Introduction

The United Nations (UN) laid out 17 Sustainable Development Goals (SDGs) on 25 September 2015. Of these, SDG 11 is aimed at “making cities and human settlements inclusive, safe, resilient and sustainable”, and it requires coordinated development in land use and land cover, demographics, and other sectors1. Various social, demographic, and economic factors collectively determine the growth and development of urban spaces2. However, rapid urban land expansion due to population growth presents serious challenges to sustainable development. It is responsible for air pollution and resource depletion on both local and global scales3,4. These challenges pose significant obstacles to achieving Sustainable Development Goal 11, particularly in developing nations where explosive population growth can lead to inadequate urban infrastructure5. Improving urban areas is crucial to addressing these challenges and achieving sustainable development goals. Along with the growth of urban land, the formulation of a sustainable development plan is even more essential to achieve the trade-off between land expansion and population changes (TLEPC). Such a trade-off strategy could help to avoid wasted resources.

Assessing the rate of urban land expansion in relation to population growth is important, but it does not consider the trade-offs involved. Additionally, limited monitoring data hinder the exploration of the spatial interactions between global population and land use changes6. To address these issues, we have developed a multiple-source remote satellite image-driven framework that explores the trade-offs and their impacts from 2010 to 2100. Through our research, we aim to address the following questions: What is the significance and purpose of the trade-off development strategy, and how does it impact the ecological service system and food production under different scenarios? Moreover, we analyse the most sustainable path for global development by utilizing the trade-off development strategy and identify regions where land consumption improvements are the most expected. Our study aims to fill the gaps in knowledge on this subject and provide insights into how trade-off policies can achieve sustainable urban development while accommodating population growth.

To make accurate projections about urban land expansion and population growth, it is essential to consider likely future development scenarios. The shared socioeconomic pathways (SSPs) model involves five scenarios—sustainability7, middle-of-the-road8, regional rivalry9, inequality10, and fossil-fuel development11— compared with the United Nations’ single scenario for simulating future global urban land changes. These scenarios can be used to explore changes in population, economy, land use, energy, and carbon emissions in the 21st century. The assumptions of the above-mentioned SSPs can be referenced in the works7,8,9,10,11. Given that land use change is significant in addressing societal challenges related to global ecosystem and climate system damage12, the effects of urban expansion under SSPs on habitat and terrestrial biodiversity (including amphibians, mammals, and birds) were explored in a recent study13. Interestingly, the study found that urban expansion has far less of an impact on natural areas compared with farmland expansion because urban sprawl generally consumes previously uncultivated land14. This underscores the importance of taking into account multiple factors when analysing the future development scenarios and their potential environmental impacts. Urban expansion affects fragile natural habitats, including in drylands, which threatens the achievement of the UN SDG 15 “life on land”4. The effect of uneven distribution of land expansion on the greening of built-up areas (BUAs) at different economic levels was analysed in refs. 15,16. The relationship between future LEPC at 1 km resolution under SSPs was explored in2. However, the above-mentioned methods might be neglected at the global scale to find the best solution for achieving the balance.

Our study focuses on exploring the interactions between population and land at a global scale, specifically at the country level. Rather than solely considering the development strategy of land or population, we have built upon previous research that has uncovered a deep relationship between the two. To achieve this, we propose a data-based simulation framework that can analyse the TLEPC under different shared socioeconomic pathways (SSPs). The framework utilizes multi-objective functions that aim to minimize urban land consumption while maximizing per capita built-up areas (PBUAs). We also employ Monte Carlo simulation to reach Pareto optimality. In addition, we quantify the impact of the trade-off index on the total loss of non-built-up areas by analysing the scenario simulation results at both regional and national levels in the 21st century. Furthermore, we thoroughly discuss the policy implications for the coordination index, and we use multiple sources of data in our analysis to provide robust findings. Overall, our study aims to provide insights into the complex relationships between land and population on a global scale and can inform policy decisions regarding urban development that prioritizes sustainability and the preservation of natural areas.

