Main

Phasing out fossil energy from transport, heating and cooling and other major emitting processes is key to achieving carbon-neutral cities1,2. However, an issue often overlooked in energy systems and environmental engineering is how to spatially organize and use nature-based solutions (NBS), which can play a critical role in addressing the causes and consequences of climate change3,4. In terms of reducing carbon emissions, most attention has been paid to the direct effects, that is, carbon sequestration in vegetation, soil5 and wetlands6. However, it is estimated that carbon sequestration can offset only a limited proportion of total anthropogenic carbon emissions, especially in urban settings7. To understand the full climate neutrality potential of NBS, comprehensive impacts should be estimated and quantified, including direct and indirect impacts on social and economic systems8.

Carbon emissions mitigation through NBS involves ecosystem services and green infrastructure (GI) approaches that support human wellbeing, saving resources and costs, and sequestering carbon emitted from human activities4,9. For example, urban agriculture in combination with the greening of streetscapes can promote pro-environmental behaviours, for example, nudging local recreational bicycle trips instead of long-distance driving, while also providing educational and participatory opportunities that promote consumer preferences for foods and products of lower environmental impacts and carbon emissions10,11. Additionally, microclimate regulation, combined with the aesthetic ecosystem services of NBS, can promote cycling and walking, prevent urban sprawl, alleviate dependence on automobiles, and reduce heating and cooling load12,13,14. When all benefits are considered, NBS can contribute much more to urban climate neutrality goals than mere carbon sequestration effects.

Cities are ideal experiment and innovation hubs for technologies, instruments and policies for carbon neutrality15. For example, the European Union (EU) has pledged to decrease net emissions by 57% by 2030 (compared with 1990 levels), with the inclusion of land-use and carbon sequestration goals16. This climate action is to be led by 100 cities in EU member states that have pledged to achieve climate neutrality by 2030 (ref. 17). At the same time, the use of NBS is in line with EU climate policy and goals for addressing the challenges of climate change18. However, the current policy programme does not address or identify opportunities for NBS to mitigate carbon emissions beyond direct sequestration. Moreover, due to the limited land resources in cities, the spatial allocation and configuration of urban NBS need to be optimized to achieve maximum carbon emissions reductions.

In this Analysis, we assessed and quantified five potential carbon emissions reduction mechanisms for different types of NBS in EU cities. On the basis of sector-wise carbon emissions and the local context of 54 major European cities, we spatially allocated these five categories of NBS to each city and estimated the emissions reduction potential for each sector and city. One NBS implementation was allocated on each land use grid (30 m × 30 m) but could be functional in different categories (that is, GI that saves energy consumption while also functioning as carbon sequestration). We then compared the estimated emissions reduction potential against the 2030 climate neutrality targets of the 54 cities, to assess the potential contributions of prioritized NBS to these targets.

Multiple pathways of urban NBS to reduce carbon emissions

We first identified types of NBS linked to the effects and mechanisms of carbon emissions reduction. From established definitions of NBS in the literature19,20,25, we end by suggesting that the EU Commission should implement NBS-based spatial urban planning as a key strategy to achieve carbon-neutral cities.

Methods

We applied a four-step approach in estimating and projecting the contributions of NBS to the climate-neutral goals of the 54 EU cities. First, we conducted a systematic meta-analysis to estimate the effects of five categories of NBS implementations on carbon emissions reductions from three sectors (transport, residential and industrial), in addition to carbon sequestration. Second, we spatially allocated NBS implementations to 30 m × 30 m land use grids for each of the 54 cities. The allocation was based on carbon emission disaggregated on land-use grids, and on socioeconomic context factors for different places in each city. Third, we calculated and summarized the carbon emissions reduction potential for the NBS implementations allocated to each of the 54 cities. Finally, we compared and projected the contribution of the NBS implementations to the climate neutrality goals of the 54 European cities for 2030.

