Settlement and infrastructure patterns influence energy use and CO2 emissions almost as much as economic activity

doi:10.21203/rs.3.rs-1210465/v1

PDF

Physical Sciences - Article

Settlement and infrastructure patterns influence energy use and CO2 emissions almost as much as economic activity

https://doi.org/10.21203/rs.3.rs-1210465/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted 17 Jan, 2022

You are reading this latest preprint version

Spatial patterns of settlements and transport infrastructures are known to influence per-capita energy use and CO2 emissions at the urban level1–4. At the national level, other potential determinants of energy use and CO2 emissions, primarily GDP, received much attention5–7, whereas the role of settlements and infrastructure patterns was disregarded due to lacking data. We present a set of novel national-level indicators derived from global satellite- and crowd-sourced maps to characterize extent and spatial patterns of settlements and infrastructures. We quantify these indicators for 113 countries and statistically analyze the results along with final energy use and territorial CO2 emissions, as well as factors usually considered in IPAT (impact = population x affluence x technology) approaches. We find that our indicators have similar explanatory power for energy use and CO2 emissions as GDP and IPAT factors. Covariates include built-up area, urban population density, the spatial clustering of built-up land, as well as transport networks, in addition to the effect of GDP. We conclude that the extent and spatial layout of settlements and infrastructures strongly affects resource use and emissions. Resource-sparing settlements and transport networks are under-appreciated options for reducing energy use and CO2 emissions.

Rapidly progressing climate heating driven by growing greenhouse gas (GHG) emissions is a major concern⁸. GHGs mainly result from energy-related combustion of fossil fuels⁹. The question emerges, which factors drive energy use and emissions, and to what extent their trajectories can be altered. A widespread approach classifies drivers of emissions and resource use into population, affluence and technology^10–12. These drivers are captured in the STIRPAT (Stochastic Impacts by Regression on Population, Affluence and Technology) framework¹⁰, derived from the classical IPAT (impact = population x affluence x technology) approach^12,13.

Economic activity (‘affluence’ in STIRPAT, usually measured as Gross Domestic Product or GDP) is acknowledged as a major driver of resource use and greenhouse gas (GHG) emissions^5–7. The discussion mainly focuses on whether GDP can be ‘decoupled’ from emissions, i.e., whether energy use or emissions can be reduced while GDP is growing. This may be possible e.g. through more efficient technologies, but the debate is so far inconclusive^14,15. While a recent study found that policies implemented in the last decade achieved reductions in CO₂ emissions in some growing economies¹⁶, a meta-analysis yielded little evidence for reductions in energy use or GHG emissions consistent with ambitious climate targets in growing economies¹⁷. Other potential drivers received less attention than GDP. Population density has been studied with varied outcomes. Some studies found that low population density favors high per-capita resource use^18,19, other analyses found no²⁰ or opposite²¹ effects. Relevant factors also include climatic conditions that drive energy demand for heating or cooling of buildings²², fuel prices affecting demand for transport energy²³, and the urbanization rate (urban population as percent of total population).

At the urban scale, the influence of population density and the spatial layout of urban areas on cities’ resource demand has been widely studied^{1–4,22,24,25}. There are several reasons why the extent and spatial layout (density and form) of settlements and infrastructures – henceforth denoted as ‘material stock patterns’²⁶, i.e. the spatial patterns of society’s material stocks in infrastructures and buildings – should affect energy use and CO₂ emissions. The accumulation of material stocks requires massive amounts of resources such as steel-reinforced concrete, mortar, bricks, timber, plastics, glass, gravel, sand, etc.^27–30. Heating, cooling or lighting of buildings and production processes in industrial plants require much energy^1,4,22,31, as does mobility of goods and people on roads and railways^30,32.

Despite these insights from urban studies, material stock patterns are largely ignored in debates on national-level decoupling^15,17 and STIRPAT analyses. We found only one STIRPAT study³³ considering material stock patterns in analyzing transport-related emissions. Hence little is known on the effects of material stock patterns on energy use and CO₂ emissions beyond the city level. This results from a scale mismatch: maps of material stock patterns provide fine-grained spatial detail²⁷ that cannot be interpreted at the national scale, where STIRPAT and decoupling research is conducted. For such analyses, spatial data need to be aggregated to the national level in a manner that preserves key information.

