined an urban agglomeration as the extended tional measure, persons per square kilometer within an area comprising the built-up area of a central place administrative region, does not account for the vastly (i.e. a city) and any suburbs or small cities linked by different sizes of municipal boundaries. While the continuous urban land (UN 2012). To delineate results show a great degree of variability (figure 2), there agglomerations for this analysis, we collected the most are two common trends across nations: (a) urban den- detailed administrative boundary data available (typi- sities are high (mean 2010, 5850 persons/km2), and (b) cally county or finer) for c. 2010, to reflect the most urban densities increase 15–30% from 2000 to 2010, recent units used for governing. We then assigned any adding between 270 and 2020 persons/km2 in ten years. administrative unit containing part of the contiguous Although on average, urban densities are decreasing in built-up area of the city to its agglomeration, so that China (from 6150 to 5290 persons/km2 across 677 the agglomeration boundary is made up of one or cities), there is considerable variability here as well: more official administrative units. This was repeated roughly half of Chinese agglomerations are decreasing for each city >100 000 persons, resulting in 1036 in density, while the remaining half witnessed no agglomerations across 17 countries (figure A1). For change or an increase in urban density, similar to other each agglomeration, we estimated the built-up extent agglomerations in the region. for 2000 and 2010 from the satellite-based maps, as well as the 2000 and 2010 urban population within the 4.2. City-level results: the view from above built-up extent from the population density maps. More than one-third of all urban land and urban The 1036 agglomerations were then stratified into five population in East–Southeast Asia falls into 30 large categories (UN 2012) based on their 2010 agglomera- agglomerations (figure 3). By 2010, the Pearl River tion population: >10 million; 5–10 million; 1–5 Delta agglomeration climbed to >41 million inhabi- million; 500 000–1 million; and 100 000–500 000. tants and 6970 km2 of urban land, surpassing Tokyo To understand regional urbanization trends (31 million persons, 5570 km2 urban land) as the within the 30 largest agglomerations, we also con- largest urban agglomeration on Earth. An additional ducted a separate analysis measuring urban expansion 12 of the top agglomerations are located in China, for all established cities within the administrative core, including Shanghai and Beijing, with 3480 and within the urban agglomeration defined by the built- 2720 km2 of urban land, and populations of 24 and 16 up extent, and directly adjacent or near the urban million persons, respectively, in 2010. China also agglomeration boundary (within 120 km of each contains the agglomerations with the greatest urban agglomeration’s center, following distance recom- land expansion, 2000–2010, with a median increase of mendations from the peri-urbanization literature, 463 km2, compared to a median of 217 km2 for all 30 Webster et al 2002). For this analysis, we followed cities. The Chinese agglomerations have witnessed 4 Environ. Res. Lett. 10 (2015) 034002 A Schneider et al Figure 2. Country-and city-level urban densities in East–Southeast Asia. On average, the majority of urban agglomerations in the region are becoming more dense, as shown in the box plots for urban densities (population/km2 of built-up land, 25th–75th percentiles) for 14 countries, 2000 and 2010 (not shown: Brunei, Mongolia, Timor-Leste). For comparison, the urban densities for 18 large agglomerations are included. significant population increases as well, adding a of a large number of independently-governed cities. median 2.5 million persons to each large agglomera- For instance, the Manila agglomeration has 17 cities in tion during the last decade. Several large agglomera- its administrative core (where resources and planning tions outside of China have major population are concentrated) and another 15 cities on the out- increases (Tokyo, Jakarta, Manila), but not surpris- skirts. Alternatively, many large agglomerations have a ingly, none have the scale of new development small core area governed as one unit, with expansion witnessed in China. that spills into the jurisdiction of nearby county-or The growth of these ‘mega-agglomerations’ is not city-level governments (e.g. Shanghai, Seoul, Hanoi). the whole story, however. The region has an additional To understand how cities within an agglomeration 