Meta-analysis of carbon emissions reduction potential of NBS

To better communicate the effects of NBS, we integrated findings from previous studies on the level of benefit of different types of NBS for carbon neutrality. We applied the qualitative meta-summary techniques proposed by Sandelowski and Barroso33 to summarize the mechanisms proposed in the literature. Meta-summary techniques are particularly useful for this purpose as they synthesize a combination of qualitative and quantitative research findings. Researchers have approached the subject of carbon emissions mitigation through a variety of analytical methods that include statistical modelling, simulation, case studies, surveys and historical data analysis. We used the meta-summary method to (1) extract relevant statements on findings from each article; (2) reduce these statements to abstract findings that included the direction and intensity of carbon mitigation effect, and the local social and economic conditions in which these methods are applied; and (3) thematize and categorize findings into key NBS strategies that were considered in subsequent analysis.

We performed a literature search in the Web of Science database on 16 July 2022, using multiple search queries combining keywords associated with NBS and carbon emissions issues (Supplementary Material 3 Table A3.1). From the list of papers retrieved (578 articles), we eliminated those repeated between the NBS groups and selected papers by reviewing titles and abstracts to check if they: (1) focused on pathways towards zero carbon emissions (rather than/in addition to estimating the gross volumes and trends in carbon emissions); (2) assessed or quantified the efficacy of carbon emissions reduction strategies; (3) investigated NBS strategies (not non-NBS approaches such as the use of non-fossil fuel energy sources, low-carbon subsidy policies and so on); and (4) reported transferable metrics (such as percentage change, value per unit, and elasticity) that could be applied to different places.

The full text in the remaining 54 articles was reviewed against the research questions. In this selection phase, we looked for NBS interventions that involve direct mitigation (for example, climate regulation and carbon sink) or indirect mitigation (for example, interventions that could influence human behaviour towards low-carbon travel), aiming to include cases representing different NBS approaches to the greatest extent possible. We also intentionally covered research conducted in multiple countries and regions of the world. In total, 22 articles were included in the review (with the selection flow chart in Supplementary Material 3 Fig. A3.1).

The basic characteristics of the 22 studies are summarized in Supplementary Material 3 Table A3.2. One study focused on urban reforestation at a global scale, while the other studies were conducted in eight countries/regions, namely Europe (n = 4), China (n = 6), the United States (n = 4), Canada (n = 3), South Korea (n = 2), Japan (n = 1), New Zealand (n = 1) and Malaysia (n = 1). All studies were published between 2010 and 2020.

To synthesize results from different studies, a meta-analysis requires common measures of effect size. Our selected metrics were: (1) the percentage of emissions that can be offset by carbon sequestration of NBS, and (2) the carbon reduction rate, expressed as tons per hectare per year. We chose these two metrics not only because they were both widely employed in the literature, thus reducing the amount of unit conversion required, but also because they offer flexibility for evaluating carbon emission reduction effects through different mechanisms. In this study, we used percentage metrics for regional-scale sequestration effects from GI, street trees and green pavements, urban green spaces and agriculture, and habitat preservation and remediation, while using carbon reduction rate in tons per hectare per year for local-scale carbon reduction effects through direct and indirect interventions such as implementing street trees and green pavements, urban green spaces and agriculture, and green buildings. We extracted statements indicating the relationship between studied NBS and carbon reduction and synthesized the indicators to describe the impact of NBS on carbon emissions. Supplementary Material 2 Table A2.2 describes the statements from our selected studies and the location- or environment-related variables for NBS design, with the synthesized results of the average, upper and lower boundaries of carbon reduction potential for each type of NBS (Supplementary Material 2 Fig. A2.1).

There were some limitations in the NBS meta-analysis presented in this paper. First, differences in study design, participants, interventions and outcomes made it difficult to compare the results across studies. This limitation is grounded in the nature of the meta-analysis process itself, which relies on data from multiple studies and is only as good as the quality of those studies. However, inclusion of as many studies as possible worldwide is important, to mitigate any methodological issues nested in one study. Second, to provide more accurate results, future studies should examine the assumptions behind the values more critically. Criteria such as soil type, the time lag for tree growth, and carbon emissions during NBS implementation work should be included in the discussion. Third, we based our study on Europe, but drew on experiences gained in case studies in many countries outside Europe, to a large extent due to the scarcity of literature. Adapting values from one place to another proved more difficult than learning from analyses conducted within a single context. However, our aim was to provide a thorough quantitative evaluation of the carbon mitigation benefits of different types of NBS, which has not been done previously. By synthesizing data from a wide range of studies conducted in various cities around the world, some of which have similar urban densities, social structure and behavioural characteristics as European cities, we were still able to gain valuable insights into develo** effective NBS strategies for European cities.

