Put emissions on the map - geospatial approach to air pollution emission analysis

Air pollution poses one of the most harmful environmental human health risks. To assess the health impacts of air pollution, georeferenced emission inventories are needed. The spatial interpolation, i.e. spatial distribution of the emissions in these inventories is essential for any impact assessment, but assessing the quality and uncertainty of the distribution is difficult, and there is a lack of scientific literature on the topic. There is no clear real-world quantity with which to validate the spatial distribution of emissions, as, for example, air quality measurements consider concentrations of the pollutants, not the emissions. The aim of this research is to utilize geospatial approach to improve methods of spatial distribution of air pollution emissions. To study this problem, this work considers and answers the following research questions: 1. How to assess the uncertainty of the spatial distribution of air pollution emissions? 2. What are the best data to use for proxies for spatial distribution of air pollution emissions in the most relevant source sectors? 3. What role does spatial resolution have in health impact assessment? For the first question, the lack of reference data for the spatial distributions meant that no one-forall method to assess the uncertainty was found. Comparison of the distributions of the emissions in different inventories was deemed as the best available method to assess the quality of the spatial proxies. To quantify the similarity of the distributions, an index of agreement was utilized. The index proved to be a good tool to support the comparison of spatial distributions, alongside visual analysis of maps and assessment of the data used for the proxies. For the second question, three source sectors were focused on: residential wood combustion, road transport, and machinery and off-road. Developed Nordic emission inventory was compared to both local and European level inventories. While European level inventories were able to produce similar spatial distributions as the Nordic inventory based on national methods, this was on resolution that was coarser than what the Nordic inventory offers, meaning modifiable areal unit problem (MAUP) needs to be considered for the emission data. In general, for the spatial distribution the difference between urban and rural areas in the spatial proxies was found to be crucial. Furthermore, local characteristics were important to take into consideration. Regarding the third question, health impact estimates from local emissions were concluded to be sensitive to the assessment resolution, and the effect was strongest in urban areas. The results highlight the importance of spatial distribution of the emissions on high resolution especially in urban areas. This work bridges the gap between geospatial theory and air pollution emission inventory framework. The spatial proxies used to describe the spatial distribution of the emissions have been assessed with practical geospatial methods to help analyse the ways the proxies can be developed. The work also describes the state-of-the-art data recommendations for spatial proxies for the three emissions sources sectors.