To address this research gap, we developed national-level indicators of material stock patterns focused on features of building and settlement patterns that, based on urban studies, might be potential determinants of resource use and emissions (Figure 1). We calculated the indicators presented in Table 1 for 113 countries comprising 91.2% of the world population and 97.3% of global GDP. Yearly total per-capita final energy use (abbreviated as TFC) was used as resource indicator, complemented by yearly per-capita CO₂ emissions (abbreviated as CO₂). To test material stock pattern indicators against other potential determinants of TFC and CO₂, we assembled a cross-country dataset of STIRPAT indicators including economic activity (GDP/cap/yr, abbreviated GDP), population density (DENS) and urban population as percent of total population (UPOP) as development indicator³⁴. Heating-degree days (HDD) served as a proxy for climate dependency of energy demand²² and the price of gasoline (PGAS) represented energy prices that are strongly related with settlement patterns²³. Extensive variables were expressed as per-capita values to facilitate country comparisons and remove countries’ population numbers from the analysis.

National-level indicators of material stock patterns

We test three hypotheses based on aggregated indicators of material stock patterns (Figure 1), represented by three types indicators: (1) Built-up land is represented by two area indicators, one as fraction of a nation’s inhabited land, the other area per inhabitant. Other indicators describe patterns of built-up areas, including their spatial clustering, form and distribution. (2) Road indicators that describe the density (length per unit area) of roads in urban and rural regions and the relations between urban and rural road lengths and densities. (3) Railway indicators were defined in the same manner as those for roads (Table 1).

Table 1. Indicators of the extent and pattern of built-up land and transport infrastructures. These indicators condense spatially explicit information in maps to national-level indicator values assumed to co-determine a nation’s per-capita level of energy use or CO₂ emissions. Global maps showing the indicator values are in Extended Data Figures 4-6.

Name	Abbreviation	Description and interpretation	Unit
(1) Indicators for the extent and pattern of a nation’s built-up land
Fraction of built-up land	BL_fract	Built-up land (buildings & infrastructures) as % of the inhabited land area.	m²/m²
Built-up land per capita	BL_cap	Built-up land per capita.	m²/cap
Dispersion of built-up land	BL_disp	Index based on the average distance of each patch of built-up land to the nearest adjacent patch. High values indicate strong dispersion.	--
Monocentricity of built-up land	BL_mono	Area of the largest contiguous built-up patch as % of the sum of the areas of the ten largest patches. High values indicate dominance of one large center.	m²/m²
Compactness of built-up land	BL_comp	Index describing how ‘round’ or ‘compact’ the shapes of a nation’s built-up land patches are on average.	--
Urban population density	UPdens	Urban population per unit of urban built-up land	cap/m²
(2) Indicators of road density and distribution
Road density	RD_total	Length of roads per unit area of inhabited land.	m/m²
Urban road density	RD_urban	Length of roads in urban areas per unit area of urban areas.	m/m²
Rural road density	RD_rural	Length of roads in rural areas per unit area of rural areas; proxy of rural accessibility and connectivity between urban centers.	m/m²
Ratio of urban-to-rural road lengths	RL_urb-rur	Ratio of urban to rural road lengths, indicating the extent to which roads are concentrated in cities.	--
Ratio of urban-to-rural road density	RD_urb-rur	Ratio of RD_urban and RD_rural, indicating the difference between urban and rural road densities.	--
(3) Indicators of railway density and distribution
Railway density	RWD_total	Length of railways per unit area of inhabited land.	m/m²
Urban railway density	RWD_urban	Length of railways in urban areas per unit area of urban areas.	m/m²
Rural railway density	RWD_rural	As RD_rural for railways.	m/m²
Ratio of urban-to-rural railway lengths	RWL_urb-rur	As RL_urb-rur for railways.	--
Ratio of urban-to-rural railway density	RWD_urb-rur	As RD_urb-rur for railways.	--

