Built structures influence patterns of energy demand and CO2 emissions across countries

Abstract

Built structures, i.e. the patterns of settlements and transport infrastructures, are known to influence per-capita energy demand and CO₂ emissions at the urban level. At the national level, the role of built structures is seldom considered due to poor data availability. Instead, other potential determinants of energy demand and CO₂ emissions, primarily GDP, are more frequently assessed. We present a set of national-level indicators to characterize patterns of built structures. 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 commonly included in national-level analyses of determinants of energy use and emissions. We find that these indicators are about equally important for predicting energy demand and CO₂ emissions as GDP and other conventional factors. The area of built-up land per capita is the most important predictor, second only to the effect of GDP.

Fig. 1. Workflow of this study.

National-level indicators of the extent and spatial patterns of settlements and infrastructures (material stock patterns) were derived from global maps, here illustrated using Paraguay, the UK, Kenia, and Bangladesh as examples. Results were statistically analyzed along with the conventional factors assumed to co-determine energy use and CO₂ emissions. The main aim was to test the hypotheses formulated in the lower-right box. Copyright for administrative boundaries: © Eurogeographics.

Fig. 2. Correlation analyses of total final energy demand per capita (TFC) and per-capita CO₂ emissions (CO₂) with conventional factors and material stock pattern indicators.

a Pearson’s zero correlation coefficients of correlations between TFC (left) and CO₂ (right) and material stock pattern indicators as well as conventional factors. Natural logarithms of the variables were analyzed. Squaring the correlation coefficients gives the percentage of the cross-country variation of CO₂ or TFC explained by the respective indicator alone. Correlations were not significant for variables marked with an asterisk (p < 0.1). b Semi-partial correlations between material stock pattern indicators and TFC (left) and CO₂ (right) controlling for GDP and DENS. Distance from the vertical axis indicates the correlation coefficient of the semi-partial correlation, and distance from the horizontal axis is the correlation coefficient of the bivariate (uncontrolled) correlation. Red contours indicate insignificant results (significance level p < 0.1).