Spatiotemporal changes in inter-city sustainability impacts linked to emission challenges worldwide

Abstract

Atmospheric emissions pose significant challenges to achieving the Sustainable Development Goals (SDGs). As key sources of emissions, cities are also critical actors in driving sustainable transitions. Their performance on emission-related SDG indicators is shaped not only by local factors but also by inter-city interactions. However, the spatial and temporal heterogeneity of these interactions remain poorly understood. Here, we investigate inter-city SDG interactions across 4,116 cities worldwide from 2010 to 2020, and simulate SDG performance under five interaction scenarios, including isolated and collaborative conditions. Results reveal that a 10% improvement in SDG 11.6 (particulate matter), SDG 12.4 (gaseous pollutants), and SDG 13.2 (greenhouse gases) in other cities can boost the focal city’s performance by 0.450%, 0.390%, and 0.200%, respectively. Most cities (93.51%) are influenced by other cities. Inter-city coordination can improve SDG performance by 10.81% compared to isolated efforts. These findings highlight the importance of coordinated strategies to enhance urban sustainability.

Emission-related SDG performance is not just a local issue but shaped by interconnected urban systems, as impacts like greenhouse gases and air pollution cross borders. Inter-city coordination improves outcomes by 11% compared to isolated efforts.

Introduction

The Sustainable Development Goals (SDGs), proposed by the United Nations (UN) in 2015, represent a global call to action aimed at addressing the global environmental, social, and economic challenges by 2030¹. As engines of economic growth, innovation, and societal progress, cities are at the heart of global sustainable development. Over the past decade, rapid industrialization and urbanization in cities fueled development and prosperity². The 1000 largest cities worldwide contributed 60% of global GDP in 2023. However, this expansion has also brought significant environmental challenges, including increased emissions of particulate matter, gaseous pollutants, and greenhouse gases^3,4. Among the many SDG indicators, atmospheric ones—such as SDG 11.6 (particulate matter), SDG 12.4 (gaseous pollutants), and SDG 13.2 (greenhouse gases)—warrant special attention due to their direct and immediate links to major threats to global sustainable development, such as climate change, deteriorating air quality, and serious risks to human health^5,6. SDG performance of emission-related indicators is not merely a local issue but a shared responsibility shaped by interconnected urban systems, as the consequences of atmospheric emissions, such as climate change and air pollution, extend far beyond city or national boundaries, making it a significant global challenge that demands coordinated international governance. Many countries are actively implementing policies and regulations to mitigate the impacts of these emissions^7–9. For instance, China has introduced an action plan to improve air quality, aiming for a reduction of about 10% in PM_2.5 concentrations and over 10% in NO_x emissions by 2050, compared to 2020 levels. Additionally, 27 European Union member states have set an intermediate target of reducing net greenhouse gas emissions by 55% by 2030, compared to 1990 levels. City-level governments play a central role in this process by translating national SDG targets into actionable strategies tailored to urban needs and realities^10,11.

In the context of globalization, cities are increasingly interconnected and interdependent, forming complex networks of interaction with other cities^11–13. Conventional sustainability initiatives often solved issues locally while creating new problems elsewhere, undermining global sustainability^14,15. The metacoupling framework, which analyzes human–environment interactions within systems such as cities, between neighboring systems, and across distant systems, offers a comprehensive conceptual basis for understanding and managing these interactions across regions^14,15. It promotes a transition from place-focused governance to governance that considers flows between places^14,15. The performance of emission-related SDG indicators can be influenced not only by local conditions but also by the SDG performance of other cities nearby and far away^16,17. For example, the pollution haven hypothesis suggests that regions with stringent environmental regulations may outsource pollution-intensive industries to areas with weaker regulations instead of reducing them, leading to unintended emissions increases in other neighboring and distant areas^18,19. For another example, if a multinational company brings clean technology to one city, nearby cities in its supply chain may also adopt similar practices to stay aligned²⁰. In this way, improvements in emission-related SDG indicators in one city can indirectly promote similar improvements in the interconnected cities²¹. These examples highlight the widespread cross-border impacts on emission-related SDGs, which often extend beyond city boundaries^11,22, emphasizing the need for collaborative, cross-boundary strategies to tackle shared environmental challenges.

However, inter-city SDG impacts related to emission-related SDG indicators across global cities remain largely underexplored, especially their spatial and temporal variations, leaving a critical gap in understanding how these dynamics influence the achievement of the SDGs. Inter-city SDG impacts refer to the situation that the emission reductions or increases in one city that can create synergies or trade-offs affecting the progress of emission-related SDG indicators in other cities. Depending on the nature of these impacts, they may create inter-city synergies that enhance progress or inter-city trade-offs that hinder it^23,24. More importantly, the magnitude and direction of inter-city SDG impacts are dynamic, changing over time and across locations. The magnitude can either strengthen or weaken, while the direction may shift from synergies to trade-offs and vice versa, or from inter-city relationships to no interaction. Most previous studies have focused on examining impacts between different SDGs within a specific place (e.g., city, country, or region)^25,26. However, they have largely overlooked the impacts across regional borders, such as how climate action efforts (SDG 13) in one city affect SDG progress of other cities. Addressing SDG challenges in isolation can lead to solutions that benefit one location at the expense of others. In contrast, accounting for inter-regional interactions can more effectively support global progress toward the SDGs. Although several recent studies explored the SDG impacts between regions, these impacts assumed that spatial relationships are constant over time and space¹¹ (Fig. 1a). Consequently, findings from previous studies primarily reflect the overall pattern of inter-city SDG impacts among all regions across all periods, and cannot capture how the magnitude and direction of inter-city SDG impacts fluctuate in each individual region in each period (Fig. 1b). Without accounting for the spatial and temporal heterogeneity of inter-city SDG impacts, policymakers are not only unable to design region-specific policies but also incapable of addressing challenges that vary over time, limiting their effectiveness in proposing targeted solutions tailored to particular regions and periods.

Fig. 1. The conceptual framework of inter-city SDG impacts.

a Inter-city SDG impacts of global cities across all time periods. b Inter-city SDG impacts on each city in each period. The blue and red colors represent inter-city synergistic and trade-off relationships, respectively. The conceptual framework of inter-city SDG impacts shown in Fig. 1 was built upon the metacoupling framework^14,15. In this study, the mechanisms behind these intercity influences are not based on physical flow of emissions (e.g., air pollution transport), but are instead delivered through institutional and socioeconomic linkages, such as trade, investment, labor migration, policy diffusion, shared infrastructure networks, and technology diffusion. For illustrative purposes, we used four cities to demonstrate their inter-city interactions. For example, taking city A as the focal city, its SDG indicator performance can be influenced by its neighboring cities B and C and the distant city D if they are connected through shared energy systems. The term focal city refers to the city at the center of the analysis, while other connected cities are assessed in terms of how their actions affect the focal city’s outcomes. In period 1, the inter-city SDG impacts on city A are positive (synergy), suggesting that the improvements in SDG performance of cities B, C, and D can contribute to better outcomes for city A. However, this shifts to a negative interaction (trade-off) in period 2, and further intensifies into a stronger trade-off in period 3. Base map is from publicly available Natural Earth vector map data (www.naturalearthdata.com/). The icons are obtained from Flaticon (www.flaticon.com/).

