Dramatic uneven urbanization of large cities throughout the world in recent decades

The study addresses how global large cities have developed in recent decades regarding built-up area (BUA) expansion, population growth, and urban greenness change, and how these urbanization features relate to economic levels. Motivated by UN projections of continued rapid urban population growth and the centrality of urban sustainability to the 2030 Sustainable Development Goals, the authors note that urbanization processes differ across socioeconomic contexts. Developed countries typically exhibit urban expansion aligned with population growth and robust infrastructure and services, while many developing countries face infrastructure deficits, overcrowding, and environmental challenges. Despite numerous regional and global studies of urbanization, a global, comparative perspective specifically characterizing large cities has been lacking. The authors therefore quantify and compare urbanization characteristics for 841 large cities worldwide to inform sustainable urban development.

Prior research has examined multiple aspects of urbanization from regional to global scales, including urban land expansion, environmental impacts (e.g., urban heat islands), and policy dimensions. Earlier global syntheses suggested that before 2000 urban land tended to expand faster than urban population growth. There is extensive literature on urban green space benefits for public health and environmental justice, as well as studies on infrastructure and sustainability challenges in rapidly urbanizing regions such as Africa, Latin America, India, and China. However, the authors identify a gap: an integrative, quantitative global analysis focused on the recent decades' urban expansion, population dynamics, and greenness change across large cities and economic strata.

Study scope and city definition: Using the MODIS MCD12Q1 (Collection 6.0) annual land cover product at 500 m resolution, urban and built-up (BUA) pixels were identified (≥30% impervious surface). Edge-adjacent BUA pixels (4-connectivity) were merged into contiguous urban patches. Large cities were defined as urban patches with BUA >100 km². City clusters (e.g., China’s Yangtze River Delta, YRD) were considered single large cities for analysis.

Temporal scope: Urban expansion was measured from 2001 to 2018. Population data were analyzed for 2000, 2005, 2010, and 2015.

Economic classification: Countries were grouped by World Bank 2018 GNI per capita thresholds into high income (H, >$12,375), upper-middle income (UM, $3,996–$12,375), lower-middle income (LM, $1,026–$3,995), and low income (L, <$1,025).

BUA expansion metrics: Built-up area expansion (BUAE) for each city was computed as BUA2018 − BUA2001. BUAE levels: E1 (≤1 km²), E2 (1–50 km²), E3 (>50 km²), based on distributional thresholds (excluding extreme outliers). City BUA size classes in 2018 were also categorized: 100–580 km², 580–2000 km², and >2000 km².

Population data and metrics: The 1 km GPWv4 gridded population (UN-adjusted) from SEDAC provided city population estimates for 2000–2015 within the urban extents. Urban population growth (UPG) per city was computed as Pop2015 − Pop2000, with growth rates UPGr = (Pop2015 − Pop2000)/Pop2000. The BUA growth rate BUr = (BUA2018 − BUA2001)/BUA2001. Their ratio Rpg (also denoted RPB) = UPGr/BUr was calculated for each city to compare relative rates of population versus area growth.

Urban greenness change: MODIS MOD13A1 v6 EVI at 500 m resolution was used as a greenness indicator. For each pixel within the 2018 BUA extent, the annual maximum EVI (EVImax) was derived from 16-day composites for 2001–2018. A Mann–Kendall non-parametric trend test assessed monotonic EVImax trends; pixels with significant positive trends (P<0.05) were counted as greening. The ratio R_greening per city was defined as the area of greening pixels divided by the 2018 BUA. Cities were grouped by R_greening: R1 (≤0.01), R2 (0.01–0.1], R3 (>0.1). Most greening pixels were within stable BUAs (urban throughout 2001–2018); pixels converted from vegetated rural land to urban were effectively excluded by requiring significant positive trends post-urbanization.

Population living in greening BUA (PGB): Using GPWv4 2015 data, the number of residents within BUA pixels exhibiting significant greening was aggregated for each city and analyzed by economic group and country.

Geospatial data: Country administrative boundaries were from GADM v3.4. All raster trend computations were implemented via Python and GDAL; city-level statistics were compiled across the 841 large cities.

Statistical comparisons: Group means across income categories are reported with SEM and significance (e.g., P<0.01) where applicable.