Pairwise and semi-partial regressions

In terms of their Pearson correlation coefficients in pairwise regressions, several indicators of material stock patterns are as strongly correlated with TFC and CO₂ as common STIRPAT factors (Figure 2a). GDP is positively correlated with both TFC and CO₂. HDD and the fraction of urban population also show the expected pattern, while PGAS and DENS are largely uncorrelated. Almost all spatial indicators are correlated with both TFC and CO₂. The extent of built-up land (BL_cap and BL_fract) is positively correlated with both TFC and CO₂, as are total and rural road density and the dispersion of built-up land and most railway-related indicators. The correlation coefficients of BL_cap with TFC and CO₂ are both ~0.7; BL_cap is the second-best predictor of both TFC and CO₂ after GDP. As expected from the urban literature, urban population density (UP_dens) is inversely correlated with CO₂ and energy, whereas the share of urban population (UPOP) is a strongly positively correlated with CO₂ and TFC. Inverse relations prevail for BL_mono, BL_comp and the urban-to-rural relations of infrastructure density.

Semi-partial correlations of the material stock pattern indicators controlled for GDP and population density (DENS) are shown in Figure 2b. The part of each material stock pattern indicator correlated with GDP and DENS is removed, revealing the strength of the linear correlation between TFC or CO₂ and the remaining part of the respective variable. The distance from the vertical axis either to the right (positive correlation) or to the left (inverse correlation), depicts the additional explanatory power of the respective indicator over a model considering only GDP and DENS. Several indicators provide additional explanations over GDP and DENS alone. Built-up land (both BL_cap and BL_fract) is positively correlated with TFC and CO₂, as are the density of rail and road infrastructures, especially in rural areas. Like in urban studies, UPOP is inversely correlated with TFC and CO₂, as is urban road density (not significant) and BL_mono. PGAS, which had not been significant in the bivariate correlations, emerges as an important factor, which is also observed for RD_urban and other indicators. UP_dens loses importance, most likely due to its high correlation with GDP.

Multivariate models

The ability of the material stock pattern indicators to add insights beyond STIRPAT is further analyzed in Table 2, which summarizes an extended multivariate analysis, taking STIRPAT as reference. We performed stepwise regressions; variables were selected either only from STIRPAT indicators (rows labelled STIRPAT) or from all indicators (rows labelled ALL). Table 2 presents the statistically best parsimonious model in each category based on p values of an F-statistic in which models were tested with and without the respective variable in each step. Variables that were not significant in the semi-partial analysis (Fig 2b) were excluded. The final model is robust to the use of BIC as an inclusion/exclusion criterion. Hence, the best model was selected in terms of its ability to explain international differences in TFC or CO₂ with the least number of factors. In addition to analyses that included GDP as independent variable (part A of Table 2), we also performed an analysis in which GDP was excluded (part B of Table 2). This analysis tested whether the cross-country patterns of TFC and CO₂ could be explained without resorting to economic activity as an explanatory variable. GDP is a global variable that is connected with many of the factors we analyze, thus this analysis reduces global multicollinearity problems and shows, which indicators have additional explanatory power in this multivariate setting. All variables are taken as logarithms; hence coefficients can be interpreted as elasticities, i.e. as the relative change in TFC, respectively CO₂, associated with a 1% change in the respective independent variable.

Table 2. Results of a stepwise multivariate regression analysis selecting the best parsimonious model in each category. A) including B) excluding GDP as explanatory variable. Rows STIRPAT show models constructed using only STIRPAT indicators, rows ALL show models where both STIRPAT and material stock pattern indicators were included as potential predictors. R²max reports the R² of a model that includes all variables. Cells left empty indicate that the respective variable was not selected. The table includes the estimated OLS parameter, where *, **, *** indicates significance at 10, 5, 1%; red marks non-significant results.