This study investigates the inter-city SDG impacts on major SDG indicators related to atmospheric emissions, including two related to SDG 11.6 (particulate matter), two to SDG 12.4 (gaseous pollutants), and two to SDG 13.2 (greenhouse gases) across cities worldwide. First, to enhance data comparability and coverage, we develop a global city-level SDG indicator system covering 4116 cities from 103 countries, accounting for over 90% of global GDP, over 11 years (2010-2020) based on a range of satellite products. While traditional statistical databases provide valuable information, they can sometimes face limitations in terms of consistency, availability, and coverage, especially when analyzing multiple regions or cities^27–29. Second, we investigate inter-city SDG interactions from neighboring cities through a detailed lens, analyzing not only overarching patterns but also the specific influences observed in each city in a period. The inter-city SDG impacts of all indicators are used to construct a spatial interaction index (ranging from 0 to 100), which quantifies the magnitude of impacts received by each city each year. Lastly, we quantify the performance of emission-related SDG indicators under five scenarios considering varying inter-city interaction conditions, including a business-as-usual scenario (Scenario 1), an isolated scenario (Scenario 2), and two collaborative scenarios (Scenarios 3-4), and one non-collaborative scenario (Scenario 5).

Results

Inter-city SDG impacts of cities worldwide over 2010-2020

We first evaluated the performance of six emission-related SDG indicators across cities worldwide. These indicators include SDG 11.6 (PM_2.5), SDG 11.6 (PM₁₀), SDG 12.4 (SO₂), SDG 12.4 (NO_x), SDG 13.2 (CO₂), and SDG 13.2 (CH₄). We then employed a spatial autoregressive model to quantify inter-city interactions in SDG performance of these emission-related indicators across global cities. The spatial autoregressive parameter in the model served as the key parameter for measuring these interactions. Inter-city synergistic effects were identified when this parameter was statistically significant and positive, indicating that improvements in emission-related SDG performance in other cities can contribute to enhancements in the focal city’s performance. Conversely, inter-city trade-off effects were observed when the parameter was statistically significant and negative, suggesting that better performance in other cities corresponded with a decline in the focal city’s outcomes.

The results show that a 10% improvement in SDG 11.6 (PM_2.5) and SDG 11.6 (PM₁₀) in other cities was found to boost a focal city’s performance on these two indicators by 0.470% and 0.430%, respectively, averaging 0.450% for SDG 11.6 (particulate matter) (Table 1). In other words, reductions in per capita emissions in other cities can contribute to mitigating per capita emissions in the focal city. Likewise, improvements in SDG 12.4 indicators (SO₂ and NOₓ) led to increased performance on these indicators of 0.510% and 0.270%, respectively, averaging 0.390% for SDG 12.4 (gaseous pollutants) (Table 1). In contrast, the effect on SDG 13.2 (greenhouse gases) was smaller: a 10% improvement in SDG 13.2 (CO₂) and SDG 13.2 (CH₄) emissions in other cities raised the focal city’s performance by only 0.140% and 0.260%, respectively—an average of 0.200% (Table 1). Overall, the magnitude of inter-city SDG impacts for both SDG 11.6 (particulate matter) and SDG 12.4 (gaseous pollutants) was ~twice as strong as those for SDG 13.2 (greenhouse gases) (Table 1). Although the inter-city SDG impacts for SDG 13.2 (greenhouse gases) were comparatively weaker, they still demonstrated positive spatial impacts, suggesting that localized strategies remained essential for greenhouse gas mitigation, supported by broader regional and global efforts to maximize impact.

Table 1.

Inter-city SDG impacts of emission-related SDG indicators across cities worldwide over 2010-2020

	SDG 11.6		SDG 12.4		SDG 13.2
	Particulate matter		Gaseous pollutants		Greenhouse gases
	PM2.5	PM10	SO2	NOx	CO2	CH4
Inter-city interaction	0.047***	0.043***	0.051***	0.027***	0.014***	0.026***
	(32.394)	(30.625)	(18.230)	(19.790)	(16.437)	(15.349)
Economic development	0.439***	0.447***	0.422***	0.539***	0.632***	0.539***
	(94.949)	(99.113)	(44.176)	(123.525)	(239.696)	(96.212)
Population density	−0.272***	−0.296***	−0.255***	−0.301***	−0.259***	−0.313***
	(−61.365)	(−68.604)	(−27.872)	(−72.007)	(−102.723)	(−58.262)
Urbanization	0.360***	0.336***	0.399***	0.245***	0.179***	0.230***
	(88.169)	(84.350)	(47.259)	(63.629)	(77.077)	(46.534)
Environmental protection	−0.418***	−0.430***	−0.557***	−0.326***	−0.111***	−0.538***
	(−8.000)	(−8.449)	(−5.158)	(-6.664)	(-3.729)	(-8.516)
Test for spatial fixed effects	184094***	190249***	149132***	159494***	225276***	173359***
Test for time fixed effects	9548***	9489***	4062***	21253***	21819***	1789***
Corr-squared	0.966	0.968	0.871	0.972	0.989	0.955
N	4116	4116	4116	4116	4116	4116
T	11	11	11	11	11	11
Log-likelihood	28676	29867	-4194	31356	54139	20021

Notes: Asymptotic t-statistic was provided in parentheses. Significance at the 1% level is denoted by ^***.

The dependent variables of the below six regression models are SDG 11.6 (PM_2.5 emission per capita), SDG 11.6 (PM₁₀ emission per capita), SDG 12.4 (SO₂ emission per capita), SDG 12.4 (NO_x emission per capita), SDG 13.2 (CO₂ emission per capita), and SDG 13.2 (CH₄ emission per capita). The model incorporates two categories of factors influencing emission-related SDG indicators: external and internal factors. The external factor is represented by the inter-city interaction variable, which captures the influence of other cities on the focal city’s SDG performance. The internal factors pertain to the characteristics of the focal city itself and include four control variables: economic development (GDP per capita), population density (population per unit area), urbanization level (urban and built-up lands per capita), and environmental protection situation (the share of forest and wetland out of total land). Detailed descriptions, the rationale for their inclusion, and data sources for these variables are provided in Table S1.