Spatial allocation of NBS implementations

Spatial allocation of the most cost-effective NBS for emissions reduction was based on sectoral carbon emissions in each land use grid (30 m × 30 m) for the transport, residential and industrial sectors in each city. Certain NBS types were considered most effective in reducing emissions in each sector, such as green buildings for residential emissions, road greening for transport emissions, and green–blue infrastructure for industrial emissions.

Global Carbon Grid (GID) data from the Global Infrastructure Emissions Database (http://gidmodel.org.cn/) were disaggregated to assess the sources of sectoral carbon emissions in each land use grid. The GID establishes 0.1° × 0.1° CO2 emissions maps (year 2019) for six source sectors: power, industry, residential, transport, ship** and aviation34,35. In this study, we assessed the carbon mitigation effects of spatially allocating NBS within three sectors: industrial, residential and transport. Socioeconomic and ecosystem co-variates used to disaggregate carbon emissions into a land use grid in our study included population density, building density, land use structure, industrial and commercial units, and road networks, to enable identification of the emissions sources at a fine spatial scale. The data were acquired from EU or global data sources that are open to the public, including Urban Atlas, EuroStat and OpenStreet Map. Detailed descriptions of the data and sources are provided in Supplementary Material 2 Table A2.1. Verification of the sectoral emissions data can be found in Supplementary Material 1.

GID sectoral carbon emission data were disaggregated to the land use grid to identify the emissions sources. For transport emissions, there is a positive correlation between vehicle-kilometres travelled and road classes, and between vehicle-kilometres travelled and population density36. In this study, we first used binary dasymmetric map** (road = 1, non-road = 0) to distribute emissions to roads, and then adjusted the emissions for each road segment cell by road classes and population density. Transport carbon emissions were calculated as:

$${E}_{t}=\mathop{\sum }\limits_{s=1}^{S}\frac{{C}_{t}\times {E}_{s}}{{C}_{s}}$$
(1)

where Et are disaggregated to emissions source, but not adjusted, emissions in target zone t; Es are emissions in source (large) zone s; Ct is the count of road cells in target zone t; and Cs is the count of road cells in source zone s, including motorways, primary roads and secondary roads.

The adjusted emissions for each land use grid were then calculated as:

$${e}_{t}={E}_{t}\times {W}_{{{\mathrm{class}}}}\times {W}_{{{\mathrm{pop}}}}$$
(2)

where et is the adjusted transport emissions in kt CO2-eq per land use grid; Wclass is the weight of road classes (we used a standardized posted speed on each road segment as a proxy and normalized the values to 0–1); and Wpop is the weight of population density, normalized to 0–1.

For residential emissions, we constructed a relationship between residential carbon emissions and population and building density, using population density data from EuroStat and the urban fabric density classifications (11100–11300) in the Urban Atlas database as our model inputs. The relationship took the form:

$${e}_{{r}}=f({{{\mathrm{Den}}}}_{{{\mathrm{pop}}}},{{{\mathrm{Den}}}}_{{{\mathrm{building}}}})$$
(3)

where er is residential emissions in kt CO2-eq per land use grid; Denpop is population density; and Denbuilding is building density, normalized to 0–1.

When estimating the function f(.) between residential carbon emissions and population and building density, we tested different functional forms (linear model, local polynomials and tree-based machine learning model), with a cross-validation method for model selection (70% observations as training data, 30% as test data). We found that the random forest model produced the lowest root mean square error. Thus we applied the random forest model to obtain residential CO2 emissions estimates.