Results

Characteristics of spatial distribution for global PBUAs: global and local scales

In order to better understand the variations in urban land expansion and population changes across different countries and scenarios, we analysed the PBUAs at a country level with a 1/8° resolution. Our findings reveal that the distribution of PBUAs remains unequal across countries, as shown in Fig. 1. High-income countries with an average of 317.99 ± 176.49, while upper-middle-income, lower-middle-income, and low-income countries have lower PBUAs, with averages of 231.61 ± 181.94, 116.88 ± 146.48, and 189.81 ± 162.84, respectively. We also found that 60% of middle-income countries with high PBUAs are located in Europe, while 71.42% of low-income countries with high PBUAs are located in Africa and Asia. In terms of urbanization rates, we observed that high-income countries have the highest rate at 81.53 ± 10.79, while low-income countries have the lowest rate at 40.69 ± 13.08. Our findings demonstrate that the global inequality in PBUAs is likely to decrease in the future only if countries adopt sustainable urban development plans based on the principles of human settlement.

Fig. 1: Map of PBUAs in 2020.
figure 1

The 153 countries in the global economy can be categorized into five income categories: high-income countries (26.14%), middle-high-income countries (32.03%), middle-low-income countries (26.14%), low-income countries (3.92%), and a category for the remaining countries (11.76%). These countries have various PBUAs, with details varying by type and surrounding cities or states. a Represents the PBUAs of these countries, while b–g provides additional information on different types of PBUAs and their respective surroundings.

Full size image

Projecting urban expansion and population for the period 2010–2100 under the SSPs

Population growth and urban expansion are affected by various socio-economic factors. However, coordination between population growth and urban expansion remains lacking, especially in high-income and lower-middle-income countries, where there are currently high and low levels of PBUAs, respectively. The extent of PBUAs may vary depending on policy choices, urban land use, and population control efforts. Figure 2 highlights two trends: as population increases with urban expansion, the ratios of population and land expansion under SSPs 15 are 3.8977 ± 1.4342, 2.3452 ± 0.4291, 4.0865 ± 0.7365, 3.3264 ± 1.3070, and 2.8133 ± 1.1307, respectively. The evolution of PBUAs can vary, with SSP1 reaching its lowest point at 2040~2050, while SSP2~SSP5 can rapidly increase after the 2040 year. Regulating and controlling the increased ratios of population and land expansion can help reduce urban land expansion after the mid-21st century.

Fig. 2: Projecting future urban expansion and populations through the 21st century.
figure 2

SSP scenarios are (1) sustainability, (2) middle of the road, (3) regional rivalry, (4) inequality, and (5) fossil-fuelled development. Graphs refer to a urban population, b urban land expansion, c coordination between population and land expansion, and d per capita GDP of BUAs.

Full size image

Based on these projections, the per capita GDP of BUAs could experience considerable growth by 2100. A comparison with 2010 indicates potential growth of nearly 520%, 258%, 202.67%, 182.6%, and 319.1%. There is no significant difference in urbanization rates between countries of varying income levels following SSP1, SSP4, and SSP5 when compared with the SSP3 development path (refer to Supplementary files Table 1). However, with SSP1, significant variations in urbanization rates are mainly observed in low-income and lower-middle-income countries, of which 73.3% are located in Africa and Asia. Insignificant changes in urbanization rates are primarily seen in high-income countries following SSP1, of which 63.6% are located in Western Europe and South America.