101 large agglomerations, each with populations view and govern themselves, we measure urban between 1 and 5 million persons, totaling >207 mil- expansion for all established cities within 120 km of lion. Although rates of expansion in these areas are on the city core for the top 30 agglomerations. Here we par with the 30 large agglomerations (>3%), the aver- delineate each core according to its 2010 municipal age rates of population increase surpass those of the area, and standardize the size of each small city extent top 30, at >3.4%. These trends are also apparent in using adaptive radial zones corresponding to each agglomerations 100 000–1 million: small cities in city’s 2010 population. Myanmar, Indonesia, Vietnam and the Philippines, On average, >60% of 2010 urban land and >71% especially, have added population without much of new development 2000–2010 are located outside expansion (figure 4). Nearly all trajectories are headed the core administrative area, but within the urban in the same general direction, with an average increase agglomeration defined by this study (figure 5). The of 970 persons/km2 for the 2000–2010 period. results also highlight three distinct urban typologies for large agglomerations: (1) a core surrounded by 4.3. City-level results: governance and policy rapidly growing cities, with expansion rates that perspectives decline with distance (e.g. Hangzhou, Guangzhou, While the agglomeration provides a consistent way to Chengdu, Jakarta); (2) a core with numerous nearby compare metropolitan areas since they are defined by cities, but with limited growth due to geophysical fac- built-up extent, many agglomerations are comprised tors (e.g. Manila, Bangkok, Kuala Lumpur); and (3) a 5 Environ. Res. Lett. 10 (2015) 034002 A Schneider et al Figure 3. Amounts of (a) urban land and (b) urban population, 2000–2010, for 30 large urban agglomerations in East–Southeast Asia. The figures illustrate the 26 agglomerations with the largest 2010 populations (all >5 million), and four capital cities included for regional representation (Yangon, Phnom Penh, Pyongyang, Vien-tiane). Agglomerations are labeled by largest city (see table A5). core with few nearby cities (Hanoi, Bangkok). Some of The trend toward increasing urban densities is these latter areas are witnessing peri-urbanization clear in nearly all countries, and at multiple scales. At (expansion up to 100 km from the core, Kontgis the country level, Japan and South Korea lead the et al 2014), but this trend may not be fully captured in region with highly urbanized populations (80–90%) satellite-based estimates or census data due to its spread across multiple large urban agglomerations small, patchy nature. covering 3–5% of each country’s land area. Although growth has tapered off in these countries, their aggre- gate urban densities are still climbing. On average, 5. Discussion and conclusions population growth rates for large, middle-income countries (China, Indonesia, the Philippines, Thai- This research presents new evidence that East–South- land) are high (3.5%) relative to their average rates of east Asia is undergoing unprecedented urbanization urban expansion (2.6%). Cities of all sizes are growing and urban expansion, coincident with well-established in these countries, with higher rates of population trends of rapid industrialization, economic growth, growth for small cities than for large agglomerations and globalization. These results were generated using during the last decade. China is clearly a unique case, directly comparable, spatially-detailed datasets however. At the country level, Chinese cities appear to derived from multiple sources of remote sensing and be decreasing in density, a result that is expected given disaggregated census data, with close attention to how the central government’s planning and policy initia- urban land, urban expansion, urban population and tives focused on small cities outside major metropoli- agglomeration boundaries were defined and operatio- tan areas (Lin 1999). Results at the city level reveal that nalized. When the factors limiting comparative urban half of all Chinese cities have urban densities that analysis are addressed, the results reveal that urban increase or remain unchanged. agglomerations across East Asia are experiencing East–Southeast Asia is also home to several low or increasing urban densities. While these trends are not low-middle income countries with 30% of their total surprising for some scholars and local experts, they do populations living in urban areas, including Vietnam, contradict established