• Global expansion: Total global BUA rose from 7.47×10^5 km² in 2001 to 8.00×10^5 km² in 2018, equivalent to about 1,130 standard football fields (7,140 m²) per day. Among the 841 large cities, total BUA increased from 2.70×10^5 to 3.08×10^5 km², representing 36.1% and 38.5% of global BUA in 2001 and 2018, respectively.
• Country contributions to BUAE (2001–2018): China (47.5% of global increase), United States (9.0%), Nigeria (5.0%), India (3.6%), Indonesia (2.8%), Russia (1.8%), Mexico (1.7%), Malaysia (1.6%), Vietnam (1.5%), Ghana (1.3%).
• Distribution and economic groups: Of 841 large cities, H=353, UM=340, LM=127, L=21. Largest 2018 BUA shares among large cities: United States (27.0%), China (19.0%), Japan (6.0%), Brazil (4.0%), Germany (2.9%), India (2.8%), Canada (2.5%), Australia (2.0%), Mexico (1.9%), Russia (1.8%).
• BUAE per city by income (2001–2018): High-income mean 12.6 km² (lowest; significantly different from others, P<0.01); Upper-middle-income mean 38.0 km² (over 3× high-income; highest). There were 61 cities with BUAE >50 km²; 51 in China; Moscow was the only such city in Europe.
• Population growth (2000–2015): 80% of large cities (676/841) increased population; 22 cities grew by >2 million (mainly Asia and Africa). Average population growth per city: Low-income ≈5.0×10^5, Lower-middle ≈5.0×10^5, Upper-middle ≈3.0×10^5, High-income ≈1.0×10^5. In 105 cities with BUAE >50 km², 55% (58/105) had faster population growth rates than area expansion rates. Overall, Rpg >1 in 543 cities (64.5%), indicating population growth outpaced horizontal expansion in most places post-2000.
• Population density (2015): The 20 densest cities were 15 in Asia and 5 in Africa. Mean density: Low-income 6.2×10^3 km⁻² versus High-income 2.0×10^3 km⁻². Densest large cities included Dhaka, Kathmandu, Manila, Karachi, Istanbul, and Cairo, which combined rapid population growth (>2M) with limited BUA expansion (<200 km²). Four megaregions/cities had >5M growth (YRD, PRD, Beijing, Jakarta) but maintained relatively lower densities due to rapid expansion.
• Urban greenness change (2001–2018): 325 cities (R_greening >0.1) had significant greening on >10% of their BUA pixels, concentrated in East Asia, Europe, and North America. Top contributors to total greening BUA among large cities: China 32%, United States 19%, Japan 7.7%. Average R_greening by income: UM 0.12 per city (highest), H 0.10, L ≈0.03 (lowest). Many cities exhibited a “fried egg” pattern with significant downtown greening and less greening or browning in suburbs. The top greening cities included YRD, PRD, Tokyo, Miami, Beijing, Chicago, Seoul, Tianjin, São Paulo, and Osaka.
• Population living in greening BUA (2015): Of 1.16 billion residents across the 841 large cities, 192 million lived in greening BUA pixels. Average PGB per city: UM ≈4.0×10^5, H ≈1.5×10^5, LM ≈1.1×10^5, L ≈0.24×10^5. By country: China ~108 million (56.2% of total PGB) across 150 large cities; United States ~10.98 million (151 cities); India ~5.2 million (34 cities). Urumqi had the highest city-level R_greening (0.53).
• Absolute greenness levels (2018): Average city EVImax (EVI_city) was highest in high-income countries (0.37) and lowest in low-income countries (0.28). Cities in the highest greenness quartile (Q4) contained only 12% of the total large-city population and were predominantly (over 71%) in high-income countries; about 69% of the total population lived in lower-greenness quartiles (Q1–Q2).

The analysis reveals pronounced economic stratification in recent urbanization. First, population growth has generally outpaced horizontal expansion (Rpg>1 in 64.5% of cities), reversing pre-2000 patterns where land expanded faster than population. This mismatch is most acute in lower-income settings, where high population growth coincides with relatively constrained expansion, resulting in substantially higher population densities and potential challenges such as slums, overcrowding, and environmental stress. In contrast, upper-middle-income countries have expanded faster on average, better accommodating population growth.

Second, increases in urban greenness have not kept pace with rapid population growth in many lower-income cities. While high-income cities already had relatively high greenness by 2001 (and thus saw modest increases), upper-middle-income cities achieved the greatest gains in greenness since 2001 (highest R_greening), reflecting investments in parks and green infrastructure amid rapid development. Low-income cities exhibited both the lowest absolute greenness and the smallest share of greening BUAs, underscoring equity gaps in access to urban green space. Given cities’ role in emissions and the health and climate co-benefits of greening, these findings highlight the importance of integrating green infrastructure into growth planning, especially in developing regions.

Third, China stands out for rapid and large-scale urbanization: it contributed nearly half of global BUAE, posted the largest population increases, and accounted for 32% of total greening BUA among large cities, with over 100 million residents living in greening areas. The prevalent “fried egg” downtown greening pattern suggests substantial retrofitting of core urban areas with parks and vegetation, potentially improving health and environmental outcomes.

Overall, the study answers the research questions by (i) quantifying recent patterns of BUA expansion, population growth, and greenness changes across 841 large cities, and (ii) demonstrating strong relationships between these urbanization features and national income levels, with implications for sustainable urban planning and equity.

This study provides a global, quantitative assessment of recent urbanization across 841 large cities, documenting uneven patterns of expansion, population growth, and greenness by economic group. Key contributions include: (1) evidence that, since 2000, population growth has outpaced horizontal expansion in most large cities; (2) identification of substantial greening gains in upper-middle-income cities and persistent greenness deficits in low-income settings; and (3) characterization of China’s outsized role in global urban expansion and greening. The approach demonstrates how satellite-based land cover and vegetation indices integrated with gridded population and economic classifications can systematically monitor urbanization and inform policy.

Future research should refine city definitions to better capture smaller and denser cities, especially in Africa and India; incorporate more granular subnational economic data; and extend analyses to include vertical growth, infrastructure provision, and additional environmental and social outcomes. Continued global monitoring is essential to support equitable, sustainable urban development aligned with the SDGs.

• Economic classification data are coarse and country-level; substantial within-country heterogeneity (e.g., eastern vs. western China) limits attribution of urbanization characteristics solely to national income groups.
• The definition of “large cities” as BUA >100 km² excludes many smaller but populous cities, particularly in Africa and India, potentially biasing representation of urbanization dynamics.
• Greenness trends were assessed within the 2018 BUA extent and based on EVImax; while this reduces phenology effects, it may not capture all forms of green infrastructure change and does not directly account for vertical greening.
• Population data were available through 2015, introducing temporal mismatch with BUA and greenness (through 2018). Vertical urban growth (e.g., high-rise construction) was not directly measured, which could influence interpretations of density and expansion dynamics.
• Reliance on remote sensing products entails classification uncertainties (though reported urban accuracy is relatively high), and the Mann–Kendall approach detects monotonic trends but not nonlinear changes.