A) GDP included as explanatory variable

B) GDP excluded as explanatory variable

CO₂

TFC

CO₂

TFC

STIRPAT

ALL

STIRPAT

ALL

STIRPAT

ALL

STIRPAT

ALL

Intercept

-4.29***

-3.98***

2.73***

3.19***

-8.99***

-5.50***

-2.10***

0.76

GDP/cap

0.58***

0.44***

0.60***

0.55***

DENS

0.02

-0.05*

0.07

-0.10

0.00

-0.06

UPOP

0.56***

0.46**

-0.05

1.86***

1.13***

1.29***

0.71***

HDD

0.11***

0.06***

0.05***

0.04***

0.16***

0.04

0.11***

PGAS

-0.69***

-0.75***

-0.33***

-0.40***

-0.37**

-0.55***

0.00

-0.20*

BL_cap

0.27***

0.19***

0.34***

0.36***

RD_urb

-0.56***

RD_rur

0.22***

RD_urb-rur

-0.38***

RWD_tot

0.16***

0.93***

RWD_urb

0.16***

RWD_rur

-0.69***

BIC

190.82

168.23

104.12

67.50

256.15

205.50

221.83

159.69

AdjR²

0.81

0.85

0.89

0.65

0.79

0.55

0.74

R²

0.82

0.86

0.85

0.89

0.66

0.80

0.56

0.76

R²MAX

0.88

0.92

0.83

0.81

Models including material stock pattern indicators always have higher explanatory power than STIRPAT models, even in part A, where the STIRPAT models has strong explanatory power. If GDP is excluded, the STIRPAT model explains only 55% (adjusted R²) of the variation for TFC and 65% for CO₂. By comparison, the best models in the ALL categories explain 74% of the variation for TFC and 79% for CO₂ when GDP is excluded (Table 2B). These latter models are almost as good as STIRPAT models that include GDP, which explain 85% of the variation for TFC and 81% for CO₂. BL_cap is always selected in ALL models, significant at the 1% level. DENS is not significant in any ALL model, and has little explanatory power in STIRPAT models. Other material stock pattern indicators are selected less frequently, but sometimes have high explanatory power. Urban road density is strongly and inversely related with TFC, in line with urban findings that connectivity reduces transport energy²². In models excluding GDP (Table 2B), RD_rur and RWD_total are strongly positively related with CO₂, whereas RWD_rural is inversely related with CO₂, all of which are expected patterns. In the TFC model, the effect of RD_urb-rur can be explained by the positive effect of a high urban road density, while we lack a straightforward explanation for the modestly positive relation between TFC and RWD_urban.

The analysis confirms that the extent and spatial patterns of buildings and infrastructures play an important role in co-determining the level of resource use, here TFC and CO₂, hence insights from urban studies^1–4 hold at the national scale. Despite the unavoidable loss of information resulting from aggregation of maps to national-scale indicators, they still represent important characteristics of material stock patterns that strongly affect cross-country patterns of resource use and emissions. The indicators have additional explanatory power over STIRPAT approaches and allow building multivariate ‘physical economy’ models that explain international patterns of energy use and emissions well, even if GDP is excluded. Altogether, material stock patterns determine levels of resource use and emissions to a similar extent as GDP does. Population density plays a smaller role than widely assumed, while neglecting material stock patterns in attempts to understand the determinants of cross-country differences in resource use and GHG is a serious omission. Our indicators are a good starting point for considering such patterns in future analyses, which could be useful in analyzing possibilities for decoupling resource use from GDP.

The material stock pattern indicator with the most consistent explanatory value across all analyses is built-up land per capita (BL_cap). This is plausible because infrastructures and buildings require energy for being built and used, which results in CO₂ emissions in current fossil-fuel dominated energy systems. Higher BL_cap also means larger floor size and more space between destinations, which all raises energy demand in buildings and transport. These findings corroborate previous analyses (using different models) suggesting that challenges for climate-change mitigation strongly depend on the future accumulation of additional material stocks in buildings and infrastructures^35–37. This is worrying as material stocks are growing globally largely in unison with GDP^29,37. In terms of the importance of specific spatial patterns, the analysis reveals that relationships found in urban-scale studies are also important at the national scale: high population densities in cities²⁵ and strong connectivity of urban transport infrastructures help reducing energy use and CO₂ emissions, whereas high dispersion of built-up land and large rural transport infrastructures drive them upwards.

While more research is warranted, our results have implications for countries pursuing ambitious climate targets⁹. In general, build-up of large material stocks is identified as a currently neglected driver of energy use and emissions. Efforts to ensure societal wellbeing while constraining land conversion to settlements and infrastructures help moving towards zero-carbon societies. Improved design of settlement and infrastructure patterns – e.g., high connectivity with infrastructures favoring walking and cycling in cities – is commendable. Reducing fragmentation of landscapes and constraining urban sprawl, reducing expansion of highways and other investment in infrastructures that favor energy-intensive mobility patterns emerge as options that could help reducing energy use. These options could contribute to low-energy futures with high wellbeing³⁸ that would facilitate staying within a 1.5° global warming target without large-scale implementation of CO₂ reduction options such as bioenergy with carbon capture and sequestration³¹.