For industrial emissions, we applied binary dasymmetric map** (industry = 1, non-industry = 0) to distribute emissions to industrial and commercial complexes that cause major carbon emissions. We used the land cover type 12100 in the Urban Atlas land use database to represent this category, which contains sites including industrial activities, major commercial sites, energy plants and sewage treatment plants. The equation for industrial carbon emissions was:

$${e}_{i}=\mathop{\sum }\limits_{m=1}^{M}\frac{{C}_{i}\times {E}_{m}}{{C}_{m}}$$
(4)

where ei is industrial carbon emissions in land use grid i; Em is carbon emissions in the original 10 m × 10 km grid m; Ci is count of industry land use grid in target zone i; and Cm is count of industry land use grid in the original 10 m × 10 km grid.

In addition to carbon emissions at the scale of land use grid, we used socioeconomic variables (population density, building density, road networks and land use structure) and biophysical variables (ecosystem services and vegetative sequestration) to determine the allocation of NBS. Socioeconomic variables were those listed in Supplementary Material 2 Table A2.1. For local ecosystem services, we used the indicator for the percentage of natural and semi-natural areas as potential GI from the ESPON GRETA report37. In particular, the ecosystem service of carbon sequestration potential of vegetation was identified using Corine Land Cover data from 2018, by assigning each vegetative land cover category a sequestration potential based on the empirical parameters provided by Page et al.7. The conditions used for determination of each NBS approach are described in details (Supplementary Material 2 Table A2.4), while their prioritized application sites are also spatially allocated (Fig. 4). The above allocation and evaluation strategies of NBSP enable the most carbon emission reduced on per land use grid allocations of NBSP.

Projecting NBS contributions to 2030 climate-neutral goals of European cities

We estimated the percentage of carbon emissions that could be saved by NBS for each sector (residential, industrial and transport) from the meta-analysis results (Supplementary Material 2 Fig. A2.1), with a summation of NBS implemented in each city determined by the allocation:

$${r}_{l,n}=\mathop{\sum }\limits_{i=1}^{I}\frac{{E}_{l,i}\times {R}_{l}}{{E}_{l,i}}$$
(5)

where rl,n is the percentage emissions reduction for sector l in city n; El,i is carbon emissions from sector l in land use grid i; and Rl is the emissions reduction metrics for sector l.

Our emissions reduction metrics take into account the presence of impervious surfaces and the fact that NBS cannot practically fully cover a land-use grid. In such cases, we estimate local parameters to convert area-based metrics from the meta-analysis to the percentage of carbon reduction that can be achieved in different areas such as regions, cities and buildings. This is to account for the specific conditions of the area and to ensure that the estimated carbon reduction is appropriate for the scale of the intervention. This process is especially useful for indirect pathways and small-scale interventions such as green buildings, as the percentage of reduction effects of these approaches on residential energy use and associated carbon emissions has been well documented.

Note that the carbon sequestration effects of some NBSP implementations (mainly urban forests and parks) allocated on existing natural areas (such as grasslands or forests) may have already been captured by existing vegetation. For those, habitat preservation and bio-remediation are allocated as the NBSP, which can lead to an improvement in the quality of these areas in terms of both biodiversity and carbon sequestration. Meanwhile, for NBSP that reduces anthropogenic carbon emissions through indirect pathways, our estimated emission reductions were almost additional even though some vegetation was already in place. For example, street trees were mainly already present around highways featuring high transportation emissions. However, improved intersection design coupled with streetscape features such as street greenery and green pavement, implemented in proximity or in parallel to the land use grid among the highest transportation emission, can promote walking and biking modal choices to mitigate intra-city vehicle travel demand.