Internal factors, such as economic development, population density, urbanization level, and environmental protection, also shaped the SDG performance of the focal city. Rising economic development and rapid urbanization often lead to higher per-capita emissions of PM_2.5, PM₁₀, SO₂, NO_x, CO₂, and CH₄ (Table 1). This result was consistent with some previous studies^30,31, largely because economic development typically intensifies production and consumption activities, and many cities may not yet have reached the turning point of the environmental Kuznets curve²⁹. At the same time, urbanization expands built-up areas and increases infrastructure demand, both of which place additional pressure on the environment. As a result, performance on SDG 11.6 (particulate matter), SDG 12.4 (gaseous pollutants), and SDG 13.2 (greenhouse gases) tend to deteriorate. In contrast, stronger environmental protection efforts and higher population density generally reduce per-capita emissions (Table 1). It can be interpreted that stronger environmental protection measures can help to curb emissions of particulate matter, gaseous pollutants, and greenhouse gases. Meanwhile, dense urban areas can promote more efficient land use and wider adoption of public transit, all of which lower emissions per person. These factors therefore contribute to improved performance on SDG 11.6, SDG 12.4, and SDG 13.2.

Dominance of cities experiencing inter-city synergistic effects

We conducted a geographically and temporally weighted autoregressive model (GTWAR), which can capture variations in inter-city interactions across different cities and times. The spatial relationship of each city can be classified as inter-city synergistic effects, inter-city trade-off effects, or no impacts. Throughout 2010-2020, for the six emission-related SDG indicators, the number of cities experiencing inter-city synergistic effects was ~twice that of cities facing inter-city trade-offs (Fig. 2a, Fig. 3, and Fig. 4). Specifically, in 2010, the share of cities experiencing inter-city synergistic effects was 43.70%, 2.29 times higher than that of cities facing trade-off effects (19.13%) (Fig. 2a, Figure S1 and Figure S2). The remaining 37.17% of cities in 2010 were not affected by the SDG performance of other cities (Fig. 2a), as indicated by the zero values in Figure S1 and Figure S2. In 2015, the share of cities with synergistic effects declined to 38.50%, 1.87 times higher than those with trade-off effects (20.62%) (Fig. 2a). In 2020, the proportion of cities with synergistic effects slightly increased to 38.84%, 2.20 times that of cities experiencing trade-off effects (17.62%) (Fig. 2a, Fig. 3, and Fig. 4). For example, Hong Kong received inter-city synergistic effects from other cities in advancing SDG 13.2 (CO₂ emissions) (Fig. 4b). This pattern can be attributed to technological diffusion and industrial upgrading in other cities within the Greater Bay Area, which promote cleaner production practices in Hong Kong through supply chain linkages, coordinated industrial restructuring policies, and regional initiatives such as the Hong Kong-Guangdong Cleaner Production Partners Recognition Scheme. For some cities that show trade-off effects with others, such as Beijing (Fig. 3a and b), the results suggest that the improvement in SDG 11.6 (PM_2.5) and SDG 11.6 (PM₁₀) in Beijing may be accompanied by deterioration in its neighboring cities. This pattern can be interpreted as the result of industrial relocation and the spatial redistribution of pollution-intensive activities, whereby stricter environmental regulations and industrial upgrading in the core city shift polluting production to surrounding cities with lower regulatory pressure or production costs. Among the six SDG indicators, SDG 12.4 (NO_x) had the largest share of cities experiencing inter-city synergistic effects, at 43.52%, ~2.64 times higher than the share of cities facing trade-off effects (16.45%) (Fig. 4a). In contrast, SDG 13.2 (CO₂) had the smallest share of cities with inter-city synergistic effects, at 30.99%, which was only 1.32 times higher than the share of cities facing trade-off effects (23.39%) (Fig. 4b).

Fig. 2. Direction of inter-city SDG impacts between 2010 and 2020.

a Share of cities exhibiting no, trade-off, and synergistic effects. b Number of cities that experienced changes in the direction of inter-city SDG impacts. c Sample list of cities that underwent directional changes in inter-city SDG impacts. Due to space limitations, we selected only a subset of cities that experienced directional changes. These cities were intentionally chosen to ensure representation across all five continents (Africa, Europe, the Americas, Oceania, and Asia), allowing for a globally balanced illustration of the observed patterns. Blue represents a positive spatial autoregressive coefficient (synergistic effect), red indicates a negative coefficient (trade-off effect), and blank denotes no spatial relationship.

Fig. 3. Inter-city interaction of SDG indicators related to PM_2.5, PM₁₀, and SO₂ of each city in 2020.

a SDG 11.6 (PM_2.5). b SDG 11.6 (PM₁₀). c SDG 12.4 (SO₂). The value of inter-city interaction is captured by the spatial autoregressive coefficient, which indicates the percentage change in a local city’s SDG performance resulting from improvements in other cities’ SDG performance. For instance, if the coefficient is 0.01, then a 1% improvement in SDG 11.6 (PM_2.5) in neighboring cities would lead to a 0.01% increase in the focal city’s SDG performance. Base map is from publicly available Natural Earth vector map data (www.naturalearthdata.com/).

Fig. 4. Inter-city interaction of SDG indicators related to NO_x, CO₂, and CH₄ of each city in 2020.

a SDG 12.4 (NO_x). b SDG 13.2 (CO₂). c SDG 13.2 (CH₄). The value of inter-city interaction is captured by the spatial autoregressive coefficient, which indicates the percentage change in a local city’s SDG performance resulting from improvements in other cities’ SDG performance. For instance, if the coefficient is 0.01, then a 1% improvement in SDG 12.4 (NO_x) in neighboring cities would lead to a 0.01% increase in the focal city’s SDG performance. Base map is from publicly available Natural Earth vector map data (www.naturalearthdata.com/).

Shift in direction of inter-city SDG impacts

Between 2010 and 2020, an average of 76.87% of cities exhibited a consistent pattern in their inter-city SDG impacts across the six emission-related SDG indicators (Fig. 2b), suggesting a relatively stable dynamic in regional sustainability linkages. Notably, inter-city synergistic effects were more common, with 32.28% of cities consistently benefiting from positive spillover effects from their neighbors (Fig. 2b). In contrast, 15.15% of cities were persistently affected by trade-offs, indicating potential regional conflicts or imbalances in SDG development (Fig. 2b). Meanwhile, 29.44% of cities received no inter-city SDG impacts, implying limited cross-border influence (Fig. 2b). These findings highlight the importance of regional coordination, as cities with sustained synergies may serve as models for collaborative policy-making, while those facing ongoing trade-offs may require targeted interventions to mitigate negative externalities. For SDG 11.6, 29.88% of cities consistently experienced inter-city synergistic effects for PM_2.5, while 16.52% faced consistent trade-offs (Fig. 2b). Similarly, for PM₁₀, 36.20% of cities experienced persistent synergies, whereas 14.21% encountered persistent trade-offs (Fig. 2b). For SDG 12.4, consistent synergistic effects influenced 34.57% of cities for SO₂ and 35.42% for NO_x, while trade-offs affected 14.33% and 12.29% of cities, respectively (Fig. 2b). In SDG 13.2, 25.90% of cities faced sustained synergies and 18.71% sustained trade-offs for CO₂, while 31.68% experienced sustained synergies and 14.84% sustained trade-offs for CH₄ (Fig. 2b).