empirical work that shows— Myanmar, Laos, and Cambodia. During the last dec- with similar attention to consistency in data and ades, these countries have witnessed major shifts from definitions but with results modeled using static c. predominantly subsistence agrarian economies to 2000 urban maps—that cities are universally declining increasingly commercialized agriculture, leading to in density (Angel et al 2011, 2012). rapid urbanization of rural populations (Hall 6 Environ. Res. Lett. 10 (2015) 034002 A Schneider et al Figure 4. Mean urban land and urban population, 2000–2010, for agglomerations 100 000–5 million (not shown due to lack of cities in these size classes: Brunei, Cambodia, Laos, Mongolia, North Korea, Singapore, Thailand, Timor-Leste). et al 2011). The rates of urban population growth at cities where the majority of urban growth is con- the country level average 4.6% annually, primarily due centrated. From these results, we therefore conclude to the extraordinary growth of just a few large cities that cities as they are experienced on the ground (i.e. (>1 million). Ho Chi Minh City, Hanoi, Yangon, and contiguous built-up regions) are often not the same as Vientiane, for example, have all witnessed rapid popu- how they are governed. Given rising urban densities, lation growth, adding an average 1.4 million persons, continued expansion, and a lack of coordinated gov- 2000–2010. The results here reveal limited urban ernance, the question for governments and planners expansion, though, which has led to an average becomes whether adequate services, infrastructure, increase in urban density of 870 persons/km2. housing, and employment are available or can be pro- Finally, this work also examines how differences in vided to incoming populations. administrative boundaries and urban definitions There are several potential sources of uncertainty impact how we characterize, monitor, and understand in this study that should be noted. With respect to the urban change. We defined 30 large agglomerations by remote sensing data, the 250 m pixel size combined contiguous urban land, but evaluated rates/amounts with the population threshold of 100 000 makes it dif- of growth within the core administrative area and for ficult to capture all small settlements. In China, Indo- the individual cities comprising these agglomerations nesia, and Vietnam, villages are spectrally distinct and (figure 5). Most administrative cores contain multiple sufficiently large (>1 km2), and disaggregated popula- cities on average, while an additional 2–21 cities exist tion data are available. Accordingly, they are well- within the built-up area of the agglomeration, but out- mapped with our methods (figure 1). In Laos, Cambo- side the jurisdiction of the core. It is in these outer dia, and North Korea, villages are small and comprised 7 Environ. Res. Lett. 10 (2015) 034002 A Schneider et al Figure 5. Scatter plots illustrating the size and growth rates of small and mid-sized cities located within the administrative core (gray area), within the urban agglomeration defined by the built-up extent (left of the dashed line), or directly adjacent or near the urban agglomeration boundary (right of the dashed line). A sample of results for 12 of the 30 largest agglomerations is shown here. Urban agglomerations are labeled by largest city (see table A5 for a list of cities within each agglomeration). of local materials that are spectrally similar to sur- monitoring remains a critical limitation of both popu- rounding land cover types. These countries have no lation data sources and remote sensing for land use reliable village population estimates, and conse- planning. New datasets (crowd sourcing, social media, quently, the results may under-report urban land or etc) and advances in radar/lidar have the potential to growth. On average, the total land area and population significantly change how we monitor urban change of these settlements is a fraction of the urban extent (Frolking et al 2013, Tsou et al 2013). and urban population in each country, and should Urban growth has increased in scope, scale, and therefore have a limited effect on interpreting the complexity in recent decades, and has become one of results of this study. Finally, the urban extent does not the most important challenges of the 21st century. The include low-density settlements (e.g. 30–40% built- urban expansion and urban growth datasets11 pre- up), although these areas may function as urban space. sented here provide a valuable, practical, and con- If we relax the 50% threshold, higher rates and