Mapping Global Fossil Fuel Combustion CO₂ Emissions at High Resolution by Integrating Nightlight, Population Density, and Traffic Network Data

Quantification of global fossil fuel carbon dioxide (CO₂) emissions at fine spatial resolution is emerging as a critical need in climate change research and policy-making. Numerous studies have constructed CO₂ emission inventories by using spatial proxies, such as radiance-calibrated nightlight and population density, to downscale national emission data into finer spatial scales. However, only using nightlight imagery and population density datasets cannot sufficiently explain the spatial characteristics of potential emission sources from transportation. In this study, we integrated nighttime imagery, population density, and traffic network data to estimate CO₂ emission, and performed a linear regression model with corrective measure to create a high-resolution global grid of fossil fuel carbon emissions in 2010. The experimental results show that the model which considers all these factors (nighttime lights, population density, and traffic network data) exhibited a more reasonable distribution of CO₂ emission than those involving only one or two factors. Besides, in contrast to previous studies on the actual statistical data of CO₂ emissions at the level of subadministrative units of mainland China and the United States, the correlation coefficient of our inventory is significantly larger than those of the other two inventories, while both the mean absolute error and root-mean-squared error of our inventory are the smallest values among the three inventories. The inventory established in this study shows strong agreement with subnational-level CO₂ emission statistics. The resulting dataset and corresponding methods would be of immediate use to global climate managers and policy-making communities.