However, as shown in Figs. 2b, 23.13% of cities experienced shifts in direction of these inter-city SDG impacts between 2010–2020, including shifting from inter-city relationship to no interaction, or even from synergies to trade-offs and vice versa. These temporal shifts can be closely linked to changes in energy efficiency, which influenced the temporal evolution of inter-city SDG interactions. Cities undergoing an energy transition were particularly likely to experience significant change in their inter-city SDG impacts. For instance, cities in the state of Alabama with coal mining and related industries such as Colbert, Lauderdale, and Pickens were part of economic development regions established by the Alabama Legislature in 2015 to support the energy transition³². Over time, their inter-city SDG impacts of SDG 11.6 (PM_2.5) shifted from positive to negative, implying a transition from inter-city synergistic effects to trade-off effects (Fig. 2c). Similarly, the coal-producing counties of Harrison, Lewis, and Marion were designated as key service areas by the Mountaineer Trail Network Recreation Authority, established by the West Virginia Legislature in 2019 to support economic revitalization and energy transition efforts. In these areas, inter-city SDG impacts for SDG 12.4 (SO₂) shifted from negative to positive over the same period, indicating a transition from inter-city trade-off effects to synergistic effects (Fig. 2c).

Dynamic change in spatial interaction index over time

We proposed a spatial interaction index (0–100) to measure the overall magnitude of influence that a city receives from others based on six emission-related SDG indicators (Fig. 5). A higher index value indicates stronger inter-city SDG impacts on a city’s sustainable development performance. In 2020, up to 93.51% of cities worldwide had a spatial interaction index greater than 0, indicating that the vast majority were affected by the performance of at least one emission-related SDG indicator of other cities (Fig. 5a). In contrast, only 6.49% of cities in 2020, and 3.77% in 2010, had an index of zero, suggesting complete independence from external SDG influence (Fig. 5a). Among the six emission-related SDG indicators, SDG 12.4 (SO₂) and SDG 12.4 (NO_x) had the highest share of cities experiencing inter-city SDG impacts in 2010, at 64.82% and 65.35%, respectively (Figure S3d and e). However, by 2020, these proportions declined to 59.47% and 55.52%, indicating a potential improvement in localized emission control (Fig. 5d, e). For SDG 13.2, the proportion of cities experiencing inter-city effects from CO₂ and CH₄ emissions was 57.71% and 63.41% in 2010 (Figure S3f and g), decreasing to 51.98% and 58.03% in 2020 (Fig. 5f and g). For SDG 11.6, inter-city SDG impacts related to PM_2.5 and PM₁₀ were observed in 61.25% and 64.99% of cities in 2010 (Figure S3b and c), dropping to 54.79% and 59.03% in 2020 (Fig. 5b, c).

Fig. 5. Spatial interaction index and scores of inter-city SDG impacts of six emission-related SDG indicators of each city in 2020.

a Spatial interaction index. b Score of inter-city SDG impacts of SDG 11.6 (PM_2.5). c Score of inter-city SDG impacts of SDG 11.6 (PM₁₀). d Score of inter-city SDG impacts of SDG 12.4 (SO₂). e Score of inter-city SDG impacts of SDG 12.4 (NO_x). f Score of inter-city SDG impacts of SDG 13.2 (CO₂). g Score of inter-city SDG impacts of SDG 13.2 (CH₄). Base map is from publicly available Natural Earth vector map data (www.naturalearthdata.com/).

From 2010 to 2020, the average spatial interaction index of cities worldwide showed a downward trend, decreasing by 16.86% (Figure S3a, Fig. 5a, and Table S9). Examining the components of this index, both the synergy scores and trade-off scores also declined over the study period, by 17.63% and 23.91%, respectively (Table S9). Specifically, 75.66% of cities worldwide experienced a decreasing spatial interaction index over time, indicating a decline in the SDG impacts they received from other cities (Figure S3a, Fig. 5a). In contrast, 21.31% of cities saw an increasing spatial interaction index (Figure S3a, Fig. 5a). This indicates a decline in the magnitude of inter-city environmental linkages in emission-related production activities over time, with the performance of these six emission-related SDG indicators becoming less influenced by the SDG performance of other cities. These results point to a transition toward more independent urban sustainability trajectories, with both advantages and potential drawbacks. On the one hand, the overall decline in the average spatial interaction index suggests that cities may be becoming more self-sufficient in pursuing environmental goals and more resilient to external shocks or negative impacts from neighboring cities. On the other hand, the decline may also imply a weakening of inter-city synergies. Reduced spatial interaction can diminish opportunities for shared learning, policy diffusion, and coordinated sustainability actions. In this sense, while greater independence may strengthen local resilience, it can also lead to the loss of collective benefits that stem from cross-city collaboration and integrated sustainability governance.

Spatial interaction index by city development level and geographical location

To examine how spatial interaction index vary across different levels of development, this study categorized cities into four groups based on two different metrics that were used to measure the development level: one is economic development level and the other is Human Development Index (HDI). Both of them divide the global cities into four groups: very high, high, medium, and low. From 2010 to 2020, all groups exhibited a declining trend in their average spatial interaction indexes, suggesting a general weakening of inter-city SDG linkages over time (Fig. 6a, b, d, and e). Interestingly, cities in lower economic development groups consistently exhibited stronger spatial impacts. Specifically, the average spatial interaction index of cities in the low economic development group was 13.30 in 2010 and 11.92 in 2020, followed by the medium group (12.72 in 2010 and 10.49 in 2020), the high group (12.12 in 2010 and 9.44 in 2020), and the very high group (11.84 in 2010 and 9.70 in 2020) (Fig. 6a, b, d, and e). When classified by HDI, we observed a similar pattern: relatively less developed cities experienced stronger inter-city impacts from other cities. Specifically, cities in the medium HDI group consistently demonstrated the highest level of spatial impacts, with an average index of 15.55 in 2020, followed by the low (13.14), very high (10.17), and high HDI groups (10.15) (Fig. 6b and e). A plausible explanation for this pattern is that more developed cities (e.g., those in the very high and high economic development or HDI groups) often possess stronger local institutions, more advanced regulatory frameworks, and more resilient infrastructure, which can buffer or internalize external environmental impacts^33,34, thereby weakening spatial spillovers. In contrast, less developed cities are more deeply embedded in global and regional production networks, with rapidly expanding industrial sectors and increasing dependence on inter-city resource flows. This higher degree of economic integration may intensify cross-city interactions. Additionally, cities in the low and medium economic development or HDI groups may still be in transitional stages of development, where institutional capacity and environmental regulation remain insufficiently mature, allowing externalities to propagate more strongly across urban systems³⁵.