sistent way to monitor a broad range of issues, amounts of urban land would be likely. including impacts to local-regional climate (Kauf- One additional area of uncertainty is related to the mann et al 2007), pollution levels (Grimm et al 2008), availability of population data. Locations with less- water quality/availability (McDonald et al 2011), ara- than-ideal data include Malaysia, Thailand, Laos, ble land (Lambin and Meyfroidt 2011) as well as the Myanmar, and North Korea (table A3); results for livelihoods and vulnerability of populations in the these countries should be considered in light of this region (Solecki et al 2011). These datasets are unique bias. In addition, population estimates have greater in that they represent the first comprehensive map- uncertainty when the administrative unit is large rela- ping of urban expansion and growth for all cities tive to urban extent, and rural populations within the >100 000 in East–Southeast Asia, and they also form unit are dense (Hay et al 2005). In these areas (e.g. the basis of ongoing work to examine land and popu- Indonesia), population densities may be over- lation trends globally for all cities and agglomerations. estimated. Finally, the approach here does not capture growth within existent urban areas, including redeve- 11 All datasets are publically available at www. lopment or vertical growth. The lack of within-city landcoverchange.com. 8 Environ. Res. Lett. 10 (2015) 034002 A Schneider et al While uncertainties may always be present no matter Science and Technology Directorate, Department of the data source, spatially-and temporally-detailed Homeland Security, and the Fogarty International maps of urban expansion and population growth Center, National Institutes of Health, and is also based on the best available data are nevertheless cri- supported by grants from the Bill and Melinda Gates tical for researchers, urban planners, land managers, Foundation (#49446, #1032350). The funders had no and government officials interested in a sustainable role in study design, data collection and analysis, urban future. decision to publish, or preparation of the manuscript. The authors also wish to thank Caitlin Kontgis, Mutlu Ozdogan, and four anonymous reviewers for their Acknowledgments helpful comments on an earlier draft of this manuscript. AS acknowledges funding support from the World Bank for preparation of datasets. AJT acknowledges funding support from the RAPIDD program of the Appendix Figure A1. The distribution of urban agglomerations assessed in this research, including (a) 131 agglomerations >1 mil, (b) 164 cities between 500,000 and 1 mil, and (c) 741 cities between 100,000 and 500,000 persons. The number of agglomerations in each population size category is shown by country in (d). Table A1. City point and raster datasets used to define the study extent for satellite image processing of urban expansion, as well as to define the 1036 urban agglomerations used for analysis. Location Dataset Producer Citation Notes Global GRUMP city points CIESIN, IFPRI, CIAT Center for International Earth Science Point dataset of 67,935 Information Network (CIESIN), cities, towns and Columbia University, International settlements. Food Policy Research Institute (IFPRI), World Bank, Centro Internacional de Agricultura Tropical (CIAT) 2004 Glo- bal Rural-Urban Mapping Project (GRUMP): Settlement points (2000) http://sedac.ciesin.columbia.edu 9 Environ. Res. Lett. 10 (2015) 034002 A Schneider et al Table A1. (Continued.) Location Dataset Producer Citation Notes Global Urban agglomerations UN Department of United Nations (UN) Department of Eco- Point dataset of 633 with >750,000 inha- Economic and nomic and Social Affairs Population cities >750,000 bitants, 2011 Social Affairs Popu- Division 2013 Urban agglomerations persons. lation Division with >750,000 inhabitants in 2011 http://esa.un.org/unup/GIS-Files/ gis_1.htm Global Universe of cities Angel, Lincoln Insti- Angel S 2012 Planet of Cities (Cambridge, Point dataset of 3,943 tute of Land Policy MA: Lincoln Institute of Land Policy cities >100,000 Publications persons. China Chinese city point data Chinese Academy of Chinese Academy of Sciences 2011 City Point dataset of 664 Sciences points Beijing, China cities. Global Google Earth popu- Google Google Earth Pro v7.1. 