1. Güneralp, B. et al. Global scenarios of urban density and its impacts on building energy use through 2050. PNAS 114, 8945–8950 (2017).

2. Kenworthy, J. R. Passenger Transport Energy Use in Ten Swedish Cities: Understanding the Differences through a Comparative Review. Energies 13, 3719 (2020).

3. Newman, P. W. G. Sustainability and cities: extending the metabolism model. Landscape and Urban Planning 44, 219–226 (1999).

4. Seto, K. C. et al. Human Settlements, Infrastructure and Spatial Planning. in Climate Change 2014: Mitigation of Climate Change. (eds. Edenhofer, O. et al.) 923–1000 (Cambride University Press, 2014).

5. Ayres, R. U. & Warr, B. The Economic Growth Engine: How Energy And Work Drive Material Prosperity. (Edward Elgar, 2009).

6. Hall, C. A. S. & Klitgaard, K. A. Energy and the Wealth of Nations. An Introduction to Biophysical Economics. (Springer, 2017).

7. Stern, D. I. The role of energy in economic growth. Annals of the New York Academy of Sciences 1219, 26–51 (2011).

8. IPCC. Climate Change 2021, The Physical Science Basis. (WMO, UNEP, 2021).

9. IPCC. Global Warming of 1.5 °C. (WMO, UNEP, 2018).

10. Dietz, T., Shwom, R. L. & Whitley, C. T. Climate Change and Society. Annual Review of Sociology 46, 135–158 (2020).

11. Jorgenson, A. K. et al. Social science perspectives on drivers of and responses to global climate change. WIREs Climate Change 10, e554 (2019).

12. Liddle, B. Impact of population, age structure, and urbanization on carbon emissions/energy consumption: evidence from macro-level, cross-country analyses. Popul Environ 35, 286–304 (2014).

13. Ehrlich, P. R. & Holdren, J. P. Impact of Population Growth. Science 171, 1212–1217 (1971).

14. Jackson, T. & Victor, P. A. Unraveling the claims for (and against) green growth. Science 366, 950–951 (2019).

15. Vadén, T. et al. Decoupling for ecological sustainability: A categorisation and review of research literature. Environmental Science & Policy 112, 236–244 (2020).

16. Quéré, C. L. et al. Drivers of declining CO 2 emissions in 18 developed economies. Nat. Clim. Chang. 9, 213–217 (2019).

17. Haberl, H. et al. A systematic review of the evidence on decoupling of GDP, resource use and GHG emissions, part II: synthesizing the insights. Environ. Res. Lett. 15, 065003 (2020).

18. Shafiei, S. & Salim, R. A. Non-renewable and renewable energy consumption and CO2 emissions in OECD countries: A comparative analysis. Energy Policy 66, 547–556 (2014).

19. Steinberger, J. K., Krausmann, F. & Eisenmenger, N. Global patterns of materials use: A socioeconomic and geophysical analysis. Ecological Economics 69, 1148–1158 (2010).

20. Parikh, J. & Shukla, V. Urbanization, energy use and greenhouse effects in economic development: Results from a cross-national study of developing countries. Global Environmental Change 5, 87–103 (1995).

21. Rahman, M. M. Do population density, economic growth, energy use and exports adversely affect environmental quality in Asian populous countries? Renewable and Sustainable Energy Reviews 77, 506–514 (2017).

22. Creutzig, F., Baiocchi, G., Bierkandt, R., Pichler, P.-P. & Seto, K. C. Global typology of urban energy use and potentials for an urbanization mitigation wedge. PNAS 112, 6283–6288 (2015).

23. Creutzig, F. How fuel prices determine public transport infrastructure, modal shares and urban form. Urban Climate 10, 63–76 (2014).

24. Newman, P. W. G. & Kenworthy, J. R. Transport and urban form in thirty‐two of the world’s principal cities. Transport Reviews 11, 249–272 (1991).