Fig. 6. Spatial interaction index by group and scenario analysis.

a Box chart of spatial interaction index in 2020 by economic development level. b Box chart of spatial interaction index in 2020 by HDI level. c Box chart of spatial interaction index in 2020 by continent. d Components of spatial interaction index by economic development level in 2010 and 2020. e Components of spatial interaction index by HDI level in 2010 and 2020. f Components of spatial interaction index by continent in 2010 and 2020. g Scenario analysis of six emission-related SDG indicators. The line and square inside the box chart of a, b, and c shows the median and average of the data, respectively. The box itself spans from the first quartile to the third quartile. The error bar in d, e, and f was obtained based on the standard error of the mean. The number of cities for each economic development level is 1029 cities. For the HDI categories, the sample sizes are 3,166 (very high), 765 (high), 152 (medium), and 33 (low). For the continent categories, the sample sizes are 46 (Africa), 579 (Asia), 536 (Europe), 2,926 (Americas), and 29 (Oceania). The y-axis units of g are as follows: tonne per capita for SDG 11.6 (PM_2.5 and PM₁₀), SDG 12.4 (SO₂ and NO_x), and SDG 13.2 (CH₄); and thousand tonnes per capita for SDG 13.2 (CO₂). Scenarios 1–5 refer to the business-as-usual situation, an isolated condition, a partial collaborative strategy (where trade-off effects shift to synergistic effects), a full collaborative strategy (where both trade-off and neutral effects shift to synergistic effects), and a non-collaborative strategy, respectively.

In 2020, cities in Africa and Asia exhibited higher average spatial interaction indexes compared to those in Europe, America, and Oceania (Fig. 6c, and f). This indicates that cities in these continents were more strongly influenced by the SDG performance of other cities, suggesting a higher degree of interdependence in sustainable development dynamics. Such heightened spatial impacts may reflect both opportunities and vulnerabilities. On one hand, cities in Africa and Asia could benefit more from positive spillover effects of regional cooperation; on the other hand, they may also be exposed to adverse impacts stemming from underperformance in other areas. In Asia, the top five cities with the highest spatial interaction indexes were predominantly located in Gansu Province, China, including Baiyin, Lanzhou, Wuwei, and Jinchang, with index values ranging from 48.99 to 53.89 (Table S14). These cities are either resource-based—relying heavily on the extraction and processing of natural resources such as forests, minerals, oil, and natural gas—or are closely connected to such cities. The clustering of cities with high spatial interaction indexes within certain regions, such as Gansu Province, suggests that the observed inter-city linkages may not solely reflect spatial proximity but also deeper structural and developmental dynamics. In this context, spillover effects and industrial agglomeration effects should not be viewed as separate or competing explanations; rather, they are interrelated and potentially causal. Spillover effects, such as the diffusion of technologies, shared infrastructure, and policy learning among neighboring cities, can serve as the underlying mechanisms that drive industrial agglomeration over time. Conversely, once industrial agglomeration emerges, it can further reinforce inter-city interactions through intensified trade, labor mobility, and environmental linkages. The high spatial interaction indexes observed in Gansu Province likely reflect such a co-evolutionary process, in which inter-city interactions contribute to the formation of industrial clusters, while industrial agglomeration, in turn, amplifies spatial dependence among cities.

SDG performance under varying inter-city interaction scenarios

We assessed the performance of six emission-related SDG indicators under five inter-city interaction scenarios (see “Methods” for details), including business-as-usual scenario (Scenario 1), isolated scenario (Scenario 2), and two collaborative scenarios (Scenarios 3 and 4), and one non-collaborative scenario (Scenario 5). Under Scenario 1, which accounts for inter-city SDG impacts, per capita emissions of PM_2.5, PM₁₀, SO₂, NO_x, CO₂, and CH₄ decreased by 2.03%, 2.98%, 5.54%, 4.89%, 1.65%, and 1.94%, respectively—an average reduction of 3.17% compared to the scenario without inter-city effects (Scenario 2) (Fig. 6g). In Scenario 3, where all trade-off relationships were transformed into synergistic ones, reductions reached 5.66%, 6.51%, 9.24%, 7.09%, 9.54%, and 6.40%, averaging 7.41% (Fig. 6g). Scenario 4, which transforms all no impacts and trade-off relationships into synergistic ones, yielded the most significant improvement with average reductions of emission per capita at 10.81% (Fig. 6g). Shifting from isolated efforts (Scenario 2) to collaborative urban strategies (Scenarios 3 and 4) can help to reduce environmental burdens (Fig. 6g). However, transitioning from isolated efforts (Scenario 2) to non-collaborative urban strategies (Scenario 5) can significantly increase emission burdens. Scenario 5 (non-collaborative strategy) involved converting both no impacts and synergistic effects into trade-offs, resulting in per capita emission increases of 9.20%, 11.20%, 14.91%, 8.65%, 7.79%, and 9.05%, with an average increase of 10.13% (Fig. 6g). This highlights the potential negative consequences of non-collaborative strategies.