2013 Layers: popu- City point location used lated places lated places http://www.google.com/ to verify, geolocate, earth and update city points. Global MODIS 500 m map of University of Wiscon- Schneider A, Friedl M, Potere D 2010 Map of 88,578 urban global urban extent sin-Madison Mapping urban areas globally using patches >1 km2 used MODIS 500m data: New methods and to verify, geolocate, datasets based on urban ecoregions and update city Remote Sens Environ 114 1733-1746 points. http://sage.wisc.edu Table A2. Remote sensing data sources used to map urban extent and urban expansion, 2000-2010. Spatial Location Dataset Producer Citation resolution East Asia MODIS 500 m nadir BRDF-adjusted NASA MODIS Schaaf C B et al 2002 First operational BRDF 500 m reflectance, 7 spectral bands, 8-day Land Team, albedo nadir reflectance products from composites (MCD43A2, Boston MODIS Remote Sens. Environ. 83 135-148 MCD43A4) University East Asia MODIS 250 m enhanced vegetation NASA Goddard Gao F, Morisette J, Wolfe R, Ederer G, Pedelty J, 250 m index 8-day annual and tiled pro- Space Flight Masuoka E, Myeneni R, Tan B, Nightingale J ducts (MOD09Q1G_EVI) Center 2008 An algorithm to produce temporally and spatially continuous MODIS-LAI time series IEEE Geoscience Remote S 5 60-64 Tan B, Morisette J, Wolfe R, Gao F, Ederer G, Nightingale J, Pedelty J 2011 An enhanced TIMESAT algorithm for estimating vegeta- tion phenology metrics from MODIS data IEEE J Sel Top App 4 361-371 Global Training exemplar database Boston Uni- Friedl, M., et al. 2009 MODIS Collection 5 global 1-30 m versity, Uni- land cover: algorithm refinements and char- versity of acterization of new datasets Remote Sens. Wisconsin- Environ. 114 168-182 Madison Schneider A, Friedl M, Potere D 2010 Mapping urban areas globally using MODIS 500m data: New methods and datasets based on urban ecoregions Remote Sens. Environ. 114 1733- 1746 Mertes C M, Schneider A, Sulla-Menashe D, Tatem A, Tan B 2014 Detecting change in urban areas at continental scales with MODIS data Remote Sens. Environ. in review East Asia Test sites for accuracy assessment University of Mertes C M, Schneider A, Sulla-Menashe D, 250- Wisconsin- Tatem A, Tan B 2014 Detecting change in 500 m Madison urban areas at continental scales with MODIS data Remote Sens. Environ. in review Acronyms: MODIS, Moderate Resolution Imaging Spectroradiometer, BRDF, bidirectional reflectance distribution function, NASA, National Aeronautics and Space Administration. 10 Environ. Res. Lett. 10 (2015) 034002 A Schneider et al Table A3. Population data sources used to map population density, 2000-2010, for each country. Country or Years of data region Official name Statistical agency Link availablea Level of data Cambodia Kingdom of Cambodia National Institute of Statistics, http://www.nis.gov.kh 1998, 2008 province Cambodia China People’s Republic of China National Bureau of Statistics, China http://www.stats.gov.cn 2000, county, muni- 2005, 2010 cipality China Data Center, University of http://chinadatacenter.org 2000, 2010 county Michigan Hong Kong Hong Kong Special Admin- Census and Statistics Department, http://www.censtatd. 2001, 2011 district istrative Region, China Hong Kong SAR, China gov.hk North Democratic People's Central Bureau of Statistics, DPR http://www.geohive.com/ 2005, 2008 province Korea Republic of Korea Korea cntry/northkorea.aspx Indonesia Republic of Indonesia Biro Pusat Statistik, Indonesia http://www.bps.go.id 2000, province 2005, 2010 Japan Japan Statistics Bureau, Management and http://www.stat.go.jp 2000, district Coordination Agency, Japan 2005, 2010 Laos Lao People's Democratic Lao Department of Statistics http://www.nsc.gov.la 1995, 2005, province Republic 2009, 2011 Malaysia Malaysia Department of Statistics, Malaysia http://www.statistics. 2000, district gov.my 2005, 2010 Mongolia Mongolia National Statistical Office, http://www.nso.mn 2000, 2010 aimag, soum Mongolia Myanmar Republic of the Union of Department of Population, https://www.mnped. 1983, district Myanmar Myanmar gov.mm 2002, 2004 Philippines Republic of the Philippines National Statistics Office, http://www.census.gov.ph 2000, province, Philippines 2007, 2010 municipality South Republic of Korea National Statistical Office, Republic http://kostat.go.kr 2000, city, district Korea of Korea 2005, 2010 Singapore Republic of Singapore Statistics Singapore http://www.singstat.gov.sg 2000, 2010 region, district Taiwan Republic of China, Taiwan Department of Household Registra- http://www.stat.gov.tw 2000, county tion Affairs, Taiwan 2006, 2010 Thailand Kingdom of Thailand National Statistical Office, Thailand http://www.nso.go.th 2000, 2010 changwat Vietnam Socialist Republic of General Statistical Office, Vietnam http://www.gso.gov.vn 1999, province Vietnam 2009, 2011 a Maps of population density were produced for c 2000 and c 2010 using