Analysis of synergistic driving factors for SSPs

Factor analysis revealed three distinct groups of synergistic driving factors (see Table 1), which highlight the trade-offs between population, land use, and economy (refer to Supplementary Fig. 3a). Supplementary files provide a detailed analysis of the correlation coefficient of the performance indicators from Table 1 (refer to Supplementary Fig. 2). Positive relationships among population change rate (PCR), land change rate (LCR), and GDP change rate (GCR) identify synergistic factors between land use and population. Supplementary Fig. 3b provides a visual illustration of the relationship between PCR and LCR. Positive relationships between GCR/LCR, per capita GDP change rate (ΔPGDP), and per capita GDP/land change rate (ΔPLGDP) signify synergistic factors between land use and GDP. The synergistic factor between population and GDP can be identified by positive relationships (population density change rate (ΔPPOP)) and negative relationships (PCR/LCR). Factors 1–3 display a percentage of the variance of 90.329%, 9.664%, and 0.00319%, respectively (Fig. 3).

Table. 1 Synergistic factor analysis of population, economy, and land (maximum rotation of variance) to obtain the estimated load of each indicator of three orthogonal axes, according to 153 countries in the world, 2010–2020.
Full size table
Fig. 3: Modelling of LEPC under SSP1 Scenario.
figure 3

a Distribution of population and land change rates 2020–2100. b Data-driven modelling of land use and population. c LCR/PCR in 2020. d PBUAs in 2020 and e urbanization rate 2010~2100 following SSP1.

Full size image

Optimization result of trade-off between urban expansion and population under the SSPs

We conducted an analysis of the trade-off development paths between urban expansion and population by examining SSP scenarios 1~5. Through the iterative determination of global urban land consumption by country, we were able to derive the optimal possible scope of BUA development within each country. Our findings indicate that countries with a rate of urban expansion higher than their rate of population growth have a high potential to reduce their land consumption in the future. The TLEPC is based on the concept of developing population and land expansion at appropriate speeds.

We found that five sustainable development paths can benefit from the trade-off development strategy. Compared with the other SSP scenarios, SSP1 might be the best development path with the least land consumption, compared with the total land consumption by 2100. Moreover, we conducted several simulation experiments to explore the potential reduction in urban land use, determining that following SSPs 1–5 can reduce the land consumption by 8.07% (−0.08681 million km2), 28.2% (−0.61 million km2), 13.76% (−0.187 million km2), 23.92% (−0.48 million km2), and 55.28% (−1.92 million km2), respectively (see Fig. 4).

Fig. 4: Optimization results of TLEPC following SSPs 1~5, using the proposed modelling framework.
figure 4

a SSP1, b SSP2, c SSP3, d SSP4, e SSP5, f optimization results of SSPs 1~5. These graphs show that the optimal plan, with trade-offs between land use and population following various scenarios, can reduce land consumption in BUAs.

Full size image

Discussion

This study presents long-term predictions for future TLEPC on global, regional, and national scales using five different SSP scenarios. These simulations aim to improve the usefulness of scientific approaches in formulating urban land use plans17. Additionally, the proposed modelling framework can be applied at smaller scales, including cities, counties, or industrial parks, for more localized and specific analyses. The growth rates of urban land use following SSPs 1–5 during the period 2010–2100 were found to be 61.7%, 208.6%, 104.2%, 187.11%, and 317.9%, respectively. The growth rates in Asia and Africa were notably higher, ranging from 1.5 to 2.9 times those of other regions under SSPs 1–5. In terms of PBUA, the findings of this study align with previous research conducted in other countries, such as China and Japan, with an accuracy of 90%18. The TLEPC development strategy is also aligned with existing urban development planning strategies. By 2050, the PBUAs of founding European Union states (Germany, France, Italy, Netherlands, Belgium, and Luxembourg) are expected to be respectively 1.89, 1.858, 1.84, 2.01, and 1.72 times the PBUAs of Brazil, Russia, India, China, and South Africa (referred collectively to as BRICS) under SSPs 1–5. Additionally, typical megacities were observed for their urban land change mechanisms using light-at-night images19, as illustrated in Fig. 5.

Fig. 5: Urban land changes in 12 megacities.
figure 5

The percentage of BUAs is indicated by the change in pixels, 2000–2018. Compared to the megacities in other BRICS countries (e.g., Moscow and Delhi), Megacities in China (e.g., Beijing and Shanghai) with rapid urbanization rate. Megacities in the EU (e.g., Rome and Berlin) show slower urbanization, similar to that of New York.