Inter-city impacts beyond geographical proximity

Defining “neighboring” cities solely based on geographical proximity cannot capture other important types of linkages between cities, particularly among cities that are far away. In many cases, institutional dissimilarity, such as divergent governance structures or misaligned policy orientations, can separate cities more clearly than physical distance. Conversely, institutional similarity, reflected in shared governance features or coordinated policies, may create close relationships that transcend geography. Therefore, this study moves beyond geographical proximity by examining both institutional dissimilarity and similarity, which correspond to “distant” and “neighboring” relationships, respectively, in an institutional sense (see Note S2 in Supplementary Information). The results indicate that inter-city impacts between cities with institutional dissimilarity were generally weaker than those between cities with institutional similarity. Specifically, a 10% improvement in SDG 11.6 (PM₂.₅) and SDG 11.6 (PM₁₀) in other cities with institutional dissimilarity increased the focal city’s performance by 0.26% and 0.27%, respectively—both smaller than the corresponding effects (0.49% and 0.44%) for institutionally similar cities (Table S2 and Table S3). Similarly, a 10% improvement in SDG 13.2 (CO₂) and SDG 13.2 (CH₄) in dissimilar cities enhanced the focal city’s performance by 0.02% and 0.07%, both smaller than the effects observed for similar cities (0.15% and 0.27%) (Table S2 and Table S3). For SDG 12.4, a 10% improvement in SO₂ and NOₓ performance in other cities with institutional dissimilarity changed the focal city’s performance by 0.11% and −0.03%, respectively (Table S3). In contrast, improvements in institutionally similar cities can lead to stronger positive impacts of 0.54% and 0.29% (Table S2). The negative impact for SDG 12.4 (NOₓ) under institutional dissimilarity suggested potential policy misalignment or industrial relocation effects, where emission reductions in one country may coincide with increased production and related emissions in another. These results highlight the importance of institutional coherence in strengthening inter-city SDG interactions. Stronger positive interactions among institutionally similar cities suggest that harmonized regulatory frameworks, shared environmental standards, and coordinated governance structures can amplify SDG progress across borders. Conversely, weaker or even negative interactions among institutionally dissimilar cities indicate that fragmented policies may diminish cooperation benefits and, in some cases, shift environmental burdens elsewhere.

Discussion

This study investigated the spatial and temporal heterogeneity of the inter-city SDG impacts related to emission-related SDG indicators across 4116 of the world’s cities from 2010 to 2020. Our findings revealed that inter-city SDG impacts play a role in shaping cities’ emission-related SDG performance, underscoring the importance of viewing urban sustainability through a networked, interdependent lens rather than an isolated and city-specific perspective. Widespread inter-city influence (93.51% of cities affected in 2020) indicated that the majority of cities were not acting in silos when it came to emissions-related SDG outcomes. This interconnectedness opens up opportunities for collaborative planning and policy coordination, particularly in regions sharing similar environmental challenges or economic ties.

The shift in the direction of inter-city SDG impacts highlighted that cities were not confined to static roles within global emission networks; rather, they can transition from being sources of inter-city trade-offs to contributors to synergistic outcomes, and vice versa. This temporal change can be attributed to several factors. First, structural shifts in cities’ economic activities and production systems can play a significant role. Cities often undergo transitions in their industrial composition, such as moving from heavy industries toward service-oriented sectors or adopting cleaner technologies. For example, when a city reduces its reliance on carbon-intensive industries, the negative spillovers it imposes on other cities within the same supply chain may diminish, shifting interactions from trade-offs toward neutral or even synergistic outcomes. Conversely, industrial expansion can intensify cross-city environmental impacts, moving relationships from synergy back toward trade-off. Second, the strength and consistency of policy interventions and environmental governance reforms can also drive these shifts. For example, cities changing to relatively weak environmental regulations may function as “pollution shelter” attracting pollution-intensive production displaced from cities with stricter jurisdictions^36,37.

By comparing five inter-city interaction scenarios, we demonstrated that shifting from isolated urban efforts to more collaborative strategies can improve the SDG performance. Specially, fully converting trade-offs and no impacts into significant synergies can boost SDG performance by 10.81% compared to isolated efforts. However, transitioning from isolated efforts to non-collaborative urban strategies can significantly increase emission burdens by an average increase of 10.13%. Policymakers should recognize the value of fostering cross-city synergies and mitigating trade-offs, especially in regions with dense urban networks, to accelerate progress toward global sustainability targets. One prominent example is carbon leakage, where companies relocate production to countries with less stringent climate regulations, thereby shifting emissions abroad instead of reducing them. To combat this issue, the European Union introduced the Carbon Border Adjustment Mechanism. Additionally, technology diffusion programs, such as the C40 Cities’ Clean Air Accelerator, facilitate the sharing of low-emission technologies among global cities. Cross-city learning platforms like ICLEI and the Global Covenant of Mayors provide mechanisms for cities to adopt best practices from peers facing similar environmental challenges. Regulatory harmonization efforts, such as the European Union’s shared air quality standards and climate targets, align cities under a unified environmental framework, reducing cross-border externalities.

There are several limitations of this study. First, to ensure the accuracy of data derived from global gridded datasets, cities with an area smaller than 50 square kilometers were excluded. Small cities, whose inter-city interactions may differ, are not captured in the analysis. Second, world cities are intrinsically linked through trade and the associated embodied emissions. The lack of fine-scale city-level data, such as MRIO tables and detailed emission inventories, limits the ability to analyze embodied emissions associated with trade of global cities. This study only demonstrates the inter-city interaction of embodied emissions of Chinese cities, as shown in the supplementary information. In the future, with the advancement of data integration and urban accounting systems, such analyses could be conducted to examine how the embodied emissions of one city are influenced by those of other cities through interconnected trade relationships.

Methods

Coverage of cities worldwide

We utilized Functional Urban Areas (FUA) and Global Administrative Areas (GADM) to account for the varying definitions of “a city” across countries worldwide³⁸. The FUA, as defined by the Organization for Economic Co-operation and Development (OECD) and the European Union, are contiguous local units encompassing a central city and its surrounding areas, with boundaries determined by population density and travel-to-work patterns³⁹. For OECD countries, we used the OECD FUA, which offers higher quality data for these regions⁴⁰. For non-OECD countries, we relied on the Global Human Settlement FUA. In China, GADM level-2 administrative areas were used for prefecture-level cities, while in the United States, they were applied to counties. To ensure the accuracy of data obtained from global gridded datasets, which have spatial resolutions ranging from 100 meters to 0.1 degrees, we excluded the city area smaller than 50 square kilometers. Thus, a total of 4116 cities covering 103 countries from 2010 to 2020 were included in the analysis. The spatial distribution of these cities is depicted in Fig. 3.

Inter-city SDG impacts of emission-related SDG indicators across cities worldwide

This study proposed four interconnected steps to assess how the performance of emission-related SDG indicators of other cities influenced that of the focal city from two perspectives, the overall pattern across all cities and periods, and the city-specific pattern in each period (Fig. 1). The first perspective examines the overall situations of cities worldwide, encompassing relationships that have no impact, synergistic, or characterized by trade-offs, as shown in our previous study²⁹. While this provides a broad understanding, it does not provide insights into the specific locations of these impacts or how they change over time. In contrast, the second perspective complements the first by providing a more detailed view of how individual cities interact with others in each period.