all available census data. Where data were not available, population data were adjusted forward or backward using inter-censal UN population growth rates (Tatem et al 2007, Linard et al 2013). Table A4. Changes in urban land and urban population for agglomerations >100,000 in East-Southeast Asia.a,b,c,d Average Average Ratio of urban annual annual rate land increase Area within Urban Urban Urban Urban rate of of change, to urban administrative land land population population change, urban population boundary 2000 2010 2000 2010 urban population change (m2/ (km2) (km2) (km2) (persons) (persons) land (%) (%) persons) China 9,453,309.3 98,819.4 126,661.1 453,257,034 598,918,893 2.5 2.8 191:1 Japan 372,468.1 19,270.5 20,094.5 76,080,201 87,527,422 0.4 1.4 72:1 Indonesia 1,890,972.7 12,635.5 13,921.9 83,535,095 118,351,117 1.0 3.5 37:1 Thailand 514,093.0 4,616.1 5,365.6 15,451,438 19,947,409 1.5 2.6 167:1 Malaysia 329,424.2 4,644.3 5,364.4 11,566,137 17,074,669 1.5 4.0 131:1 Vietnam 328,385.3 4,200.9 5,098.2 22,854,276 33,863,070 2.0 4.0 82:1 South 100,229.2 2,835.9 3,232.4 24,958,293 28,271,528 1.3 1.3 120:1 Korea Philippines 295,987.7 2,332.9 2,907.9 19,397,798 26,882,521 2.2 3.3 77:1 Taiwan 36,223.7 1,782.9 2,043.3 13,801,713 14,801,705 1.4 0.7 260:1 Myanmar 670,746.8 1,838.4 2,030.1 8,452,657 11,235,349 1.0 2.9 69:1 North 122,755.1 852.6 906.6 4,189,762 4,693,317 0.6 1.1 107:1 Korea Mongolia 1,566,250.3 683.1 764.4 840,233 1,209,552 1.1 3.7 220:1 Singapore 755.4 337.3 403.5 2,539,073 3,412,239 1.8 3.0 76:1 11 Environ. Res. Lett. 10 (2015) 034002 A Schneider et al Table A4. (Continued.) Average Average Ratio of urban annual annual rate land increase Area within Urban Urban Urban Urban rate of of change, to urban administrative land land population population change, urban population boundary 2000 2010 2000 2010 urban population change (m2/ (km2) (km2) (km2) (persons) (persons) land (%) (%) persons) Cambodia 181,354.0 218.3 290.9 1,195,233 1,806,264 2.9 4.2 119:1 Laos 229,878.0 162.0 222.6 296,091 629,370 3.2 7.8 182:1 Brunei 528.5 144.4 180.3 155,880 230,304 2.0 4.0 481:1 Timor- 369.4 28.4 28.4 115,901 180.737 0.0 4.5 0:1 Leste Total 16,092,832.9 155,230.1 189,307.1 738,415,036 968,624,426 2.0 2.8 148:1 a Urban extent maps produced at 250m resolution (Mertes et al 2014). In these maps, pixels containing at least 50% constructed surfaces are considered urban. b Population data were estimated from the WorldPop population distribution maps (Linard et al 2013) for built-up areas within the urban expansion map. c Administrative boundary data provided by GADM (2012). d Agglomerations were defined by the administrative units corresponding to the contiguous built-up land area of cities over 100,000. Table A5. List of independent cities comprising the 30 large agglomerations assessed in this research. Ranka Agglomerationb Country Cities, towns >100,000 included in agglomerationc 1 Pearl River China Conghua Guangzhou Luoyang Shunde Zhuhai Delta Daling Heshan Nanhai Sihui Dongguan Huiyang Pingshan Xinhui Foshan Huizhou Qingyuan Zengcheng Gaoming Jiangmen Sanshui Zhaoqing Gaoyao Kaiping Shenzhen Zhongshan 2 Tokyo Japan Abiko Hachioji Kashiwa Misato Tachikawa Ageo Hadano Kasukabe Mitaka Takasaki Akishima Higashikurume Kawagoe Musashino Tama Asaka Higashimurayama Kawaguchi Nagareyama Toda Ashikaga Hino Kawasaki Narashino Tokorozawa Atsugi Hiratsuka Kiryu Niiza Tokyo Chiba Hoya Kisarazu Noda Tsuchiura Chigasaki Ichihara Kodaira Odawara Urawa Chofu Ichikawa Koganei Ome Urayasu Ebina Iruma Kokubunji Omiya Utsunomiya Fuchu Isehara Koshigaya Ota Yachiyo Fujimi Isesaki Kumagaya Sagamihara Yamato Fujisawa Iwatsuki Machida Sakura Yokohama Fukaya Kamagaya Maebashi Sayama Yokosuka Funabashi Kamakura Matsudo Soka Zama 3 Shanghai China Kunshan Shanghai Suzhou Taicang Wujiang 4 Beijing China Beijing Sanhe 5 Bangkok Thailand Bangkok Nakhon Pathom Pak Kret Samut Prakan Khlong Nonthaburi Phra Thanya Buri Luang Pradaeng 6 Osaka Japan Akashi Ibaraki Kawanishi Neyagawa Takatsuki Amagasaki Ikeda Kishiwada Nishinomiya Tondabayashi Daito Itami Kobe Osaka Toyonaka Habikino Izumi Kyoto Sakai Uji Higashiosaka Kadoma Matsubara Sanda Yao Himeji Kakogawa Mino Suita Hirakata Kawachinagano Moriguchi Takarazuka 12 Environ. Res. Lett. 10 (2015) 034002 A Schneider et al Table A5. (Continued.) Ranka Agglomerationb Country Cities, towns >100,000 included in agglomerationc 7 Nagoya Japan Anjo Ise Kuwana Okazaki Toyota Gifu Kakamigahara Matsusaka Seto Tsu Handa Kariya Nagoya Suzuka Yokkaichi Ichinomiya Kasugai Nishio Tajimi Inazawa Komaki Ogaki Tokai 8 Kuala Lumpur Malaysia Ampang Kuala Lumpur