Full size image

The spatial patterns of urban land use varied significantly across scenarios and were influenced by drivers of urbanization. Despite geographical, social, and cultural differences among countries and cities worldwide, urbanization, population growth, and climate change pose common challenges and trends20. This study aims to investigate the characteristics of land expansion and population evolution in different cities and explore the relationship between population growth and urban land use, which significantly impacts ecosystem services. Our simulation model effectively captured local urban land patterns and their evolution over time. Implementing trade-off strategies can lead to significant reductions in land consumption for both the EU and BRICS. Based on our optimization strategy, BRICS’ average PBUA is expected to remain within 152.82, 196.16, 164.92, 187.99, and 212.88 m2 for SSPs 1~5, respectively, by 2050. Considering minimum land consumption and current policies, such as those in China, SSP1 represents the most sustainable development path.

Urban land expansion can negatively impact the ecological service system, by reducing forest or grassland areas. To mitigate this effect, the proposed methods of SSPs 1–5 can lead to an increase of 0.0735 ± 0.0397 TC, 0.4247 ± 0.2725 TC, 0.1297 ± 0.0781 TC, 0.284 ± 0.168 TC, and 0.7418 ± 0.5380 TC per person in carbon sequestration potential over each 10-year period, respectively, by considering the carbon sink of forests. The proposed trade-off strategy following SSP1 (see Fig. 6) suggests that the carbon sink can be reduced to improve human settlement by up to 74.4% in African countries and 57.89% in Asia, while 67.86% of countries have the potential to reduce carbon sink. By slightly reducing PBUAs, the carbon sequestration potential can be increased slightly in 87.5% of European countries and 32.14% of countries in the Americas. Supplementary results following SSPs 2~5 are illustrated in the supplementary files (see Supplementary Figs. 4–7).

Fig. 6: Improvement of carbon sequestration potential of forests by using the SSP1 trade-off strategy.
figure 6

a Reduced built-up area with trade-off strategy to transform into forest land under the SSP1 and b reduced built-up area with trade-off strategy to transform into grassland under the SSP1. Carbon sequestration potential can be increased by 39.87% of countries, while 60.13% of countries can reduce carbon sink area to improve human settlement.

Full size image

The proposed methods of SSPs 1–5 can increase the carbon sequestration potential per person by 0.0022 ± 0.0015 TC, 0.0211 ± 0.0133 TC, 0.0051 ± 0.0038 TC, 0.0146 ± 0.0086 TC, and 0.0413 ± 0.0302 TC, respectively, over each 10-year period in grasslands. The coefficient of forest and grassland carbon sink can be obtained from21. Our proposed method is well suited for planning urban land development. Additionally, urban land expansion can impact the food production system. Compared with baseline SSPs, the cereal yield for the trade-off development path of SSPs 1–5 in 2030 can be increased by 19.798 million tons (MT), 74.589 MT, 26.532 MT, 60.965 MT, and 121.414 MT, respectively. The future yield can be estimated by using the average cereal yield from the World Bank between 2019 and 2021.

These messages are important policy warnings regarding the achievement of the TLEPC, as failing to do so could threaten ecological service systems and food security22. It is critical that future urban land use policies consider population prediction trends in order to reduce the consumption of urban land by planning optimal PBUAs that align with the TLEPC. However, this study has some limitations, including the unavailability of adequate data, which prevented us from exploring socio-economic and ecological factors. Additionally, countries with insufficient resources tend to have lower rates of urbanization, and strategies to improve these rates require further investigation. Future research should prioritize the coordination of urban land expansion and population changes in smaller regions, such as special economic parks and rural areas. We aim to address these issues by utilizing mobile signal and point-of-interest data to explore the multi-source drivers of land expansion and laws of population migration, with the goal of developing intelligent decision-making support system software to explore the spatio-temporal optimal PBUAs of different countries and cities around the world.