Step 1: Construct emission-related SDG indicators

We first evaluated the performance of six emission-related SDG indicators of cities worldwide during 2010-2020. These emission-related SDG indicators were selected mainly based on the official list of global Sustainable Development Goal indicators proposed by the UN, and supported by the reports from international institutions and published papers^10,41,42. The six indicators analyzed include SDG 11.6 (PM_2.5), SDG 11.6 (PM₁₀), SDG 12.4 (SO₂), SDG 12.4 (NO_x), SDG 13.2 (CO₂), and SDG 13.2 (CH₄), as detailed in Table S1. These indicators were sourced from a range of satellite products, including EDGAR, ODIAC, and WorldPop Global Project Population Data (Table S1). To calculate city-specific emissions, we applied the zonal statistics tool, which involves overlaying city boundary polygons with the gridded data and aggregating values within each city’s boundaries. The resulting emission data were then divided by the population within each city’s boundaries, which can be used to represent the performance of these six SDG indicators.

Step 2: Quantify overall pattern of inter-city SDG impacts

We evaluated whether there is temporal and spatial autocorrelation for these SDG indicators. If no autocorrelation is found, ordinary least squares models are sufficient for capturing spatiotemporal variations. However, if autocorrelation is present, the spatial econometric model is more suitable. To evaluate this, we employed Global Moran’s I. Table S5 shows that Moran’s I values are significantly positive. This suggests that spatial autocorrelations exist in these SDG indicators. Consequently, we used a spatial econometric model, that is spatial autoregressive model, to quantify this inter-city SDG impacts across global cities during 2010–2020⁴³, as demonstrated in the following sections:

$$y_{it}=\rho {\sum }_{j=1}^N{w}_{ij}{y}_{jt}+{\mathit{\beta }}_{\mathbf{1}}{\mathbf{X}}_{\mathbf{1it}}+{\eta }_t+{\mu }_i+{\epsilon }_{it}$$ 1

$y_{it}$ was the dependent variable. $\rho$ was the coefficient to examine whether there were inter-city SDG impacts across global cities. This study used the inverse distance weight to construct the spatial weight matrix, allowing us to represent neighboring cities based on their geographical proximity. $w_{ij}$ indicated the spatial weight between city $i$ and city $j$, which can link the focal city and other cities together. This spatial weight was determined based on Haversine formula which measures the distance between two points taking into account the curvature of the Earth. Closer proximity between cities suggests stronger inter-city connections and, consequently, higher spatial weight values. ${\mathbf{X}}_{\mathbf{1it}}$ was a list of explanatory variables. ${\mathit{\beta }}_{\mathbf{1}}$ is a vector of the coefficients of the explanatory variables. The explanatory variables included economic development, population density, urbanization, and environmental protection, with their data sources, such as MCD12Q1.061 MODIS Land Cover Type Yearly Global 500 m, detailed in Table S1. ${\eta }_t$, ${\mu }_i$, ${\epsilon }_{it}$ indicated the time fixed effects, unit fixed effects, and time-unit interaction fixed effects. The test for spatial fixed effects and test for time fixed effects were conducted to assess the presence of unobserved spatial and temporal heterogeneity in the spatial autoregressive model. The results indicate that both spatial fixed effects and time fixed effects were statistically significant at the 1% level, suggesting that two-way fixed-effects models should be used to quantify the overall pattern of inter-city SDG impacts (Table 1).

Step 3: Quantify inter-city SDG impacts received by each city in each period

We conducted a GTWAR, which can capture variations in relationships across different cities and times. GTWAR model can account for the spatial and temporal heterogeneity^44,45. We constructed the GTWAR model as follows:

$$y_i={\beta }_0\left({\mu }_i,v_i,t_i\right)+\rho \left({\mu }_i,v_i,t_i\right){\sum }_{j=1}^N{w}_{ij}{y}_j+{\sum }_{l=1}^L{\beta }_l\left({\mu }_i,v_i,t_i\right)x_{il}+{\epsilon }_i$$ 2

where $y$ represented the dependent variable, that was emission-related SDG indicator. $x_{il}$ denoted explanatory variables (Table S1). ${\sum }_{j=1}^N{w}_{ij}{y}_j$ denoted a spatially lagged dependent variable.${\mu }_i$, $v_i$, and $t_i$ represents the latitude, longitude, and year of the city $i$, respectively. $\rho$ indicated spatial autoregressive parameter, while $\beta$ was the coefficient of the explanatory variables, estimated for each city during period. To estimate parameters in the GTWAR model, we used the Euclidean distance and Gaussian distance-decay-based functions to construct an N×N spatial-temporal weight matrix, where the diagonal elements represent the spatial relationship between the kernel and neighboring samples. We used the adaptive bandwidth, which was determined by the Akaike Information Criterion.

Based on the GTWAR model, we quantified the inter-city SDG impacts covering 4116 cities each year for every emission-related SDG indicator, yielding a total of 271,656 spatial autoregressive parameters (4116 cities × 11 years × 6 indicators). These parameters captured significantly positive, significantly negative, and statistically insignificant spatial relationships. A 10% significance level was used as a threshold to identify spatial impacts that are statistically meaningful, as per the previous study²⁹. Parameters not meeting this criterion were treated as statistically insignificant and assigned a value of zero, indicating no impacts with neighboring cities²⁹.

Step 4: Develop the spatial interaction index for each city in every year

We quantified the scores of inter-city SDG impacts for each emission-related SDG indicator of each city in each year. Given that the spatial autoregressive parameters were comparable across indicators, we applied min-max normalization to the absolute values of both positive and negative parameters, scaling them to a range of 0 to 100. The normalization formula is as follows:

$$I_{a,i,t}=100\times \frac{\mid {\rho }_{a,i,t}\mid -\min \left(\mid {\rho }_{a,i,t}\mid \right)}{\max \left(\mid {\rho }_{a,i,t}\mid \right)-\min \left(\mid {\rho }_{a,i,t}\mid \right)}$$ 3

where $I_{a,i,t}$ indicated the normalized value, namely, the score of inter-city SDG impacts of indicator $a$ for city $i$ at $t$ th year. ${\rho }_{a,i,t}$ was the spatial autoregressive parameter of indicator $a$ for city $i$ at $t$ th year. After obtaining the scores for six emission-related SDG indicators, we calculated their average to represent the overall magnitude of inter-city SDGs for each city in each year. This measure, referred to as the spatial interaction index, was determined using the following equation:

$$I_{i,t}={\sum }_{a=1}^N\frac{1}{N}{I}_{a,i,t}$$ 4

where $I_{i,t}$ indicated the spatial interaction index for city $i$ at $t$ th year. The spatial interaction index, ranging from 0 to 100, quantifies the extent of inter-city SDG impacts each city experiences. It reflects the overall influence of emission-related SDG indicators on the city, with a higher index indicating stronger inter-city SDG impacts. $N$ means the number of indicators of the emission-related SDG ($N$ = 6).