25. Pomponi, F., Saint, R., Arehart, J. H., Gharavi, N. & D’Amico, B. Decoupling density from tallness in analysing the life cycle greenhouse gas emissions of cities. npj Urban Sustain 1, 33 (2021).

26. Haberl, H. et al. Contributions of sociometabolic research to sustainability science. Nature Sustainability 2, 173–184 (2019).

27. Haberl, H. et al. High-Resolution Maps of Material Stocks in Buildings and Infrastructures in Austria and Germany. Environ. Sci. Technol. 55, 3368–3379 (2021).

28. Kennedy, C. The energy embodied in the first and second industrial revolutions. Journal of Industrial Ecology 24, 887–898 (2020).

29. Krausmann, F. et al. Global socioeconomic material stocks rise 23-fold over the 20th century and require half of annual resource use. Proceedings of the National Academy of Sciences of the USA 114, 1880–1885 (2017).

30. Lamb, W. F. et al. A review of trends and drivers of greenhouse gas emissions by sector from 1990 to 2018. Environ. Res. Lett. 16, 073005 (2021).

31. Grubler, A. et al. A low energy demand scenario for meeting the 1.5 °C target and sustainable development goals without negative emission technologies. Nature Energy 3, 515 (2018).

32. Sims, R. et al. Transport. in Climate Change 2014, Mitigation of Climate Change. Contribution of Working Group III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (eds. Edenhofer, O., Pichs-Madruga, R. & Sokona, Y.) 599–670 (Cambride University Press, 2014).

33. Lim, J., Kang, M. & Jung, C. Effect of national-level spatial distribution of cities on national transport CO2 emissions. Environmental Impact Assessment Review 77, 162–173 (2019).

34. Krausmann, F., Weisz, H. & Eisenmenger, N. Transitions in Socio-Metabolic Regimes Through Human History. in Social Ecology: Society-Nature Relations Across Time And Space (eds. Haberl, H., Fischer-Kowalski, M., Krausmann, F. & Winiwarter, V.) 63–92 (Springer, 2016).

35. Cao, Z. et al. The sponge effect and carbon emission mitigation potentials of the global cement cycle. Nature Communications 11, 3777 (2020).

36. Krausmann, F., Wiedenhofer, D. & Haberl, H. Growing stocks of buildings, infrastructures and machinery as key challenge for compliance with climate targets. Global Environmental Change 61, 102034 (2020).

37. Watari, T. et al. Global Metal Use Targets in Line with Climate Goals. Environ. Sci. Technol. (2020) doi:10.1021/acs.est.0c02471.

38. Vogel, J., Steinberger, J. K., O’Neill, D. W., Lamb, W. F. & Krishnakumar, J. Socio-economic conditions for satisfying human needs at low energy use: An international analysis of social provisioning. Global Environmental Change 69, 102287 (2021).

Author contributions

HH conception of research design, data analysis, interpretation of results, article drafting, project management, funding acquisition, ML development of spatial indicators, spatial analyses, data analysis, quantification of indicators, designing figures, contribution to writing. APL conceptualization and implementation of statistical analyses, data analysis, designing figures, contribution to article writing, SM spatial and statistical analyses, data analysis and interpretation, contribution to writing, BP collation of STIRPAT indicators, data analysis and interpretation, contribution to writing, DW, FC, KHE contribution to research design, data analysis and interpretation, contribution to writing, JAD conception and supervision of statistical analyses, research design, data analysis and interpretation, contribution to writing.

Correspondence should be addressed to Helmut Haberl ([email protected]).

Competing interests

The authors declare no competing interests

Acknowledgements

This research has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (MAT_STOCKS, grant agreement No 741950).

Supplementary Information and Extended Data

This article includes supplementary information and Extended Data Figures.

There is NO Competing Interest.

SuppInfo.docx
Supplementary Information
ExtendedData.docx
Six Figures used to explain methods and document results

PDF

Version 1

posted 17 Jan, 2022

You are reading this latest preprint version

Settlement and infrastructure patterns influence energy use and CO2 emissions almost as much as economic activity

Status:

Version 1

Abstract

Figures

Main

Results

Discussion

References

Declarations

Competing Interests

Supplementary Files

Status:

Version 1