To project how NBS would contribute to the climate neutrality goals of European cities in 2030, we conducted a comparative analysis of the percentage of carbon emissions mitigation in each city with its (emissions mitigation pathway) RCP for different SSPs. Our methodology assumed that the future trajectory of urban carbon emissions in each city is influenced by broader regional socioeconomic and environmental trends, and that emissions trajectories follow regional patterns. Based on the regional projection results of the sixth climate model intercomparison project (CMIP6) for SSP scenarios38, we assigned each city a regional average carbon emissions change rate for different SSP baselines in different EU regions, as a proxy for future urban carbon emissions. While we acknowledge the complex and diverse urban emissions landscape in Europe, we apply this generalized method as a starting point to estimate the potential contribution of NBS to carbon emissions mitigation. We generate base carbon emissions for the EU cities and project carbon emissions corresponding to different RCPs in 2030 using these regional average carbon emissions change rates. The equation for total carbon emissions was:

$${E}_{s,n}={E}_{{t}_{0},n}\times \left(1+\frac{\left({E}_{{r}_{n},{t}_{2},s}-{E}_{{r}_{n},{t}_{1},s}\right)\times \left({t}_{2}-{t}_{0}\right)}{{E}_{{r}_{n},{t}_{1},s}}\right)$$
(6)

where Es,n is the carbon emissions for SSP scenario s in city n; \({E}_{{r}_{n},t,s}\) is the carbon emissions of SSP scenario s in EU region rn in year t; rn is the EU regional classification where the city n is located by CMIP6; and t0, t1 and t2 is the starting year of the carbon emissions study, the year closest to the starting year, and the ending year of the study in CMIP6, respectively.

To calculate the carbon emissions of European cities on different SSP development paths under the effect of NBS, we summarized Shared Climate Policy Assumptions (SPA) from SSP1 to SSP5 together with their respective NBS storyline. Specifically, we translated different levels of policy stringency into varying levels of reduction of NBS and conversion of land use. Since the RCPs and SSPs already consider many mitigation activities, we needed to make sure that our analysis incorporating NBS would not involve double-counting of the mitigation policy effects. To address this issue, we generated specific NBS based on the predicted changes in SPAs for SSPs (Supplementary Material 3 Table A3.3). For example, stricter preservation of natural areas in RCPs does not consider the fact that these natural areas can provide additional effects in reducing automobile travel and residential energy consumption. On the other hand, if transportation measures of a certain RCP have already been highly featured (that is, SPA for SSP.1), we reduced the NBS effects on transportation in that scenario to avoid double-counting.

While RCP and SSP consider the net balance of carbon emissions and removals from the atmosphere, they do not directly account for carbon sequestration. Carbon sequestration can be achieved through various methods, such as bioenergy with carbon capture and sequestration (BECCS), afforestation, reforestation and soil carbon sequestration. Although some RCP and SSP scenarios include BECCS, other types of carbon sequestration may not be incorporated39. To fully capture the potential impact of all carbon sequestration methods on mitigating future climate change, it is essential to perform separate sequestration potential calculations. In this study, we calculated the final sequestration potential by combining the converted rate of green space area and the proportion of existing urban carbon sequestration. The equation used for mitigating carbon emissions was as follows:

$${E}_{c,n}=\left({E}_{{t}_{0},n}\times \mathop{\sum}\limits_{i=1}^{l}\left({r}_{l,n}\times {m}_{l,{r}_{n}}\right)\right)-{S}_{n}\times {C}_{{r}_{n}}$$
(7)

where Ec,n is the carbon emissions for climate mitigation model c (including both NBS and carbon sequestration) in city n; \({m}_{l,{r}_{n}}\) is the SPA-converted mitigation rate for emissions sector l for CMIP6 classified EU region r; S is the carbon sequestration rate; and C[0, 2] is the standardized green space change index, which measures the change in green space area in a city over time, with higher values indicating a greater increase in green space area. While predicting carbon sequestration in SSP scenarios typically involves calculating the change in vegetation through land use40, we used a simplified approach7 to calculate carbon sequestration change. Detailed information is provided in Supplementary Material 3.

Based on the climate mitigation model results for urban carbon emissions for both the SSP and RCP scenarios, we further elaborated per capita carbon emissions based on the respective population change projections. This enabled us to make cross-sectional comparisons of the progress made by different cities and regions in Europe towards reaching their climate neutrality goals by 2030, and the contribution of NBS to this progress.

Reporting summary

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