Selayang Baru Shah Alam Ulu Kelang Klang Petaling Jaya Seremban Subang Jaya 9 Tianjin China Tianjin 10 Jakarta Indonesia Bekasi Cimanggis Depok Sawangan Bogor Ciomas Jakarta Serang Ciawi Ciputat Pondok Aren Serpong Cibinong Citeureup Pondokgede Tangerang 11 Hangzhou China Hangzhou Keqiao Shaoxing Xiaoshan Yuhang 12 Manila Philippines Antipolo Calamba Makati Muntinglupa San Jose del Monte Bacoor Caloocan Malabon Navotas San Juan del Monte Baliuag Cavite Malolos Paranaque Santa Rosa Binan Dasmarinas Mandaluyong Pasay Tagig Binangonan Imus Marikina Pasig Taytay Cainta Las Pinas Meycauayan Quezon City Valenzuela 13 Shantou China Anbu Chaoyang Denghai Jieyang Puning Caitang Chaozhou Fengxi Paotai Shantou 14 Seoul South Korea Ansan Koyang Osan Shihung Uiwang Anyang Kunpo Puch'on Songnam Hanam Kuri P'yongt'aek Suwon Inch'on Kwangmyong Seoul Uijongbu 15 Chengdu Chengdu Chengdu Chongzhou Guanghan 16 Shenyang China Fushun Shenyang 17 Wuhan China Wuhan 18 Hanoi Vietnam Ha Dong Hanoi 19 Singapore Singapore- Singapore Johor Bahru Malaysia 20 Ho Chi Vietnam Bien Hoa Ho Chi Minh City Thu Minh City Daut Mot 21 XI'an China Xi'an Xianyang 22 Surabaya Indonesia Gresik Sidoarjo Taman Pasuruan Surabaya Waru 23 Chongqing China Chongqing 24 Taipei Taiwan, PRC Chungho Hsintien Pingchen Tanshui Yingko Chungli Luchou Sanchung Taoyuan Yungho Hsichih Panchiao Shulin Tucheng Hsinchuang Pate Taipei Yangmei 25 Bandung Indonesia Bandung Cimahi Ciparay Margahayu Padalarang 26 Yangon Myanmar Yangon City 27 Hong Kong China Jiulong Sheung Shui Tseun Wan Kwai Chung Tai Po Tuen Mun Ma On Shan Tin Shui Wai Xianggang Sha Tin Tseung Kwan O Xianggangzi 28 Phnom Penh Cambodia Phnom Penh 29 Pyongyang North Korea Pyongyang 30 Vientiane Laos Vientiane a Rank was determined according to the 2010 agglomeration population estimated from the WorldPop population density maps produced in this work. b Agglomerations were defined by the administrative units (GADM 2012) corresponding to the contiguous built-up land area of cities over 100,000. c Cities within each agglomeration were compiled from all available city lists (table A1) and verified through local maps and urban planning documents. 13 Environ. Res. Lett. 10 (2015) 034002 A Schneider et al References transformations estimated with satellite data Geophys. Res. Lett. 32 L02405 Angel S 2012 Planet of Cities (Cambridge, MA: Lincoln Institute of McDonald R I, Green P, Balk D, Fekete B, Ravenga C, Land Policy) Todd M and Montgomery M 2011 Urban growth, climate Angel S, Parent J and Civco D 2012 The fragmentation of urban change, and freshwater availability Proc. Natl Acad. Sci. landscapes: global evidence of a key attribute of the spatial USA 108 6312–7 structure of cities 1990–2000 Environ. Urban 24 249–83 Mertes C M, Schneider A, Sulla-Menashe D, Tatem A J and Tan B Angel S, Parent J, Civco D, Blei A and Potere D 2011 The dimensions 2015 Detecting change in urban areas at continental scales of urban expansion: estimates and projections for all coun- with MODIS data Remote Sens. Environ. 158 331–47 tries, 2000–2050 Prog. Plan. 75 53–107 Montgomery M 2008 The urban transformation of the developing Angel S, Sheppard S C, Civco D L, Buckley R, Chabaeva A, world Science 319 761–4 Gitlin L, Kraley A, Parent J and Perlin M 2005 The dynamics Murakami A, Zain A, Takeuchi K, Tsunekawa A and Yokota S 2005 of global urban expansion Transport and Urban Develop- Trends in urbanization and patterns of land use in the Asian ment Department Report (Washington DC: World Bank mega cities of Jakarta, Bangkok, and Metro Manila Landscape Publications) Urban Plan. 70 251–9 Bagan H and Yamagata Y 2012 Landsat analysis of urban growth: Potere D, Schneider A, Schlomo A and Civco D 2009 Mapping how Tokyo became the world’s largest megacity during the urban areas on a global scale: Which of the eight maps now last 40 years Remote Sens. Environ. 127 210–22 available is more accurate? Int. J. Remote Sens. 30 6531–58 Balk D L, Deichmann U, Yetman G, Pozz F, Hay S I and Nelson A Quinlan J R 1993 C4 5: Programs for Machine Learning (New York: 2007 Determining global population distribution: methods, Morgan Kaufmann) applications and data Global Mapping of Infectious Diseases: Schneider A, Friedl M A and Potere D 2009 A new map of Methods, Examples, and Emerging Applications ed S Hay et al global urban extent from MODIS data Environ. Res. Lett. 4 (London: Academic) pp 119–56 044003 Cohen B 2004 Urban growth in developing countries: a review of Schneider A, Friedl M A and Potere D 2010 Mapping urban areas current trends and a caution regarding existing forecasts globally using MODIS 500 m data Remote Sens. Environ. 