Methods

Data

Panel data regarding future population, urbanization rate (percentage of urban population to total population) and gross domestic product (GDP) were collated from the SSP database (https://tntcat.iiasa.ac.at/SspDb), which aggregates information about 153 countries according to geographic location and income level for the period 2020–2100. Data about change in land use and population by country was collected from https://doi.org/10.7910/DVN/85PJ1D for 2000–2100, with the resolution of 1/8° grid cell, predicted by the Country-Level Urban Build-up Scenario (CLUE_S) model with different driving factors in terms of GDP, population, etc. The boundaries of BUAs were downloaded from https://www.naturalearthdata.com/.

Modelling framework

We developed a suite of models that reflect the spatiotemporal relationship between LEPC at national and regional scales. Our model captures the small scale of population change and land expansion of countries with a resolution of 1/8° grid cells at 10-year intervals, using multi-resource satellite images and expert knowledge (refer to Fig. 7). We conducted Monte Carlo experiments following different development strategies, resulting in five urban land-expansion scenarios: sustainability, middle of the road, regional rivalry, inequality, and fossil-fuelled development. Using the CLUE_S model, we predicted the evolutionary urbanization path for every country following SSPs 1–5 during the period of 2010–2100. We established a trade-off modelling framework that links LEPC based on the distribution properties of land use and population growth. Our objective aims to minimize land consumption in global BUAs and maximize the PBUAs, which can be achieved through multi-objective optimization. The optimal point of equilibrium can be found by using the Pareto frontier theory (refer to Fig. 4). We validated the accuracy and effectiveness of our model by using country policies.

Fig. 7: Trade-off modelling framework linking LEPC.
figure 7

The data-driven modelling utilizes the CLUE_S model for multiple scenario simulations in the 21st century. The Monte Carlo-driven simulation evolves over time, with a focus on achieving the trade-off between population and land use for each country by decade.

Full size image

Trade-off models for optimizing the urban expansion and population

We further calculate the LCR at the country scale as

$${\rm {{LCR}}}=\left(\frac{{{{\rm {BUA}}}}_{t1}-{{{\rm {BUA}}}}_{t2}}{n}-1\right)\times 100 \%$$
(1)

where t1 and t2 are the final period and the initial period, respectively. n is the time interval between these two time periods. PCR and GCR can be calculated by using the same strategy. ΔPGDP is the per capita GDP difference between t1 and t2, and this strategy also applies to ΔPPOP and ΔPLGDP. Factor analysis was used for exploring internal dependencies among many variables, and a few independent unobservable variables were used for representing the basic data property. Factor analysis was achieved by using SPSS software.

The objective function is designed to obtain the Pareto optimal solution set of the TLEPC. The set can be calculated by using

$${{\rm {Min}}}\,G\left(x\right)=\left[{g}_{1}(x),{g}_{2}\,(x)\right]$$
(2)

The constraint condition can be defined as

$${g}_{1}\left(x\right)=1/\left({U}_{2100}-{U}_{2020}\right)$$
(3)
$${g}_{2}\left(\cdot \right)={\sum }_{t=2020}^{2100}{P}_{{{\rm {urbanland}}}}$$
(4)

where g1 (·) is the objective function in terms of the minimization of urban land consumption and g2 (·) is the maximization of PBUAs. U and Purbanland are the total BUAs of global and PBUAs, respectively.

The constraint of the objective function should be considered in terms of PCR, LCR, and PCR/LCR. We solved the multi-objective optimization problem by using the Monte Carlo simulation strategy. For the developed countries with high incomes, our optimization objective was to minimize land consumption of BUAs while ensuring adequate human settlement. For regions with rapidly increasing urbanization rates, we determined the Pareto frontier for the TLEPC scheme. To determine an approximate maximum turning of slope and the minimum distance to the upper edge, we maximized the distance and implemented its corresponding plan (refer to Supplementary Figs. 3 and 4).