The spatial interaction index can be decomposed into two components: inter-city synergistic effects and trade-off component. The component of inter-city synergistic effects ($S_{i,t}$) and trade-off effects ($T_{i,t}$) were obtained based on the below equations:

$$S_{i,t}=I_{i,t}\times \frac{{\sum }_{i=1}^M{\rho }_{i,t}^{+}}{{\sum }_{i=1}^M{\rho }_{i,t}^{+}+{\sum }_{i=1}^M\mid {\rho }_{i,t}^{-}\mid }$$ 5

$$T_{i,t}=I_{i,t}\times \frac{{\sum }_{i=1}^M\mid {\rho }_{i,t}^{-}\mid }{{\sum }_{i=1}^M{\rho }_{i,t}^{+}+{\sum }_{i=1}^M\mid {\rho }_{i,t}^{-}\mid }$$ 6

${\rho }_{i,t}^{+}$ and ${\rho }_{i,t}^{-}$ respectively represent the spatial autoregressive parameter with positive and negative values.

Inter-city SDG impacts across cities by development level and geographical location

We calculated the average spatial interaction index based on city development levels measured by two different metrics: one is GDP per capita, and the other is Human Development Index (HDI). GDP per capita is a widely used indicator of economic development and is commonly applied for group classification using quartile-based methods. In this study, we classified cities into four development groups based on their GDP per capita, with 1029 cities in each group: low development (bottom 25%), medium development (25–50%), high development (50–75%), and very high development (top 25%). HDI is a composite measure that evaluates progress in three key dimensions of human development: health and longevity, education and knowledge, and a standard of living⁴⁶. As a capability-based metric grounded in Amartya Sen’s human development theory, HDI reflects not only economic performance but also the broader socio-institutional conditions that shape people’s well-being and societal development. Because these underlying capabilities influence governance quality, environmental management, institutional robustness, and technological adoption, HDI provides a theoretically meaningful framework for differentiating cities’ developmental contexts. Thus, HDI has been widely used as a basis for group classification in empirical studies^47,48. In this study, global gridded HDI datasets in 2015 at a 5 arc-minute resolution were used to classify cities into four categories, following the United Nations Development Program’s classification⁴⁹. The categories are as follows: very high performance (HDI above 0.800) with 3166 cities, high performance (HDI between 0.700 and 0.799) with 765 cities, medium performance (HDI between 0.550 and 0.699) with 152 cities, and low performance (HDI below 0.550) with 33 cities. Last, we calculated the average spatial interaction index based on geographic locations, dividing the cities into six groups: 46 cities in Africa, 579 in Asia, 536 in Europe, 2926 in America, and 29 in Oceania.

Performance of emission-related SDG indicators under five scenarios

Spatial autoregressive parameter, $\rho \left({\mu }_i,v_i,t_i\right)$, refers to the situation that the increase in 1% in other cities will result in $\rho \left({\mu }_i,v_i,t_i\right)$% in focal city. We designed five scenarios to evaluate the performance of emission-related SDG indicators of each city under varying inter-city interaction conditions by changing the spatial autoregressive parameter. Scenario 1 represented the business-as-usual case by incorporating inter-city SDG impacts. It accounted for the influence measured by each city’s average spatial autoregressive coefficient over the study period (2010–2020). This average value over the study period served as the change rate of the SDG performance. Scenario 2 was an isolated condition, assuming no inter-city SDG impacts. This meant the spatial autoregressive coefficient was set to zero, and the 2020 SDG performance values remained unchanged. Scenarios 3 and 4 represented collaborative strategies. Scenario 3 assumed that all cities previously experiencing inter-city trade-off effects shifted to synergistic effects, meaning all negative spatial autoregressive coefficients were replaced with the average of the positive coefficients. Scenario 4 built on this by assuming that all cities with either no inter-city SDG impacts or trade-off effects transitioned to synergistic effects, replacing all zero and negative coefficients with the average of the positive ones. In contrast, Scenario 5 reflected a non-collaborative strategy, assuming that both no inter-city SDG impacts and synergistic effects shifted to trade-off effects. In this case, all zero and positive spatial autoregressive coefficients were replaced with the average of the negative coefficients.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Z.X. and M.L. supervised and designed the study. H.X. wrote the original draft of the manuscript. H.X., C.Y., S.X., and J.Z. collected the data. J.L., C.A., Z.X., C.Y., S.X., M.L., J.R., Q.W., and H.X. reviewed and revised the manuscript. H.X. and S.X. prepared the figures. H.X., S.X., and C.A. interpreted the results. H.X. and S.B. established the model and compiled the code. M.L. and Q.W. contributed research resources, and were responsible for funding acquisition.

Peer review

Peer review information

Nature Communications thanks Elvira Uyarra, Matt Ziembla, and the other, anonymous, reviewers for their contribution to the peer review of this work. A peer review file is available.

Funding

This research is supported by the Hong Kong Research Grants Council’s collaborative research funds (project no. C6032-21G to M.L.), general research fund (project no. 16300424 & 16215820 to M.L.), theme-based research scheme (project no. T22-501/23 R, T22-607/24-N to M.L.), NSFC/RGC Collaborative Fund (project no. CRS_PolyU503/23 to M.L.), “1 + 1 + 1”Joint Funding Program (project no. 2025A0505000018 to M.L.), National Science Foundation (2118329 to J.L.), Michigan AgBioResearch (to J.L.), and the Excellent Young Scientists Fund from the National Natural Science Foundation of China (Grant No. 42422105 to Z.X.). All co-authors at HKUST acknowledge the support by the Otto Poon Center for Climate Resilience and Sustainability at HKUST. This research has received funding from Global STEM Professorship (to Q.W), Hong Kong SAR Government (P0039329 to Q.W), and Hong Kong Polytechnic University (P0046482 and P0038446 to Q.W). This work is part of the United Nations Educational Scientific and Cultural Organization (UNESCO) International Decade of Sciences for Sustainable Development (2024–33) and contributes to the Seamless Prediction and Services for Sustainable Natural and Built Environment (SEPRESS) Program (2025–32), an initiative endorsed under this global framework.

Data availability

The data generated in this study are available in Supplemental Information (Table S2-S15) and at GitHub repository (https://github.com/HuijuanXIAO/Spatial-Economics-Model). The SDG indicator data were obtained from the following open-access databases: the Emissions Database for Global Atmospheric Research (EDGAR), the Open-Data Inventory for Anthropogenic Carbon Dioxide (ODIAC), and the WorldPop Global Project Population Data.

Code availability

The MATLAB code and processed datasets used in this study are publicly available at GitHub repository (https://github.com/HuijuanXIAO/Spatial-Economics-Model).

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at (10.1038/s41467-026-73515-0).