114 World Dev. 32 23–51 1733–46 Dietzel C, Herold M, Hemphill J and Clarke K 2005 Spatio-temporal Schneider A and Mertes C 2014 Expansion and growth in Chinese dynamics in California’s central valley: empirical links to cities, 1978–2010 Environ. Res. Lett. 9 024008 urban theory Int. J. Geogr. Inf. Sci. 19 175–95 Schneider A and Woodcock C 2008 Compact, dispersed, fragmen- Elvidge C, Tuttle B, Sutton P, Baugh K, Howard A, Milesi C, ted, extensive? A comparison of urban growth in twenty-five Bhaduri B and Nemani R 2007 Global distribution and global cities Urban Stud. 45 659–92 density of constructed impervious surfaces Sensors 7 Seto K C and Fragkias M 2005 Quantifying spatiotemporal 1962–79 patterns of urban land-use change in four cities of Frolking S, Milliman T, Seto K and Friedl M 2013 A global footprint China with time series landscape metrics Landscape Eco. 20 of macro-scale changes in urban structure from 1999–2009 871–88 Environ. Res. Lett. 8 024004 Seto K C, Fragkias M, Guneralp B and Reilly M 2011 A meta-analysis Gaughan A E, Stevens F, Linard C, Jia P and Tatem A 2013 High of global urban land expansion PLoS One 6 e23777 resolution population distribution maps for Southeast Asia in Seto K C, Sanchez-Rodriguez R and Fragkias M 2010 The new 2010 and 2015 PLoS One 8 e55882 geography of contemporary urbanization and the environ- Grimm N B, Foster D, Groffman P, Grove J, Hopkinson C, ment Annu. Rev. Environ. Resour. 35 167–94 Nadelhoffer K, Pataki D and Peters D 2008 The changing Solecki W, Leichenko R and O’Brien K 2011 Climate change landscape: ecosystem responses to urbanization and pollu- adaptation strategies and disaster risk reduction in cities Curr. tion across climatic and societal gradients Front. Ecol. Opin. Environ. Sustain. 3 135–41 Environ. 6 264–72 Tan B, Morisette J, Wolfe R, Gao F, Ederer G, Nightingale J and Hall D, Hirsch P and Li T 2011 Powers of Exclusion: Land Dilemmas Pedelty J 2011 An enhanced TIMESAT algorithm for estimat- in Southeast Asia (Honolulu, Hawaii: University of Hawaii ing vegetation phenology metrics from MODIS data IEEE J. Press) p 320 Sel. Top. Appl. 4 361–71 Hay S I, Noor A, Nelson A and Tatem A J 2005 The accuracy of Tatem A J, Noor A M, vonHagen C, DiGregorio A and Hay S 2007 human population maps for public health application High resolution settlement and population maps for low Tropical Med. Int. Health 10 1–11 income nations: combining land cover and national census in Huang J, Lu X X and Sellers J 2007 A global comparative analysis of East Africa PLoS One 2 e1298 urban form: applying spatial metrics and remote sensing Taubenbock H, Esch T, Felbier A, Wiesner M, Roth A and Dech S Landscape Urban Plan. 82 184–97 2012 Monitoring urbanization in megacities from space Kaufmann R, Seto K, Schneider A and Zhou L 2007 Climate Remote Sens. Environ. 117 162–76 response to rapid urban growth: evidence of a human- Tsou M, Yang J, Lusher D, Han S, Spitsberg B, Gawron J, induced precipitation deficit J. Clim. 20 2290–306 Gupta D and An L 2013 Mapping social activities and Kontgis C, Schneider A, Fox J, Saksena S, Spence J and Castrence M concepts with social media and web search engines Cartogr. 2014 Monitoring peri-urbanization in the greater Ho Chi Geogr. Inf. Sci. 40 337–48 Minh City metropolitan area Appl. Geogr. 53 377–88 United Nations, Department of Economic and Social Affairs, Lambin E F and Meyfroidt P 2011 Global land use change, economic Population Division 2012 World Urbanization Prospects: The globalization, and the looming land scarcity Proc. Natl Acad. 2011 Revision No. ESA/P/WP.224 Sci. USA 108 3465–72 Wang L et al 2012 China’s urban expansion from 1990 to 2010 Lin G C S 1999 State policy and spatial restructuring in post-reform determined with satellite remote sensing Chin. Sci. Bull. 57 China, 1978–1995 Int. J. Urb. Reg. 23 670–96 2802–12 Linard C, Gilbert M and Tatem A 2010 Assessing the use of global Webster D 2002 On the edge: shaping the future of peri-urban land cover data for guiding large area population distribution East Asia Asia/Pacific Research Discussion Paper, the modelling GeoJournal 76 525–38 Urban Dynamics of East Asia Series (Stanford, CA: Stanford Linard C, Tatem A and Gilbert M 2013 Modelling spatial patterns of University) urban growth in Africa Appl. Geogr. 44 23–32 World Bank 2015 East Asia ’ s Changing Urban Landscape: Liu J, Tian H, Liu M, Zhuang D, Melillo J and Zhang Z 2005 China’s Measuring a Decade of Spatial Growth (Washington DC: changing landscape during the 1990s: large-scale land World Bank Publications) 14